AA_PRACTICA1_GRUPO_4.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": [],
      "collapsed_sections": [
        "pzs-25Bvw50W",
        "IiqUjYKEl4G0",
        "l5GNinIhnd2a"
      ],
      "toc_visible": true,
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "TPU",
    "gpuClass": "standard"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/javierarteagapuell/deeplearning/blob/main/AA_PRACTICA1_GRUPO_4.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# AA_PRACTICA1_GRUPO_4\n",
        "\n",
        "Daniel Alonso\n",
        "\n",
        "Javier Arteaga\n",
        "\n",
        "Marina Garrido\n",
        "\n",
        "Elena González"
      ],
      "metadata": {
        "id": "E5GHvTYcZxIJ"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Importar librerías"
      ],
      "metadata": {
        "id": "qHJD9Fh5qcuW"
      }
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "Kj0gniXK4uEd",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "4fa37a9e-ece3-40c3-9cbd-de35d7657cf6"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Colab only includes TensorFlow 2.x; %tensorflow_version has no effect.\n",
            "2.9.0\n",
            "GPU Available: []\n",
            "2.9.2\n"
          ]
        }
      ],
      "source": [
        "%tensorflow_version 2.x\n",
        "import tensorflow\n",
        "from tensorflow import keras\n",
        "from tensorflow.keras import layers\n",
        "import numpy as np\n",
        "print(tensorflow.keras.__version__)\n",
        "print('GPU Available:', tensorflow.config.list_physical_devices('GPU'))\n",
        "print(tensorflow.__version__)"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from keras.datasets import fashion_mnist \n",
        "from keras.utils.np_utils import to_categorical\n",
        "from keras import models\n",
        "from keras import layers\n",
        "from keras import optimizers\n",
        "from keras import losses\n",
        "from keras.layers.core import Flatten\n",
        "from keras import metrics\n",
        "import matplotlib.pyplot as plt\n",
        "from collections import Counter\n",
        "from sklearn.metrics import confusion_matrix\n",
        "import itertools\n",
        "import numpy as np"
      ],
      "metadata": {
        "id": "yCls260ob3Xo"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Cargar los datos"
      ],
      "metadata": {
        "id": "AB--zOzNbD_i"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Cargar los datos de Fashion MNIST en Keras. \n",
        "\n",
        "Hay 60K imágenes para entrenar la red y 10K para evaluar la precisión de clasificación.\n"
      ],
      "metadata": {
        "id": "PtM5VECf5oDf"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "(train_data, train_labels), (test_data, test_labels) = fashion_mnist.load_data()\n",
        "print('Imágenes de entrenamiento:', train_data.shape)\n",
        "print('Imágenes de test:', test_data.shape)"
      ],
      "metadata": {
        "id": "ceq8WM5X5pe2",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "a159ac9f-aff5-4ef9-c58f-351e31e4abcb"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Imágenes de entrenamiento: (60000, 28, 28)\n",
            "Imágenes de test: (10000, 28, 28)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Los datos tienen forma de 60K y 10K matrices (imágenes), respectivamente, de 28 filas x 28 columnas = 784 (píxeles).\n",
        "\n",
        "Se modifica la forma de los datos a 60K y 10K matrices, respectivamente, de matrices 28x28. Además, se normalizan los datos para facilitar que converja el entrenamiento, ya que el resultado se obtiene en porcentaje de probabilidad (valores entre 0 y 1)."
      ],
      "metadata": {
        "id": "oqUS89i0Rxqp"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "train_data = train_data.reshape((60000, 28 * 28))\n",
        "train_data = train_data.astype('float32') / 255\n",
        "test_data = test_data.reshape((10000, 28 * 28))\n",
        "test_data = test_data.astype('float32') / 255\n",
        "print('Imágenes de entrenamiento:', train_data.shape)\n",
        "print('Imágenes de test:', test_data.shape)"
      ],
      "metadata": {
        "id": "9FhG-daMMP5f",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "a4cba419-b12d-4989-a1d9-71d52c1a821b"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Imágenes de entrenamiento: (60000, 784)\n",
            "Imágenes de test: (10000, 784)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Se codifican las etiquetas de las imágenes de forma que lo pueda procesar la red. Cada etiqueta se convierte en un vector de 10 elementos (número de categorías) en el que el único valor distinto de ```0``` es un ```1``` en la posición de la categoría a la que corresponde."
      ],
      "metadata": {
        "id": "7vPtSQP3MtM4"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "train_labels = to_categorical(train_labels)\n",
        "test_labels = to_categorical(test_labels)"
      ],
      "metadata": {
        "id": "GW3NxRqYQQZ6"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "Se seleccionan 10K imágenes de entrenamiento como datos de validación para poder estudiar la precisión de la red frente a nuevos datos."
      ],
      "metadata": {
        "id": "DJSeRVrjVPMa"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Datos de validación\n",
        "val_data = train_data[:10000]\n",
        "val_labels = train_labels[:10000]\n",
        "\n",
        "# Datos de entrenamiento\n",
        "train_val_data = train_data[10000:]\n",
        "train_val_labels = train_labels[10000:]\n"
      ],
      "metadata": {
        "id": "VfrSdJ--VV-A"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Caso 1"
      ],
      "metadata": {
        "id": "2xL3znkukYG3"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Crear la red de neuronas"
      ],
      "metadata": {
        "id": "pfXNIXfvkYG3"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Se crea la red neuronal con una capa de entrada con 10 neuronas y función de activación ```relu```, y una capa de salida con función de activación ```softmax```."
      ],
      "metadata": {
        "id": "hd2FYxcMkYG3"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model = models.Sequential()\n",
        "model.add(layers.Dense(10, activation ='relu', name = 'Capa1', input_shape = (28*28,)))\n",
        "model.add(layers.Dense(10, name = 'CapaSalida1', activation = 'softmax'))\n",
        "\n",
        "model.add(Flatten())\n",
        "model.output_shape\n"
      ],
      "metadata": {
        "id": "E2kugFRNkYG3",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "450039e1-58b8-4aa4-f69f-490d2a69244f"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(None, 10)"
            ]
          },
          "metadata": {},
          "execution_count": 72
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Explicar model.summary()\n"
      ],
      "metadata": {
        "id": "V12EvOyAkYG3"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "El resumen del modelo indica el número de parámetros de salida de cada capa. Este número es igual al tamaño de los datos de entrada multiplicado por el número de neuronas, y sumando el número de umbrales de activación de la capa.\n",
        "\n",
        "\n",
        "En el caso de la primera capa: 7850 = 784 (tamaño de los datos) x 10 (número de neuronas) + 10 (número de umbrales de activación).\n",
        "\n",
        "En el caso de la segunda capa: 110 = 10 (tamaño de los datos) x 10 (número de neuronas) + 10 (número de umbrales de activación)."
      ],
      "metadata": {
        "id": "fTquOce_kYG3"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model.summary()\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "3b5b180a-ac6a-44d0-d120-53a01096cc89",
        "id": "oKDb1F1ukYG4"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential_6\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " Capa1 (Dense)               (None, 10)                7850      \n",
            "                                                                 \n",
            " CapaSalida1 (Dense)         (None, 10)                110       \n",
            "                                                                 \n",
            " flatten_2 (Flatten)         (None, 10)                0         \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 7,960\n",
            "Trainable params: 7,960\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Se definen las funciones del optimizador, la pérdida y la métrica."
      ],
      "metadata": {
        "id": "l_mwrJn4kYG4"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model.compile(optimizer = 'sgd',\n",
        "              loss = 'categorical_crossentropy',\n",
        "              metrics = ['accuracy'])"
      ],
      "metadata": {
        "id": "tLaGWZxgkYG4"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Entrenar la red"
      ],
      "metadata": {
        "id": "WnhOIJIskYG4"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Para entrenar la red se realizan ```5``` ciclos con lotes de ```128``` imágenes. Se compara con los datos de validación para controlar el sobreentrenamiento. "
      ],
      "metadata": {
        "id": "e2T7D3_JkYG4"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "history = model.fit(train_val_data, train_val_labels,\n",
        "                    epochs = 5,\n",
        "                    batch_size = 128,\n",
        "                    validation_data = (val_data, val_labels))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "fb02c07b-a8b5-4df9-a492-56967fb42bb1",
        "id": "NRSZiSZIkYG4"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/5\n",
            "391/391 [==============================] - 2s 3ms/step - loss: 1.8724 - accuracy: 0.3568 - val_loss: 1.4650 - val_accuracy: 0.5438\n",
            "Epoch 2/5\n",
            "391/391 [==============================] - 1s 3ms/step - loss: 1.1055 - accuracy: 0.6639 - val_loss: 0.8869 - val_accuracy: 0.7317\n",
            "Epoch 3/5\n",
            "391/391 [==============================] - 1s 3ms/step - loss: 0.8020 - accuracy: 0.7444 - val_loss: 0.7354 - val_accuracy: 0.7601\n",
            "Epoch 4/5\n",
            "391/391 [==============================] - 1s 3ms/step - loss: 0.7022 - accuracy: 0.7692 - val_loss: 0.6692 - val_accuracy: 0.7790\n",
            "Epoch 5/5\n",
            "391/391 [==============================] - 1s 3ms/step - loss: 0.6475 - accuracy: 0.7883 - val_loss: 0.6291 - val_accuracy: 0.7926\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "La pérdida indica el porcentaje de errores entre las etiquetas y las predicciones del modelo.\n",
        "\n",
        "La precisión indica el porcentaje de aciertos entre las predicciones y las etiquetas."
      ],
      "metadata": {
        "id": "zJ9bW9YKkYG4"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "test_loss, test_acc = model.evaluate(test_data, test_labels)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "5cac9e00-635e-4dc0-f5a8-65ac45447ef7",
        "id": "KFffdFRxkYG4"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 1s 2ms/step - loss: 0.6463 - accuracy: 0.7808\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import matplotlib.pyplot as plt\n",
        "\n",
        "acc = history.history['accuracy']\n",
        "val_acc = history.history['val_accuracy']\n",
        "loss = history.history['loss']\n",
        "val_loss = history.history['val_loss']\n",
        "\n",
        "epochs = range(1, len(acc) + 1)\n",
        "\n",
        "# \"bo\" es para \"punto azun\"\n",
        "plt.plot(epochs, loss, 'bo', label='Pérdida de Entrenamiento')\n",
        "# \"b\" es para \"linea continua azul\"\n",
        "plt.plot(epochs, val_loss, 'b', label='Pérdida de Validación')\n",
        "plt.title('Pérdida de Entrenamiento y Validación')\n",
        "plt.xlabel('Epochs')\n",
        "plt.ylabel('Loss')\n",
        "plt.legend()\n",
        "\n",
        "plt.show()"
      ],
      "metadata": {
        "id": "MdWqzr6vxoFh",
        "outputId": "6b28bdef-6764-4a50-90ac-37daabcc5ea8",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Predecir valores"
      ],
      "metadata": {
        "id": "r0wt2CTDkYG4"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Imprimir la matriz de confusión, que muestra los valores predichos frente a los reales."
      ],
      "metadata": {
        "id": "4MzquXO2kYG5"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Note, this code is taken straight from the SKLEARN website, an nice way of \n",
        "# viewing confusion matrix.\n",
        "def plot_confusion_matrix(cm, classes,\n",
        "                          normalize = False,\n",
        "                          title = 'Confusion matrix',\n",
        "                          cmap = plt.cm.Blues):\n",
        "    '''\n",
        "    This function prints and plots the confusion matrix.\n",
        "    Normalization can be applied by setting `normalize = True`.\n",
        "    '''\n",
        "    plt.imshow(cm, interpolation = 'nearest', cmap = cmap)\n",
        "    plt.title(title)\n",
        "    plt.colorbar()\n",
        "    tick_marks = np.arange(len(classes))\n",
        "    plt.xticks(tick_marks, classes, rotation = 45)\n",
        "    plt.yticks(tick_marks, classes)\n",
        "\n",
        "    if normalize:\n",
        "        cm = cm.astype('float') / cm.sum(axis = 1)[:, np.newaxis]\n",
        "\n",
        "    thresh = cm.max() / 2.\n",
        "    for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):\n",
        "        plt.text(j, i, cm[i, j],\n",
        "                 horizontalalignment = 'center',\n",
        "                 color = 'white' if cm[i, j] > thresh else 'black')\n",
        "\n",
        "    plt.tight_layout()\n",
        "    plt.ylabel('Observación')\n",
        "    plt.xlabel('Predicción')"
      ],
      "metadata": {
        "id": "76b0Ovx5kYG5"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "Y_pred = model.predict(test_data)\n",
        "print('Valores de predicción del modelo')\n",
        "# Convertir en predicciones de clases desde los vectores one-hot encoding \n",
        "Y_pred_classes = np.argmax(Y_pred, axis = 1) \n",
        "# Convertir en clases de observación desde los vectores one-hot encoding\n",
        "Y_true = np.argmax(test_labels, axis = 1) \n",
        "# Computar la matriz de confusión\n",
        "confusion_mtx = confusion_matrix(Y_true, Y_pred_classes) \n",
        "# Dibujar la matriz de confusión\n",
        "print('Matriz de confusión')\n",
        "plot_confusion_matrix(confusion_mtx, classes = range(10))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 366
        },
        "outputId": "92ebdb3a-8fce-4fb3-ede5-6c47d27a765e",
        "id": "feFxDVJVkYG5"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 0s 1ms/step\n",
            "Valores de predicción del modelo\n",
            "Matriz de confusión\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Se grafica el elemento 6 (```test_data[5]```) para comprobar las soluciones. "
      ],
      "metadata": {
        "id": "h5LFpB6vkYG5"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "digit = test_data.reshape(10000,28, 28)[5]\n",
        "plt.imshow(digit, cmap = plt.cm.binary)\n",
        "plt.show()\n",
        "print(test_labels[5])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 283
        },
        "outputId": "df6f63e5-12f6-434c-e930-d6f92ed14eb6",
        "id": "uwx9R-zAkYG5"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[0. 1. 0. 0. 0. 0. 0. 0. 0. 0.]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "La clase del elemento 6 es la clase 1 (pantalones)."
      ],
      "metadata": {
        "id": "rJWwgRTokYG5"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "predictions = model.predict(test_data)\n",
        "print('Predicción del elemento 6: clase', np.argmax(predictions[5]))\n",
        "print(predictions[5])\n",
        "print('Suma de las predicciones del elemento 6:', np.sum(predictions[5]))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "57dcaa7a-986b-4732-b499-64825de87e82",
        "id": "uQqd3SDkkYG5"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 0s 1ms/step\n",
            "Predicción del elemento 6: clase 1\n",
            "[1.5701486e-03 9.6329826e-01 3.6851829e-04 1.0837999e-02 1.9751115e-02\n",
            " 1.7835138e-05 4.0504690e-03 8.1603874e-05 2.3705077e-06 2.1786480e-05]\n",
            "Suma de las predicciones del elemento 6: 1.0000001\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "def plot_image(i, predictions_array, true_label, img):\n",
        "  predictions_array, true_label, img = predictions_array, true_label[i], img[i]\n",
        "  plt.grid(False)\n",
        "  plt.xticks([])\n",
        "  plt.yticks([])\n",
        "\n",
        "  plt.imshow(img, cmap = plt.cm.binary)    \n",
        "\n",
        "  predicted_label = np.argmax(predictions_array)\n",
        "  if predicted_label.all() == true_label.all():\n",
        "    color = 'blue'\n",
        "  else:\n",
        "    color = 'red'\n",
        "\n",
        "  plt.xlabel('{} {:2.0f}% ({})'.format(predicted_label,\n",
        "                                100*np.max(predictions_array),\n",
        "                                true_label),\n",
        "                                color = color)\n",
        "\n",
        "def plot_value_array(i, predictions_array, true_label):\n",
        "  predictions_array, true_label = predictions_array, true_label[i]\n",
        "  plt.grid(False)\n",
        "  plt.xticks(range(10))\n",
        "  plt.yticks([])\n",
        "  thisplot = plt.bar(range(10), predictions_array, color = '#00FF00')\n",
        "  plt.ylim([0, 1])\n",
        "  predicted_label = np.argmax(predictions_array)\n",
        "\n",
        "  thisplot[predicted_label].set_color('red')\n",
        "  thisplot[np.argmax(true_label)].set_color('black')"
      ],
      "metadata": {
        "id": "w5Rev9XZkYG6"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "plt.figure(figsize = (6,3))\n",
        "plt.subplot(1,2,1)\n",
        "plot_image(5, predictions[5], test_labels, test_data.reshape(10000,28,28))\n",
        "plt.subplot(1,2,2)\n",
        "plot_value_array(5, predictions[5],  test_labels)\n",
        "plt.show()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 211
        },
        "outputId": "a5eba9c7-ed14-48c1-e82e-1fc6be5683bc",
        "id": "ywXOTTUUkYG6"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x216 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Caso 2"
      ],
      "metadata": {
        "id": "pzs-25Bvw50W"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Crear la red de neuronas"
      ],
      "metadata": {
        "id": "LUwP7hgxLwg7"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Se crea la red neuronal con una capa de entrada con 10 neuronas y función de activación ```relu```, y una capa de salida con función de activación ```softmax```."
      ],
      "metadata": {
        "id": "yTEzceXEQs__"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model2 = models.Sequential()\n",
        "model2.add(layers.Dense(10, activation ='relu', name = 'Capa2', input_shape = (28*28,)))\n",
        "model2.add(layers.Dense(10, name = 'CapaSalida2', activation = 'softmax'))"
      ],
      "metadata": {
        "id": "6s4rp0iWSGE_"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Explicar model.summary()\n"
      ],
      "metadata": {
        "id": "O7ykuaNzMnqg"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "El resumen del modelo indica el número de parámetros de salida de cada capa. Este número es igual al tamaño de los datos de entrada multiplicado por el número de neuronas, y sumando el número de umbrales de activación de la capa.\n",
        "\n",
        "\n",
        "En el caso de la primera capa: 7850 = 784 (tamaño de los datos) x 10 (número de neuronas) + 10 (número de umbrales de activación).\n",
        "\n",
        "En el caso de la segunda capa: 110 = 10 (tamaño de los datos) x 10 (número de neuronas) + 10 (número de umbrales de activación)."
      ],
      "metadata": {
        "id": "UAxZHzXQTk9Y"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model2.summary()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "jQlpbPf-Qdiv",
        "outputId": "e3ccef84-e596-4efa-a4c1-a5de3494b203"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential_7\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " Capa2 (Dense)               (None, 10)                7850      \n",
            "                                                                 \n",
            " CapaSalida2 (Dense)         (None, 10)                110       \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 7,960\n",
            "Trainable params: 7,960\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Se definen las funciones del optimizador, la pérdida y la métrica."
      ],
      "metadata": {
        "id": "3y2UTZhbwCor"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model2.compile(optimizer = 'rmsprop',\n",
        "              loss = 'categorical_crossentropy',\n",
        "              metrics = ['accuracy'])"
      ],
      "metadata": {
        "id": "uwIBFxNZVD5j"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Entrenar la red"
      ],
      "metadata": {
        "id": "2oFAQom1VEdi"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Para entrenar la red se realizan ```5``` ciclos con lotes de ```128``` imágenes. Se compara con los datos de validación para controlar el sobreentrenamiento. "
      ],
      "metadata": {
        "id": "D2YuIwa0yGsR"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "history = model2.fit(train_val_data, train_val_labels,\n",
        "                    epochs = 5,\n",
        "                    batch_size = 128,\n",
        "                    validation_data = (val_data, val_labels))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "MepKtlWWQrt1",
        "outputId": "128d36ea-b450-4734-b883-7a5f9212258d"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/5\n",
            "391/391 [==============================] - 2s 3ms/step - loss: 1.0597 - accuracy: 0.6798 - val_loss: 0.6837 - val_accuracy: 0.7774\n",
            "Epoch 2/5\n",
            "391/391 [==============================] - 1s 3ms/step - loss: 0.5852 - accuracy: 0.8014 - val_loss: 0.5436 - val_accuracy: 0.8184\n",
            "Epoch 3/5\n",
            "391/391 [==============================] - 1s 3ms/step - loss: 0.5170 - accuracy: 0.8251 - val_loss: 0.4980 - val_accuracy: 0.8317\n",
            "Epoch 4/5\n",
            "391/391 [==============================] - 1s 3ms/step - loss: 0.4833 - accuracy: 0.8366 - val_loss: 0.4880 - val_accuracy: 0.8338\n",
            "Epoch 5/5\n",
            "391/391 [==============================] - 1s 3ms/step - loss: 0.4629 - accuracy: 0.8413 - val_loss: 0.4710 - val_accuracy: 0.8388\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "La pérdida indica el porcentaje de errores entre las etiquetas y las predicciones del modelo.\n",
        "\n",
        "La precisión indica el porcentaje de aciertos entre las predicciones y las etiquetas."
      ],
      "metadata": {
        "id": "xByMAIeTPqjV"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "test_loss, test_acc = model2.evaluate(test_data, test_labels)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "zwHoOjklPmsa",
        "outputId": "47359e38-2ba4-497c-b60d-46a2372b2818"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 1s 2ms/step - loss: 0.5016 - accuracy: 0.8279\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import matplotlib.pyplot as plt\n",
        "\n",
        "acc = history.history['accuracy']\n",
        "val_acc = history.history['val_accuracy']\n",
        "loss = history.history['loss']\n",
        "val_loss = history.history['val_loss']\n",
        "\n",
        "epochs = range(1, len(acc) + 1)\n",
        "\n",
        "# \"bo\" es para \"punto azun\"\n",
        "plt.plot(epochs, loss, 'bo', label='Pérdida de Entrenamiento')\n",
        "# \"b\" es para \"linea continua azul\"\n",
        "plt.plot(epochs, val_loss, 'b', label='Pérdida de Validación')\n",
        "plt.title('Pérdida de Entrenamiento y Validación')\n",
        "plt.xlabel('Epochs')\n",
        "plt.ylabel('Loss')\n",
        "plt.legend()\n",
        "\n",
        "plt.show()"
      ],
      "metadata": {
        "id": "mSXFknhfyQlJ",
        "outputId": "518febad-bfe6-4a6f-c18b-c7d560aa1b26",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Predecir valores"
      ],
      "metadata": {
        "id": "wRx8hZhLQo8m"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Imprimir la matriz de confusión, que muestra los valores predichos frente a los reales."
      ],
      "metadata": {
        "id": "jeSpdtd88cuS"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Note, this code is taken straight from the SKLEARN website, an nice way of \n",
        "# viewing confusion matrix.\n",
        "def plot_confusion_matrix(cm, classes,\n",
        "                          normalize = False,\n",
        "                          title = 'Confusion matrix',\n",
        "                          cmap = plt.cm.Blues):\n",
        "    '''\n",
        "    This function prints and plots the confusion matrix.\n",
        "    Normalization can be applied by setting `normalize = True`.\n",
        "    '''\n",
        "    plt.imshow(cm, interpolation = 'nearest', cmap = cmap)\n",
        "    plt.title(title)\n",
        "    plt.colorbar()\n",
        "    tick_marks = np.arange(len(classes))\n",
        "    plt.xticks(tick_marks, classes, rotation = 45)\n",
        "    plt.yticks(tick_marks, classes)\n",
        "\n",
        "    if normalize:\n",
        "        cm = cm.astype('float') / cm.sum(axis = 1)[:, np.newaxis]\n",
        "\n",
        "    thresh = cm.max() / 2.\n",
        "    for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):\n",
        "        plt.text(j, i, cm[i, j],\n",
        "                 horizontalalignment = 'center',\n",
        "                 color = 'white' if cm[i, j] > thresh else 'black')\n",
        "\n",
        "    plt.tight_layout()\n",
        "    plt.ylabel('Observación')\n",
        "    plt.xlabel('Predicción')"
      ],
      "metadata": {
        "id": "ap-ozaX0llrR"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "Y_pred = model2.predict(test_data)\n",
        "print('Valores de predicción del modelo')\n",
        "# Convertir en predicciones de clases desde los vectores one-hot encoding \n",
        "Y_pred_classes = np.argmax(Y_pred, axis = 1) \n",
        "# Convertir en clases de observación desde los vectores one-hot encoding\n",
        "Y_true = np.argmax(test_labels, axis = 1) \n",
        "# Computar la matriz de confusión\n",
        "confusion_mtx = confusion_matrix(Y_true, Y_pred_classes) \n",
        "# Dibujar la matriz de confusión\n",
        "print('Matriz de confusión')\n",
        "plot_confusion_matrix(confusion_mtx, classes = range(10))"
      ],
      "metadata": {
        "id": "XshGwiutVsi9",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 366
        },
        "outputId": "9ba691fe-a00c-4c11-e956-6e8a661c6b92"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 0s 1ms/step\n",
            "Valores de predicción del modelo\n",
            "Matriz de confusión\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Se grafica el elemento 6 (```test_data[5]```) para comprobar las soluciones. "
      ],
      "metadata": {
        "id": "04i0fC4H8buV"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "digit = test_data.reshape(10000,28, 28)[5]\n",
        "plt.imshow(digit, cmap = plt.cm.binary)\n",
        "plt.show()\n",
        "print(test_labels[5])"
      ],
      "metadata": {
        "id": "9BbG8Tn8c07E",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 283
        },
        "outputId": "c8727897-afa8-41df-922b-08ec07dee688"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[0. 1. 0. 0. 0. 0. 0. 0. 0. 0.]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "La clase del elemento 6 es la clase 1 (pantalones)."
      ],
      "metadata": {
        "id": "QAMvqbA3ZOlT"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "predictions = model2.predict(test_data)\n",
        "print('Predicción del elemento 6: clase', np.argmax(predictions[5]))\n",
        "print(predictions[5])\n",
        "print('Suma de las predicciones del elemento 6:', np.sum(predictions[5]))"
      ],
      "metadata": {
        "id": "FhoAWXL1c64n",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "a01d0d1e-621f-4639-ef7e-a5acf948b648"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 0s 1ms/step\n",
            "Predicción del elemento 6: clase 1\n",
            "[4.6655824e-04 9.9097687e-01 4.4463715e-04 1.8717691e-04 7.8032305e-03\n",
            " 1.4819129e-10 1.2012521e-04 4.3562089e-13 1.4784405e-06 1.4691415e-10]\n",
            "Suma de las predicciones del elemento 6: 1.0000001\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "def plot_image(i, predictions_array, true_label, img):\n",
        "  predictions_array, true_label, img = predictions_array, true_label[i], img[i]\n",
        "  plt.grid(False)\n",
        "  plt.xticks([])\n",
        "  plt.yticks([])\n",
        "\n",
        "  plt.imshow(img, cmap = plt.cm.binary)    \n",
        "\n",
        "  predicted_label = np.argmax(predictions_array)\n",
        "  if predicted_label.all() == true_label.all():\n",
        "    color = 'blue'\n",
        "  else:\n",
        "    color = 'red'\n",
        "\n",
        "  plt.xlabel('{} {:2.0f}% ({})'.format(predicted_label,\n",
        "                                100*np.max(predictions_array),\n",
        "                                true_label),\n",
        "                                color = color)\n",
        "\n",
        "def plot_value_array(i, predictions_array, true_label):\n",
        "  predictions_array, true_label = predictions_array, true_label[i]\n",
        "  plt.grid(False)\n",
        "  plt.xticks(range(10))\n",
        "  plt.yticks([])\n",
        "  thisplot = plt.bar(range(10), predictions_array, color = '#00FF00')\n",
        "  plt.ylim([0, 1])\n",
        "  predicted_label = np.argmax(predictions_array)\n",
        "\n",
        "  thisplot[predicted_label].set_color('red')\n",
        "  thisplot[np.argmax(true_label)].set_color('black')"
      ],
      "metadata": {
        "id": "QLoUnoqIdHbd"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "plt.figure(figsize = (6,3))\n",
        "plt.subplot(1,2,1)\n",
        "plot_image(5, predictions[5], test_labels, test_data.reshape(10000,28,28))\n",
        "plt.subplot(1,2,2)\n",
        "plot_value_array(5, predictions[5],  test_labels)\n",
        "plt.show()"
      ],
      "metadata": {
        "id": "82gJvLq1_et8",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 211
        },
        "outputId": "e20ffebe-44fb-4ae9-f090-5753d581d2c5"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x216 with 2 Axes>"
            ],
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAWcAAADCCAYAAACVHgNkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAATJElEQVR4nO3df5BeVXnA8e9hk5BAQghJSJBfSTBqUUAgOLQgUikUo4XWYYTOqCC2aitVbJ0OiqOWtjPV0U7r2JFpAxSp4A+iRR2qxGLrHy0g0aQEJZhAkBCSEJIQAiE/n/5xb2CTe95977ss2ZPs9zPzTu4+e857zvtm93nP3nPOvSkikCSV5aDh7oAkqcnkLEkFMjlLUoFMzpJUIJOzJBXI5CxJBRo13B2QSjdlypSYMWPGcHdDB6iFCxeui4ipe8dNzlIXM2bM4P777x/S55w+fTpr1qxpVXbatGmsXr16SNtXOVJKj+XintaQhkHbxNxrWR04TM6SVCCTsyQVaMjOOR/IkyZLly5txFJK2bK5eO76JWPGjMnWz5Xdvn17tuxBB7X/bM097+zZs1vXH24rVqxg3bp1+TddOgANWXJ+JSZNSnHuuec2Yn19fdmyBx98cCP2wgsvNGKdPshyZTudc5wwYUIjtnPnzmzZXPzOO+/Mli3RnDlzhrsL0j7laQ1JKpDJWZIK5DrnfjZt2pSNP/jgg43Y1KmNNeMdbdmypRFbvnx5tuzYsWMbsU7nlg855JBGbNu2bdmyvfRX0vBz5CxJBTI5S1KBTM6SVCCTsyQVyOQsSQVytUY/uQ0gkN/112mzR27nXy42adKkbP3c83ZaRZJbxdFpc8u4ceOycUllcuQsSQUyOUtSgUzOklQgk7MkFcgJwX7mz5+fjT/99NON2DHHHJMtm5vQ27VrVyOWu3pdp7K57d8AO3bsaMSeeeaZbNlVq1Y1YgsXLsyWPf3007NxSfuOI2dJKpDJWZIKZHKWpAKZnCWpQCZnSSqQqzX6mTdvXjZ+1FFHNWJHHnlktmzufn+jRjXf5scffzxbP3cB/U73K8xdmD/XFsDatWsbsfvuuy9b1tUa0vBz5CxJBTI5S1KBTM6SVCCTsyQVyAnBfpYuXZqNz5kzpxHrtKV6+/btjVhuS/b48eOz9XN3z85t0waYOHFiqxjkr/2c29ItqQyOnCWpQCZnSSqQyVmSCmRylqQCmZwlqUAjdrXGk08+2Yh1uqN2bqt2bjs05FdF5O6+3Wn7dm5LdqeVHbmVIZ1WduSet9MF/yUNP0fOklQgk7MkFcjkLEkFMjlLUoFG7ITg6tWrG7HctZQ7iYhsfNy4cY3YunXrGrHclnCAJUuWNGKbN2/Ols1NFHaa1MxdEzo3SSipDI6cJalAJmdJKpDJWZIKZHKWpAKZnCWpQCN2tcbDDz/ciI0ePTpb9tBDD239vCmlRiy3VXz58uXZ+qeeemoj1ukmAMcff3wjltsqDvm7crt9WyqXI2dJKpDJWZIKZHKWpAKZnCWpQCN2QvChhx5qxDpt337uuecasdx2aICNGzc2YlOnTm3drzPPPLMRW7RoUbZsbvJx69atrct2mjyUNPwcOUtSgUzOklQgk7MkFcjkLEkFMjlLUoFG7GqNZcuWNWITJ07Mlt22bVsj1mmr96pVqxqxK664onW/rrzyykbs+uuvz5bdtWtX6+fNrS7ptOJE0vBz5CxJBTI5S1KBTM6SVCCTsyQVaMROCG7atKkRy905G/Jbn7dv354tm4tfffXVrft1xhlntGof8hOCnSb5ctdudkJQKpcjZ0kqkMlZkgpkcpakApmcJalAJmdJKtCIXa2Ruxv1+PHjs2VzqyW2bNmSLTt9+vRGbNasWT32bk9TpkzJxnOrNY444ohs2aeffroR6/QaJA0/R86SVCCTsyQVyOQsSQUyOUtSgUbshODkyZMbsR07drSuv3nz5mz8wgsvHHSfOslNMkJ++3WnycP169c3Yr1cD1rSvuXIWZIKZHKWpAKZnCWpQCZnSSrQiJ0QzO0G3LBhQ7ZsbqIwd4NYgC9+8Yut2u80GXfQQc3Py5kzZ2bLrly5shGbOnVqtuzOnTtb1ZdUBkfOklQgk7MkFcjkLEkFMjlLUoFMzpJUoBG7WiN3N+oXXnghWza3VTsismVPPPHEVu3nVk9AfrXG61//+mzZRx99tBGbMGFCtuxTTz3ViE2aNGmgLkoaRo6cJalAJmdJKpDJWZIKZHKWpAKN2AnBk046qRG79957s2VzE4WzZ8/Olu107eW95Sb+Opk7d242/qUvfakRe/7557NlV69e3Yh1uhmspOHnyFmSCmRylqQCmZwlqUAmZ0kqkMlZkgo0YldrXHrppY3YTTfdlC07alTzbdq0aVO27N13392IXXDBBY1Yp+3fOa973euy8WOPPbYR67QKJNfes88+27oPkvYtR86SVCCTsyQVyOQsSQUyOUtSgUbshGBfX18jNnr06GzZ3PWcc/UBbrnllkYsNyGYm2TsZMqUKdl4bkv2Y489li2bew1jx45t3QdJ+5YjZ0kqkMlZkgpkcpakApmcJalAJmdJKtCIXa2Rk1vRALBly5ZGrNNKh/vuu29I+zSQ3E0AFi5cmC27ffv2Riz3uiSVwZGzJBXI5CxJBTI5S1KBTM6SVCAnBPs566yzsvFbb721Eet05+oxY8YMaZ8GMmPGjEZsw4YN2bJbt25txHbu3DnUXZI0RBw5S1KBTM6SVCCTsyQVyOQsSQUyOUtSgVyt0c9VV12Vjd9+++2NWKe7XG/cuLERe+SRRxqxWbNm9di7pgkTJjRine6ovWvXrkZs0qRJL7sPkl4ZjpwlqUAmZ0kqkMlZkgpkcpakAjkh2M/RRx+djR9++OGNWKdrP2/btq0Ry13jeSgmBHNbxXfs2JEtm9u+neurpDI4cpakApmcJalAJmdJKpDJWZIKZHKWpAKN2NUaEdGIpZSyZc8///xGbP78+dmyuRUUd9xxRyN22WWXdetiV+PHj2/EVq1alS2be725Ld2SyuDIWZIKZHKWpAKZnCWpQCZnSSrQiJ0QzE2G9fX1ZcvOnTu3Ectd4xlg3LhxjdjKlSt77F07EydObMQ6bcnOXbt5/fr1Q94nSUPDkbMkFcjkLEkFMjlLUoFMzpJUIJOzJBVoxK7W6HT37Jyzzz67Eet0Yf7c3bdXr17diC1evDhb/5RTTmndr8MOO6wRe/7557NlR48e3YjlbiIgqQyOnCWpQCZnSSqQyVmSCmRylqQCjdgJwU7Xbm7ruOOOy8YXLVrUiOUm4xYsWJCt38uE4LPPPtuIbdmypXX9NWvWtC4rad9y5CxJBTI5S1KBTM6SVCCTsyQVyOQsSQUasas1Xq5rr702G58+fXojllut8Za3vOVl9+HSSy9txKZNm5Ytm9uqfd55573sPkh6ZThylqQCmZwlqUAmZ0kqkMlZkgqUImJoniilp4DHhuTJpKbjI2LqcDQ8Z86cuP/++4f0OXu9fMBQ/Z6qPCmlhRExZ+/4kK3WGK5fHEk6EHlaQ5IKZHKWpAKZnCWpQO4QlNRRov3EZeCk5VDqvlojpRuBdwBriXhDhzLHAzcCU4H1wLuJWFl/73PA2+uSf03EN+r414CTgO8T8ck69ilgCRH/3qGdU4GriHg/KX0W+GPgBiI+XU9//yMwF3geuIKIn3V5bVcBVwMnAFOJWDfwmwGkdGHdTh8wj4i/61J+MP2aCXwdmAwsBN5DxLYudS4HPlV/9TdE3Nyl/MHAV4HTgaeBS4lY0aXO6cC/AuOAO4GPdl1GkNIngPcDO4GPEPHDLuWPAL4BzABWAO8iYsNeZU4A5gOvJmI8KZ0E/AURVwz43IM0iJVIU4DuP0v7b73haPNArpdfiRQRAz/gnIDTApYMUOZbAZfXx28NuKU+fnvAgoBRAYcG/DTgsICTA+bVZRYETAw4KuB7XfryrYBT6uPPBny83/fmBvxHQAo4M+DeFq/t1IAZASsCprQo3xewPGBWwJiAxQEndqkzmH59M+Cy+vj6gD/pUv6IgEfqfyfVx5O61PnTgOvr48sCvtGiX/fVryHVr+ltXcqfWL9HBwfMrN+7vi51Ph9wTX18TcDnBii7ud/xjwKO6/oa9sEDuP9Arrc/9XV/qZd7dD/nHPETqtHwQE4E7q6Pfwxc3C/+EyJ2EPEc8H/AhcB2YBwpHQSMphpVXQd8pmMLKU0ATiZicYcSFwNfrV/ZPcDhpHRUl9f2c7qNFvf0JmAZEY9QjWS/zkuvtZPe+lWNtN8K3F5HbgZ+v0sbvwssIGI91ShzAdX73K1fu0fXtwPnDbj4turzYUTcU/8UfrVFvy4Gvk7EViIeBZZRvYdt+9Xmte/2PeCylmWl4g3VhOBi4J318R8AE0hpch2/kJQOIaUpwG8DxxLxS+Ap4GdUv1SvBg5i4D/35wBLBvj+0cDj/b5eWceG0mDa6LXOZGAjETtewTb2rFO19Uzd9kDlVw66jfZ1phHxZH28GshfZq/pfuDNLctKxRuqCcGPA18mpSuAnwBPADuJuIuUzgD+hyoZ/y/VKBkirn6xdkrfAz5IStcCp1CNAv9lrzaOqp9DI0VEkFLbWaa1wKteye704J8P8HrD0eaBXq9haEbOEauIeCcRpwLX1rGN9b9/S8QbiTgfSMDDe9RN6WKqSa/xwAlEvAu4hJQO2auVLcDYAXrxBHBsv6+PqWNDaTBt9FrnaapTH6Naln/5/aramli3PVD5YwbdRvs6a1487VP9u7ZL+d3GUv2MDLuIGNQv6P5SbzjaPNDr5QxNck5pSn3+GOATVCs3IKW++vQGpHQycDJwV796o6lWS3yeagXA7lFSHzBmr1Z+SXX6o5PvAu8lpURKZwLP9PvzeKj8FJhNSjNJaQzVOc7vdqnTW7+q87k/Bi6pI5cDd3Rp44fABaQ0iZQmARfUsW79urw+vgS4e/eMRod+PQlsIqUz63PT723Rr+8Cl5HSwfUKlNnAfT30q81r3+01DHzaS9q/dJ01hNsCngzYHrAy4P2ZMpcE/Crg4YB5AQfX8bEBv6gf9wS8ca96VwdcUR+nuq0HOs7QV9+bUB/vvVojBfxTvSLggYA5/b63qMPzfaR+TTsCVvVbQTLnxeNmnbn161wecG2/+HUBF2XKD6Zfs+qVEcuiWqGy+/28KOC6DnWurMsvC3hfv/i8Pdp8KT62fu5ldVuz6virAu7s0MacgCX1a/ly1EsxAz4U8KEOda6tyy/dY3UH3Bnwqkz5yQH/Wf88/SjgiI7/J3uu1vhywO8N1Uz5YB9UE7FLqSY/r2lZ50aqvxA6r4jK1zuW6oP8F8CDwEdb1htL9SG5uK73Vz222wf8HPh+D3VWAA8Ai+hhRQNwONWE9UNUA7TfbFnvtXVbux+bgKtb1v1Y/b4sAW4Dxras99G6zoNt2xrw+fbFD+yQPeBjAX9UH++ZnH2MvMfu5Fwt1bsnYNRw9qdOWsuBWVR/+S2m21LLqt45wGmDSM5HAafVxxOoThm2aS8B4+vj0cC9wJk9tPvnwK2DSM7dl6s2691M/Ttfv6eHD/L/ZTXVeuJuZY8GHgXG1V9/k90DyIHrvaFOzIdQzeX9CHj1y/l52t+2b38F2FofbwY+QErXDWN/NBxSOoGUFgFr6shxwDW8tMJluLwJWBYRj0T7pZZEu+WquXpPRr3CKSKepRpZdl2hFJXN9Zej60e0aTOldAzVprJ5vfa3VymliVQfXDcARMS22D2X1ZvzgOUR0XYj0ShgXKrmYg4BVrWo8xtUexiej+rn8L95aQXboOxfyTniBSJuqY+/QMRriPj0MPdK+1rEcqpJ5hPqr39FxH8Nb6eAfbOcMyulNAM4lWoU3KZ8X6o+4NYCCyKiVT3gH4C/BHb12MUA7kopLUwpfaBlnZlUK7RuSin9PKU0L6V0aI/tQjU3dFurTkY8AXwB+DXwJPBMRNw1cC2gGjW/OaU0OVWLGeay52R4z/av5CypIaU0nmo7+9URsalNnYjYGRFvpFpB86aUUv7SDHu28w5gbUQsHEQ3z46I04C3AR9OKZ3Tos4oqtM9X4lqJdhzwDW9NJqqifuLgG+1LD+J6q+dmVRLMw9NKb27W72o9m58jmrBww+oznPv7KWvezM5S0NnXyzn3EOqVjzNB74WEd/utX59muDHdN9RCnAWcFFKaQXVKZu3ppT+rWU7T9T/rgW+Q/edolD95bGy36j+dqpk3Yu3AT+LiDVdS1Z+B3g0Ip6KiO3At4HfalMxIm6IiNMj4hxgA3svG+6RyVkaOj8FZqeUZqb2Sy0HLVVLGm8AfhkRf99DvakppcPr43HA+VSrIQYUEZ+IiGMiYgbVa7s7IrqOKlNKh6bq8gvUpyUuoMWyx4hYDTyeUnptHTqPamVKL/6Qlqc0ar8GzkwpHVK/v+dRncvvKqV0ZP3vcVTnm2/tsa978JKh0hCJiB2putLhD6lWCNwYEQ92q5dSug04F5iSUloJfCYibmjR5FnAe4AH6vPHAJ+MiDu71DsKuDml1Ec1QPtmRHy/RXuDNQ34Tn3pllHArRHxg5Z1/wz4Wv1h9wjwvraN1h8E5wMfbFsnIu5NKd1OdWmJHVRLBttuLJmfqn0d24EPD3Ly8kVDdoNXSdLQ8bSGJBXI5CxJBTI5S1KBTM6SVCCTsyQVyOQsSQUyOUtSgUzOklSg/wfdjUOgItEHhgAAAABJRU5ErkJggg==\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Caso 2.5"
      ],
      "metadata": {
        "id": "mYLCqAC3rKbZ"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Crear la red de neuronas"
      ],
      "metadata": {
        "id": "MZlfbltzrKba"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Se crea la red neuronal con una capa de entrada con 10 neuronas y función de activación ```relu```, y una capa de salida con función de activación ```softmax```."
      ],
      "metadata": {
        "id": "m3rEuHQSrKbb"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model2b = models.Sequential()\n",
        "model2b.add(layers.Dense(10, activation ='relu', name = 'Capa2', input_shape = (28*28,)))\n",
        "model2b.add(layers.Dense(10, name = 'CapaSalida2', activation = 'softmax'))"
      ],
      "metadata": {
        "id": "pgakvJbnrKbb"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Explicar model.summary()\n"
      ],
      "metadata": {
        "id": "myOSwUnWrKbc"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "El resumen del modelo indica el número de parámetros de salida de cada capa. Este número es igual al tamaño de los datos de entrada multiplicado por el número de neuronas, y sumando el número de umbrales de activación de la capa.\n",
        "\n",
        "\n",
        "En el caso de la primera capa: 7850 = 784 (tamaño de los datos) x 10 (número de neuronas) + 10 (número de umbrales de activación).\n",
        "\n",
        "En el caso de la segunda capa: 110 = 10 (tamaño de los datos) x 10 (número de neuronas) + 10 (número de umbrales de activación)."
      ],
      "metadata": {
        "id": "UVmfqP4orKbc"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model2b.summary()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "35368eb5-214e-46b4-9b76-0c276c3e2b40",
        "id": "yZpkC9OvrKbc"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential_8\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " Capa2 (Dense)               (None, 10)                7850      \n",
            "                                                                 \n",
            " CapaSalida2 (Dense)         (None, 10)                110       \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 7,960\n",
            "Trainable params: 7,960\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Se definen las funciones del optimizador, la pérdida y la métrica."
      ],
      "metadata": {
        "id": "h37sm_XVrKbd"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model2b.compile(optimizer = 'rmsprop',\n",
        "              loss = 'categorical_crossentropy',\n",
        "              metrics = ['accuracy'])"
      ],
      "metadata": {
        "id": "KENFu-InrKbd"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Entrenar la red"
      ],
      "metadata": {
        "id": "WExv33SlrKbd"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Para entrenar la red se realizan ```5``` ciclos con lotes de ```128``` imágenes. Se compara con los datos de validación para controlar el sobreentrenamiento. "
      ],
      "metadata": {
        "id": "izEo9-gQrKbe"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "history = model2b.fit(train_val_data, train_val_labels,\n",
        "                    epochs = 25,\n",
        "                    batch_size = 128,\n",
        "                    validation_data = (val_data, val_labels))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "f6f1ebb7-bc4c-40f8-d986-c1c753892a5e",
        "id": "8pH8i2xJrKbe"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/25\n",
            "391/391 [==============================] - 2s 3ms/step - loss: 0.8909 - accuracy: 0.7082 - val_loss: 0.6161 - val_accuracy: 0.7985\n",
            "Epoch 2/25\n",
            "391/391 [==============================] - 1s 3ms/step - loss: 0.5596 - accuracy: 0.8126 - val_loss: 0.5445 - val_accuracy: 0.8117\n",
            "Epoch 3/25\n",
            "391/391 [==============================] - 1s 3ms/step - loss: 0.5051 - accuracy: 0.8276 - val_loss: 0.5113 - val_accuracy: 0.8241\n",
            "Epoch 4/25\n",
            "391/391 [==============================] - 1s 3ms/step - loss: 0.4766 - accuracy: 0.8367 - val_loss: 0.4706 - val_accuracy: 0.8387\n",
            "Epoch 5/25\n",
            "391/391 [==============================] - 1s 3ms/step - loss: 0.4581 - accuracy: 0.8423 - val_loss: 0.4681 - val_accuracy: 0.8340\n",
            "Epoch 6/25\n",
            "391/391 [==============================] - 1s 3ms/step - loss: 0.4454 - accuracy: 0.8484 - val_loss: 0.4478 - val_accuracy: 0.8478\n",
            "Epoch 7/25\n",
            "391/391 [==============================] - 1s 3ms/step - loss: 0.4354 - accuracy: 0.8503 - val_loss: 0.4498 - val_accuracy: 0.8458\n",
            "Epoch 8/25\n",
            "391/391 [==============================] - 1s 3ms/step - loss: 0.4285 - accuracy: 0.8527 - val_loss: 0.4394 - val_accuracy: 0.8492\n",
            "Epoch 9/25\n",
            "391/391 [==============================] - 1s 3ms/step - loss: 0.4217 - accuracy: 0.8547 - val_loss: 0.4458 - val_accuracy: 0.8452\n",
            "Epoch 10/25\n",
            "391/391 [==============================] - 1s 3ms/step - loss: 0.4180 - accuracy: 0.8569 - val_loss: 0.4269 - val_accuracy: 0.8502\n",
            "Epoch 11/25\n",
            "391/391 [==============================] - 1s 3ms/step - loss: 0.4122 - accuracy: 0.8592 - val_loss: 0.4247 - val_accuracy: 0.8566\n",
            "Epoch 12/25\n",
            "391/391 [==============================] - 1s 3ms/step - loss: 0.4083 - accuracy: 0.8590 - val_loss: 0.4299 - val_accuracy: 0.8521\n",
            "Epoch 13/25\n",
            "391/391 [==============================] - 1s 3ms/step - loss: 0.4054 - accuracy: 0.8598 - val_loss: 0.4237 - val_accuracy: 0.8553\n",
            "Epoch 14/25\n",
            "391/391 [==============================] - 1s 3ms/step - loss: 0.4016 - accuracy: 0.8613 - val_loss: 0.4197 - val_accuracy: 0.8567\n",
            "Epoch 15/25\n",
            "391/391 [==============================] - 1s 3ms/step - loss: 0.3987 - accuracy: 0.8604 - val_loss: 0.4687 - val_accuracy: 0.8435\n",
            "Epoch 16/25\n",
            "391/391 [==============================] - 1s 3ms/step - loss: 0.3954 - accuracy: 0.8638 - val_loss: 0.4259 - val_accuracy: 0.8511\n",
            "Epoch 17/25\n",
            "391/391 [==============================] - 1s 3ms/step - loss: 0.3932 - accuracy: 0.8641 - val_loss: 0.4121 - val_accuracy: 0.8565\n",
            "Epoch 18/25\n",
            "391/391 [==============================] - 1s 3ms/step - loss: 0.3911 - accuracy: 0.8641 - val_loss: 0.4130 - val_accuracy: 0.8585\n",
            "Epoch 19/25\n",
            "391/391 [==============================] - 1s 3ms/step - loss: 0.3893 - accuracy: 0.8650 - val_loss: 0.4160 - val_accuracy: 0.8546\n",
            "Epoch 20/25\n",
            "391/391 [==============================] - 1s 3ms/step - loss: 0.3870 - accuracy: 0.8654 - val_loss: 0.4120 - val_accuracy: 0.8550\n",
            "Epoch 21/25\n",
            "391/391 [==============================] - 1s 3ms/step - loss: 0.3839 - accuracy: 0.8668 - val_loss: 0.4384 - val_accuracy: 0.8435\n",
            "Epoch 22/25\n",
            "391/391 [==============================] - 1s 3ms/step - loss: 0.3829 - accuracy: 0.8660 - val_loss: 0.4125 - val_accuracy: 0.8566\n",
            "Epoch 23/25\n",
            "391/391 [==============================] - 1s 3ms/step - loss: 0.3805 - accuracy: 0.8670 - val_loss: 0.4086 - val_accuracy: 0.8583\n",
            "Epoch 24/25\n",
            "391/391 [==============================] - 1s 3ms/step - loss: 0.3793 - accuracy: 0.8673 - val_loss: 0.4068 - val_accuracy: 0.8596\n",
            "Epoch 25/25\n",
            "391/391 [==============================] - 1s 3ms/step - loss: 0.3772 - accuracy: 0.8686 - val_loss: 0.4106 - val_accuracy: 0.8569\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "La pérdida indica el porcentaje de errores entre las etiquetas y las predicciones del modelo.\n",
        "\n",
        "La precisión indica el porcentaje de aciertos entre las predicciones y las etiquetas."
      ],
      "metadata": {
        "id": "CNaK1etArKbf"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "test_loss, test_acc = model2b.evaluate(test_data, test_labels)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "4896a22e-9b24-4c5e-f550-6ca5a453c424",
        "id": "AEKJDE1ArKbf"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 1s 2ms/step - loss: 0.4402 - accuracy: 0.8456\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "acc = history.history['accuracy']\n",
        "val_acc = history.history['val_accuracy']\n",
        "loss = history.history['loss']\n",
        "val_loss = history.history['val_loss']\n",
        "\n",
        "epochs = range(1, len(acc) + 1)\n",
        "\n",
        "# \"bo\" es para \"punto azun\"\n",
        "plt.plot(epochs, loss, 'bo', label='Pérdida de Entrenamiento')\n",
        "# \"b\" es para \"linea continua azul\"\n",
        "plt.plot(epochs, val_loss, 'b', label='Pérdida de Validación')\n",
        "plt.title('Pérdida de Entrenamiento y Validación')\n",
        "plt.xlabel('Epochs')\n",
        "plt.ylabel('Loss')\n",
        "plt.legend()\n",
        "\n",
        "plt.show()"
      ],
      "metadata": {
        "id": "tcCW6qXuyXcQ",
        "outputId": "98ee4fe1-43cc-4723-963c-a74d0a3e28e3",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Predecir valores"
      ],
      "metadata": {
        "id": "qSQtDu3HrKbf"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Imprimir la matriz de confusión, que muestra los valores predichos frente a los reales."
      ],
      "metadata": {
        "id": "syJQS7rnrKbg"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Note, this code is taken straight from the SKLEARN website, an nice way of \n",
        "# viewing confusion matrix.\n",
        "def plot_confusion_matrix(cm, classes,\n",
        "                          normalize = False,\n",
        "                          title = 'Confusion matrix',\n",
        "                          cmap = plt.cm.Blues):\n",
        "    '''\n",
        "    This function prints and plots the confusion matrix.\n",
        "    Normalization can be applied by setting `normalize = True`.\n",
        "    '''\n",
        "    plt.imshow(cm, interpolation = 'nearest', cmap = cmap)\n",
        "    plt.title(title)\n",
        "    plt.colorbar()\n",
        "    tick_marks = np.arange(len(classes))\n",
        "    plt.xticks(tick_marks, classes, rotation = 45)\n",
        "    plt.yticks(tick_marks, classes)\n",
        "\n",
        "    if normalize:\n",
        "        cm = cm.astype('float') / cm.sum(axis = 1)[:, np.newaxis]\n",
        "\n",
        "    thresh = cm.max() / 2.\n",
        "    for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):\n",
        "        plt.text(j, i, cm[i, j],\n",
        "                 horizontalalignment = 'center',\n",
        "                 color = 'white' if cm[i, j] > thresh else 'black')\n",
        "\n",
        "    plt.tight_layout()\n",
        "    plt.ylabel('Observación')\n",
        "    plt.xlabel('Predicción')"
      ],
      "metadata": {
        "id": "1JtGWUtNrKbg"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "Y_pred = model2b.predict(test_data)\n",
        "print('Valores de predicción del modelo')\n",
        "# Convertir en predicciones de clases desde los vectores one-hot encoding \n",
        "Y_pred_classes = np.argmax(Y_pred, axis = 1) \n",
        "# Convertir en clases de observación desde los vectores one-hot encoding\n",
        "Y_true = np.argmax(test_labels, axis = 1) \n",
        "# Computar la matriz de confusión\n",
        "confusion_mtx = confusion_matrix(Y_true, Y_pred_classes) \n",
        "# Dibujar la matriz de confusión\n",
        "print('Matriz de confusión')\n",
        "plot_confusion_matrix(confusion_mtx, classes = range(10))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 366
        },
        "outputId": "1b621b16-7f72-41ba-dd30-6330babdc72e",
        "id": "gkSvMpqZrKbg"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 0s 1ms/step\n",
            "Valores de predicción del modelo\n",
            "Matriz de confusión\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Se grafica el elemento 6 (```test_data[5]```) para comprobar las soluciones. "
      ],
      "metadata": {
        "id": "wyVJoZlwrKbh"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "digit = test_data.reshape(10000,28, 28)[5]\n",
        "plt.imshow(digit, cmap = plt.cm.binary)\n",
        "plt.show()\n",
        "print(test_labels[5])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 283
        },
        "outputId": "d7c6bafb-21ea-4e5e-926f-1e3817f39a71",
        "id": "kib4BvBXrKbh"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[0. 1. 0. 0. 0. 0. 0. 0. 0. 0.]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "La clase del elemento 6 es la clase 1 (pantalones)."
      ],
      "metadata": {
        "id": "XAQ1AWWHrKbi"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "predictions = model2b.predict(test_data)\n",
        "print('Predicción del elemento 6: clase', np.argmax(predictions[5]))\n",
        "print(predictions[5])\n",
        "print('Suma de las predicciones del elemento 6:', np.sum(predictions[5]))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "459e2e0e-b08f-4111-b856-cd9dafc37466",
        "id": "FBIVo56OrKbi"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 0s 1ms/step\n",
            "Predicción del elemento 6: clase 1\n",
            "[3.9456324e-03 9.9470371e-01 2.0176769e-04 3.5539505e-04 7.7885954e-04\n",
            " 4.2394350e-17 1.4595930e-05 3.5434653e-09 1.0422614e-08 3.9521919e-09]\n",
            "Suma de las predicciones del elemento 6: 1.0\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "def plot_image(i, predictions_array, true_label, img):\n",
        "  predictions_array, true_label, img = predictions_array, true_label[i], img[i]\n",
        "  plt.grid(False)\n",
        "  plt.xticks([])\n",
        "  plt.yticks([])\n",
        "\n",
        "  plt.imshow(img, cmap = plt.cm.binary)    \n",
        "\n",
        "  predicted_label = np.argmax(predictions_array)\n",
        "  if predicted_label.all() == true_label.all():\n",
        "    color = 'blue'\n",
        "  else:\n",
        "    color = 'red'\n",
        "\n",
        "  plt.xlabel('{} {:2.0f}% ({})'.format(predicted_label,\n",
        "                                100*np.max(predictions_array),\n",
        "                                true_label),\n",
        "                                color = color)\n",
        "\n",
        "def plot_value_array(i, predictions_array, true_label):\n",
        "  predictions_array, true_label = predictions_array, true_label[i]\n",
        "  plt.grid(False)\n",
        "  plt.xticks(range(10))\n",
        "  plt.yticks([])\n",
        "  thisplot = plt.bar(range(10), predictions_array, color = '#00FF00')\n",
        "  plt.ylim([0, 1])\n",
        "  predicted_label = np.argmax(predictions_array)\n",
        "\n",
        "  thisplot[predicted_label].set_color('red')\n",
        "  thisplot[np.argmax(true_label)].set_color('black')"
      ],
      "metadata": {
        "id": "3B0qQ9WxrKbi"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "plt.figure(figsize = (6,3))\n",
        "plt.subplot(1,2,1)\n",
        "plot_image(5, predictions[5], test_labels, test_data.reshape(10000,28,28))\n",
        "plt.subplot(1,2,2)\n",
        "plot_value_array(5, predictions[5],  test_labels)\n",
        "plt.show()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 211
        },
        "outputId": "7bb7bd27-11cd-4f6d-8d03-6912cc1e04b9",
        "id": "3d4DxyPQrKbj"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x216 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Caso 3"
      ],
      "metadata": {
        "id": "IiqUjYKEl4G0"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Crear la red de neuronas"
      ],
      "metadata": {
        "id": "g2tAAPrHl4G1"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Se crea la red neuronal con una capa de entrada con 10 neuronas y función de activación ```sigmoid```, y una capa de salida con función de activación ```softmax```."
      ],
      "metadata": {
        "id": "EdX8meU-l4G1"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model3 = models.Sequential()\n",
        "model3.add(layers.Dense(10, activation ='sigmoid', name = 'Capa3', input_shape = (28*28,)))\n",
        "model3.add(layers.Dense(10, name = 'CapaSalida3', activation = 'softmax'))"
      ],
      "metadata": {
        "id": "-bPneutFl4G1"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Explicar model.summary()\n"
      ],
      "metadata": {
        "id": "0NmJbS4fl4G1"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "El resumen del modelo indica el número de parámetros de salida de cada capa. Este número es igual al tamaño de los datos de entrada multiplicado por el número de neuronas, y sumando el número de umbrales de activación de la capa.\n",
        "\n",
        "\n",
        "En el caso de la primera capa: 7850 = 784 (tamaño de los datos) x 10 (número de neuronas) + 10 (número de umbrales de activación).\n",
        "\n",
        "En el caso de la segunda capa: 110 = 10 (tamaño de los datos) x 10 (número de neuronas) + 10 (número de umbrales de activación)."
      ],
      "metadata": {
        "id": "-FTX4egil4G1"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model3.summary()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "f6d70845-9783-40f5-d0f2-bcff31e8aaf4",
        "id": "6_qB0a08l4G1"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential_9\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " Capa3 (Dense)               (None, 10)                7850      \n",
            "                                                                 \n",
            " CapaSalida3 (Dense)         (None, 10)                110       \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 7,960\n",
            "Trainable params: 7,960\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Se definen las funciones del optimizador, la pérdida y la métrica."
      ],
      "metadata": {
        "id": "CEThyLqdl4G1"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model3.compile(optimizer = 'sgd',\n",
        "              loss = 'categorical_crossentropy',\n",
        "              metrics = ['accuracy'])"
      ],
      "metadata": {
        "id": "NlWa9sPel4G1"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Entrenar la red"
      ],
      "metadata": {
        "id": "thul5Sspl4G1"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Para entrenar la red se realizan ```5``` ciclos con lotes de ```128``` imágenes. Se compara con los datos de validación para controlar el sobreentrenamiento. "
      ],
      "metadata": {
        "id": "idPqmt4Wl4G1"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "history = model3.fit(train_val_data, train_val_labels,\n",
        "                    epochs = 5,\n",
        "                    batch_size = 128,\n",
        "                    validation_data = (val_data, val_labels))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "e955d2b7-9749-4cca-f52f-e445995c5cfd",
        "id": "9Pd1VScol4G2"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/5\n",
            "391/391 [==============================] - 2s 4ms/step - loss: 2.2132 - accuracy: 0.2310 - val_loss: 2.0616 - val_accuracy: 0.3602\n",
            "Epoch 2/5\n",
            "391/391 [==============================] - 1s 4ms/step - loss: 1.9545 - accuracy: 0.4360 - val_loss: 1.8392 - val_accuracy: 0.4891\n",
            "Epoch 3/5\n",
            "391/391 [==============================] - 1s 3ms/step - loss: 1.7462 - accuracy: 0.5353 - val_loss: 1.6525 - val_accuracy: 0.5731\n",
            "Epoch 4/5\n",
            "391/391 [==============================] - 1s 3ms/step - loss: 1.5757 - accuracy: 0.5902 - val_loss: 1.5012 - val_accuracy: 0.6101\n",
            "Epoch 5/5\n",
            "391/391 [==============================] - 1s 3ms/step - loss: 1.4387 - accuracy: 0.6156 - val_loss: 1.3802 - val_accuracy: 0.6311\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "La pérdida indica el porcentaje de errores entre las etiquetas y las predicciones del modelo.\n",
        "\n",
        "La precisión indica el porcentaje de aciertos entre las predicciones y las etiquetas."
      ],
      "metadata": {
        "id": "AmvMmmwal4G2"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "test_loss, test_acc = model3.evaluate(test_data, test_labels)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "db408258-8319-4dee-ed36-7816fda0c27a",
        "id": "57kpkxnKl4G2"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 1s 2ms/step - loss: 1.3849 - accuracy: 0.6245\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "acc = history.history['accuracy']\n",
        "val_acc = history.history['val_accuracy']\n",
        "loss = history.history['loss']\n",
        "val_loss = history.history['val_loss']\n",
        "\n",
        "epochs = range(1, len(acc) + 1)\n",
        "\n",
        "# \"bo\" es para \"punto azun\"\n",
        "plt.plot(epochs, loss, 'bo', label='Pérdida de Entrenamiento')\n",
        "# \"b\" es para \"linea continua azul\"\n",
        "plt.plot(epochs, val_loss, 'b', label='Pérdida de Validación')\n",
        "plt.title('Pérdida de Entrenamiento y Validación')\n",
        "plt.xlabel('Epochs')\n",
        "plt.ylabel('Loss')\n",
        "plt.legend()\n",
        "\n",
        "plt.show()"
      ],
      "metadata": {
        "id": "IRszGBtPy_iJ",
        "outputId": "6054f02c-b0bb-44d3-f4bd-3755bb12a0be",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Predecir valores"
      ],
      "metadata": {
        "id": "URWllf3ll4G2"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Imprimir la matriz de confusión, que muestra los valores predichos frente a los reales."
      ],
      "metadata": {
        "id": "_OBRybhrl4G2"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Note, this code is taken straight from the SKLEARN website, an nice way of \n",
        "# viewing confusion matrix.\n",
        "def plot_confusion_matrix(cm, classes,\n",
        "                          normalize = False,\n",
        "                          title = 'Confusion matrix',\n",
        "                          cmap = plt.cm.Blues):\n",
        "    '''\n",
        "    This function prints and plots the confusion matrix.\n",
        "    Normalization can be applied by setting `normalize = True`.\n",
        "    '''\n",
        "    plt.imshow(cm, interpolation = 'nearest', cmap = cmap)\n",
        "    plt.title(title)\n",
        "    plt.colorbar()\n",
        "    tick_marks = np.arange(len(classes))\n",
        "    plt.xticks(tick_marks, classes, rotation = 45)\n",
        "    plt.yticks(tick_marks, classes)\n",
        "\n",
        "    if normalize:\n",
        "        cm = cm.astype('float') / cm.sum(axis = 1)[:, np.newaxis]\n",
        "\n",
        "    thresh = cm.max() / 2.\n",
        "    for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):\n",
        "        plt.text(j, i, cm[i, j],\n",
        "                 horizontalalignment = 'center',\n",
        "                 color = 'white' if cm[i, j] > thresh else 'black')\n",
        "\n",
        "    plt.tight_layout()\n",
        "    plt.ylabel('Observación')\n",
        "    plt.xlabel('Predicción')"
      ],
      "metadata": {
        "id": "Iowk6cHEl4G2"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "Y_pred = model.predict(test_data)\n",
        "print('Valores de predicción del modelo')\n",
        "# Convertir en predicciones de clases desde los vectores one-hot encoding \n",
        "Y_pred_classes = np.argmax(Y_pred, axis = 1) \n",
        "# Convertir en clases de observación desde los vectores one-hot encoding\n",
        "Y_true = np.argmax(test_labels, axis = 1) \n",
        "# Computar la matriz de confusión\n",
        "confusion_mtx = confusion_matrix(Y_true, Y_pred_classes) \n",
        "# Dibujar la matriz de confusión\n",
        "print('Matriz de confusión')\n",
        "plot_confusion_matrix(confusion_mtx, classes = range(10))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 366
        },
        "outputId": "c0e03d09-be1c-4578-8f55-3ee0f64653f0",
        "id": "smiumUGxl4G2"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 0s 1ms/step\n",
            "Valores de predicción del modelo\n",
            "Matriz de confusión\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Se grafica el elemento 6 (```test_data[5]```) para comprobar las soluciones. "
      ],
      "metadata": {
        "id": "ytD9o7lnl4G2"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "digit = test_data.reshape(10000,28, 28)[5]\n",
        "plt.imshow(digit, cmap = plt.cm.binary)\n",
        "plt.show()\n",
        "print(test_labels[5])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 283
        },
        "outputId": "13866096-9429-4670-ce52-3180ba60330f",
        "id": "68xLwYf2l4G2"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[0. 1. 0. 0. 0. 0. 0. 0. 0. 0.]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "La clase del elemento 6 es la clase 1 (pantalones)."
      ],
      "metadata": {
        "id": "VHJVmjX5l4G2"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "predictions = model.predict(test_data)\n",
        "print('Predicción del elemento 6: clase', np.argmax(predictions[5]))\n",
        "print(predictions[5])\n",
        "print('Suma de las predicciones del elemento 6:', np.sum(predictions[5]))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "2034d342-7db5-4616-978a-7feca3a30b6d",
        "id": "q8E7EoEsl4G2"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 0s 1ms/step\n",
            "Predicción del elemento 6: clase 1\n",
            "[1.5701486e-03 9.6329826e-01 3.6851829e-04 1.0837999e-02 1.9751115e-02\n",
            " 1.7835138e-05 4.0504690e-03 8.1603874e-05 2.3705077e-06 2.1786480e-05]\n",
            "Suma de las predicciones del elemento 6: 1.0000001\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "def plot_image(i, predictions_array, true_label, img):\n",
        "  predictions_array, true_label, img = predictions_array, true_label[i], img[i]\n",
        "  plt.grid(False)\n",
        "  plt.xticks([])\n",
        "  plt.yticks([])\n",
        "\n",
        "  plt.imshow(img, cmap = plt.cm.binary)    \n",
        "\n",
        "  predicted_label = np.argmax(predictions_array)\n",
        "  if predicted_label.all() == true_label.all():\n",
        "    color = 'blue'\n",
        "  else:\n",
        "    color = 'red'\n",
        "\n",
        "  plt.xlabel('{} {:2.0f}% ({})'.format(predicted_label,\n",
        "                                100*np.max(predictions_array),\n",
        "                                true_label),\n",
        "                                color = color)\n",
        "\n",
        "def plot_value_array(i, predictions_array, true_label):\n",
        "  predictions_array, true_label = predictions_array, true_label[i]\n",
        "  plt.grid(False)\n",
        "  plt.xticks(range(10))\n",
        "  plt.yticks([])\n",
        "  thisplot = plt.bar(range(10), predictions_array, color = '#00FF00')\n",
        "  plt.ylim([0, 1])\n",
        "  predicted_label = np.argmax(predictions_array)\n",
        "\n",
        "  thisplot[predicted_label].set_color('red')\n",
        "  thisplot[np.argmax(true_label)].set_color('black')"
      ],
      "metadata": {
        "id": "donQW3jdl4G3"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "plt.figure(figsize = (6,3))\n",
        "plt.subplot(1,2,1)\n",
        "plot_image(5, predictions[5], test_labels, test_data.reshape(10000,28,28))\n",
        "plt.subplot(1,2,2)\n",
        "plot_value_array(5, predictions[5],  test_labels)\n",
        "plt.show()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 211
        },
        "outputId": "fe902807-aaa2-4568-f6e2-366aa5b7c5ac",
        "id": "PriRLEtYl4G3"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x216 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Caso 3.5"
      ],
      "metadata": {
        "id": "N7JYgzp1rd3G"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Crear la red de neuronas"
      ],
      "metadata": {
        "id": "QK-V2H0Rrd3G"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Se crea la red neuronal con una capa de entrada con 10 neuronas y función de activación ```sigmoid```, y una capa de salida con función de activación ```softmax```."
      ],
      "metadata": {
        "id": "8dqCjBo5rd3H"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "\n",
        "model3b = models.Sequential()\n",
        "#Capa 3b (sparse_categorical_crossentropy)\n",
        "model3b.add(layers.Dense(10, activation='sigmoid', name=\"Capa3b\",input_shape=(28*28,)))\n",
        "#Capa final\n",
        "model3b.add(layers.Dense(10, name=\"CapaSalida\",activation='softmax'))\n",
        "model3b.add(layers.Flatten())\n",
        "model3b.output_shape"
      ],
      "metadata": {
        "id": "C6cDwHcurd3H",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "768b518e-b59b-49a9-e615-38d4a0ac0f80"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(None, 10)"
            ]
          },
          "metadata": {},
          "execution_count": 120
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Explicar model.summary()\n"
      ],
      "metadata": {
        "id": "NFLzryB3rd3I"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "El resumen del modelo indica el número de parámetros de salida de cada capa. Este número es igual al tamaño de los datos de entrada multiplicado por el número de neuronas, y sumando el número de umbrales de activación de la capa.\n",
        "\n",
        "\n",
        "En el caso de la primera capa: 7850 = 784 (tamaño de los datos) x 10 (número de neuronas) + 10 (número de umbrales de activación).\n",
        "\n",
        "En el caso de la segunda capa: 110 = 10 (tamaño de los datos) x 10 (número de neuronas) + 10 (número de umbrales de activación)."
      ],
      "metadata": {
        "id": "WFxnmnlgrd3I"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model3b.summary()\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "c0713fbf-1093-466f-f263-8b51587e1f35",
        "id": "eEkK-CgVrd3I"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential_10\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " Capa3b (Dense)              (None, 10)                7850      \n",
            "                                                                 \n",
            " CapaSalida (Dense)          (None, 10)                110       \n",
            "                                                                 \n",
            " flatten_3 (Flatten)         (None, 10)                0         \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 7,960\n",
            "Trainable params: 7,960\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Se definen las funciones del optimizador, la pérdida y la métrica."
      ],
      "metadata": {
        "id": "TE6uPAGdrd3J"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Probamos que pasaría si cambiamos el optimizador por 'adam' y la funcion de perdida por 'sparse_categorical_crossentropy'. El optimizador adam es una combinacion de AdaGrad y RMSProp ya que además de RMSProp, cada factor de entrenamiento también se ve afectado por el momentum del gradiente. (https://respuestasrapidas.com.mx/que-es-optimizador-adam/) La función de pérdida sparse_categorical_crossentropy, es una medida de distancia entre probabilidades al igual que categorical_crossentropy pero este caso es usado unicamente para numeros enteros.(https://ignaciogavilan.com/catalogo-de-componentes-de-redes-neuronales-iii-funciones-de-perdida/)"
      ],
      "metadata": {
        "id": "ahlaSxFDxfF4"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model3b.compile(optimizer='adam',\n",
        "              loss='sparse_categorical_crossentropy',\n",
        "              metrics=['accuracy'])"
      ],
      "metadata": {
        "id": "0x7r1BtUrd3J"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Entrenar la red"
      ],
      "metadata": {
        "id": "r29ixCBzrd3J"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Para entrenar la red se realizan ```5``` ciclos con lotes de ```128``` imágenes. Se compara con los datos de validación para controlar el sobreentrenamiento. "
      ],
      "metadata": {
        "id": "LWuLgQZWrd3K"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model3b.fit(train_data, train_labels, epochs=5, batch_size=128)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "5c7d6ed8-5ef5-4682-b2e7-d288a3c31520",
        "id": "2otyXG0Prd3K"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/5\n"
          ]
        },
        {
          "output_type": "error",
          "ename": "InvalidArgumentError",
          "evalue": "ignored",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mInvalidArgumentError\u001b[0m                      Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-201-9f09bdbb5632>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mmodel3b\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtrain_data\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtrain_labels\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mepochs\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m5\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mbatch_size\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m128\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mvalidation_data\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m(\u001b[0m\u001b[0mval_data\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mval_labels\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/keras/utils/traceback_utils.py\u001b[0m in \u001b[0;36merror_handler\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m     65\u001b[0m     \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m  \u001b[0;31m# pylint: disable=broad-except\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     66\u001b[0m       \u001b[0mfiltered_tb\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0m_process_traceback_frames\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0me\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m__traceback__\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 67\u001b[0;31m       \u001b[0;32mraise\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mwith_traceback\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfiltered_tb\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mfrom\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     68\u001b[0m     \u001b[0;32mfinally\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     69\u001b[0m       \u001b[0;32mdel\u001b[0m \u001b[0mfiltered_tb\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/execute.py\u001b[0m in \u001b[0;36mquick_execute\u001b[0;34m(op_name, num_outputs, inputs, attrs, ctx, name)\u001b[0m\n\u001b[1;32m     53\u001b[0m     \u001b[0mctx\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mensure_initialized\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     54\u001b[0m     tensors = pywrap_tfe.TFE_Py_Execute(ctx._handle, device_name, op_name,\n\u001b[0;32m---> 55\u001b[0;31m                                         inputs, attrs, num_outputs)\n\u001b[0m\u001b[1;32m     56\u001b[0m   \u001b[0;32mexcept\u001b[0m \u001b[0mcore\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_NotOkStatusException\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     57\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mname\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mInvalidArgumentError\u001b[0m: Graph execution error:\n\nDetected at node 'sparse_categorical_crossentropy/SparseSoftmaxCrossEntropyWithLogits/SparseSoftmaxCrossEntropyWithLogits' defined at (most recent call last):\n    File \"/usr/lib/python3.7/runpy.py\", line 193, in _run_module_as_main\n      \"__main__\", mod_spec)\n    File \"/usr/lib/python3.7/runpy.py\", line 85, in _run_code\n      exec(code, run_globals)\n    File \"/usr/local/lib/python3.7/dist-packages/ipykernel_launcher.py\", line 16, in <module>\n      app.launch_new_instance()\n    File \"/usr/local/lib/python3.7/dist-packages/traitlets/config/application.py\", line 846, in launch_instance\n      app.start()\n    File \"/usr/local/lib/python3.7/dist-packages/ipykernel/kernelapp.py\", line 612, in start\n      self.io_loop.start()\n    File \"/usr/local/lib/python3.7/dist-packages/tornado/platform/asyncio.py\", line 132, in start\n      self.asyncio_loop.run_forever()\n    File \"/usr/lib/python3.7/asyncio/base_events.py\", line 541, in run_forever\n      self._run_once()\n    File \"/usr/lib/python3.7/asyncio/base_events.py\", line 1786, in _run_once\n      handle._run()\n    File \"/usr/lib/python3.7/asyncio/events.py\", line 88, in _run\n      self._context.run(self._callback, *self._args)\n    File \"/usr/local/lib/python3.7/dist-packages/tornado/ioloop.py\", line 758, in _run_callback\n      ret = callback()\n    File \"/usr/local/lib/python3.7/dist-packages/tornado/stack_context.py\", line 300, in null_wrapper\n      return fn(*args, **kwargs)\n    File \"/usr/local/lib/python3.7/dist-packages/tornado/gen.py\", line 1233, in inner\n      self.run()\n    File \"/usr/local/lib/python3.7/dist-packages/tornado/gen.py\", line 1147, in run\n      yielded = self.gen.send(value)\n    File \"/usr/local/lib/python3.7/dist-packages/ipykernel/kernelbase.py\", line 381, in dispatch_queue\n      yield self.process_one()\n    File \"/usr/local/lib/python3.7/dist-packages/tornado/gen.py\", line 346, in wrapper\n      runner = Runner(result, future, yielded)\n    File \"/usr/local/lib/python3.7/dist-packages/tornado/gen.py\", line 1080, in __init__\n      self.run()\n    File \"/usr/local/lib/python3.7/dist-packages/tornado/gen.py\", line 1147, in run\n      yielded = self.gen.send(value)\n    File \"/usr/local/lib/python3.7/dist-packages/ipykernel/kernelbase.py\", line 365, in process_one\n      yield gen.maybe_future(dispatch(*args))\n    File \"/usr/local/lib/python3.7/dist-packages/tornado/gen.py\", line 326, in wrapper\n      yielded = next(result)\n    File \"/usr/local/lib/python3.7/dist-packages/ipykernel/kernelbase.py\", line 268, in dispatch_shell\n      yield gen.maybe_future(handler(stream, idents, msg))\n    File \"/usr/local/lib/python3.7/dist-packages/tornado/gen.py\", line 326, in wrapper\n      yielded = next(result)\n    File \"/usr/local/lib/python3.7/dist-packages/ipykernel/kernelbase.py\", line 545, in execute_request\n      user_expressions, allow_stdin,\n    File \"/usr/local/lib/python3.7/dist-packages/tornado/gen.py\", line 326, in wrapper\n      yielded = next(result)\n    File \"/usr/local/lib/python3.7/dist-packages/ipykernel/ipkernel.py\", line 306, in do_execute\n      res = shell.run_cell(code, store_history=store_history, silent=silent)\n    File \"/usr/local/lib/python3.7/dist-packages/ipykernel/zmqshell.py\", line 536, in run_cell\n      return super(ZMQInteractiveShell, self).run_cell(*args, **kwargs)\n    File \"/usr/local/lib/python3.7/dist-packages/IPython/core/interactiveshell.py\", line 2855, in run_cell\n      raw_cell, store_history, silent, shell_futures)\n    File \"/usr/local/lib/python3.7/dist-packages/IPython/core/interactiveshell.py\", line 2881, in _run_cell\n      return runner(coro)\n    File \"/usr/local/lib/python3.7/dist-packages/IPython/core/async_helpers.py\", line 68, in _pseudo_sync_runner\n      coro.send(None)\n    File \"/usr/local/lib/python3.7/dist-packages/IPython/core/interactiveshell.py\", line 3058, in run_cell_async\n      interactivity=interactivity, compiler=compiler, result=result)\n    File \"/usr/local/lib/python3.7/dist-packages/IPython/core/interactiveshell.py\", line 3249, in run_ast_nodes\n      if (await self.run_code(code, result,  async_=asy)):\n    File \"/usr/local/lib/python3.7/dist-packages/IPython/core/interactiveshell.py\", line 3326, in run_code\n      exec(code_obj, self.user_global_ns, self.user_ns)\n    File \"<ipython-input-123-23c07e59ef1a>\", line 1, in <module>\n      model3b.fit(train_data, train_labels, epochs=5, batch_size=128)\n    File \"/usr/local/lib/python3.7/dist-packages/keras/utils/traceback_utils.py\", line 64, in error_handler\n      return fn(*args, **kwargs)\n    File \"/usr/local/lib/python3.7/dist-packages/keras/engine/training.py\", line 1409, in fit\n      tmp_logs = self.train_function(iterator)\n    File \"/usr/local/lib/python3.7/dist-packages/keras/engine/training.py\", line 1051, in train_function\n      return step_function(self, iterator)\n    File \"/usr/local/lib/python3.7/dist-packages/keras/engine/training.py\", line 1040, in step_function\n      outputs = model.distribute_strategy.run(run_step, args=(data,))\n    File \"/usr/local/lib/python3.7/dist-packages/keras/engine/training.py\", line 1030, in run_step\n      outputs = model.train_step(data)\n    File \"/usr/local/lib/python3.7/dist-packages/keras/engine/training.py\", line 890, in train_step\n      loss = self.compute_loss(x, y, y_pred, sample_weight)\n    File \"/usr/local/lib/python3.7/dist-packages/keras/engine/training.py\", line 949, in compute_loss\n      y, y_pred, sample_weight, regularization_losses=self.losses)\n    File \"/usr/local/lib/python3.7/dist-packages/keras/engine/compile_utils.py\", line 201, in __call__\n      loss_value = loss_obj(y_t, y_p, sample_weight=sw)\n    File \"/usr/local/lib/python3.7/dist-packages/keras/losses.py\", line 139, in __call__\n      losses = call_fn(y_true, y_pred)\n    File \"/usr/local/lib/python3.7/dist-packages/keras/losses.py\", line 243, in call\n      return ag_fn(y_true, y_pred, **self._fn_kwargs)\n    File \"/usr/local/lib/python3.7/dist-packages/keras/losses.py\", line 1861, in sparse_categorical_crossentropy\n      y_true, y_pred, from_logits=from_logits, axis=axis)\n    File \"/usr/local/lib/python3.7/dist-packages/keras/backend.py\", line 5239, in sparse_categorical_crossentropy\n      labels=target, logits=output)\nNode: 'sparse_categorical_crossentropy/SparseSoftmaxCrossEntropyWithLogits/SparseSoftmaxCrossEntropyWithLogits'\nlogits and labels must have the same first dimension, got logits shape [128,10] and labels shape [1280]\n\t [[{{node sparse_categorical_crossentropy/SparseSoftmaxCrossEntropyWithLogits/SparseSoftmaxCrossEntropyWithLogits}}]] [Op:__inference_train_function_210707]"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "La pérdida indica el porcentaje de errores entre las etiquetas y las predicciones del modelo.\n",
        "\n",
        "La precisión indica el porcentaje de aciertos entre las predicciones y las etiquetas."
      ],
      "metadata": {
        "id": "DW-cUg6trd3K"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "test_loss, test_acc = model3b.evaluate(test_data, test_labels)"
      ],
      "metadata": {
        "id": "rsUBHHr5rd3L"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "print('test_loss:', test_loss)\n",
        "print('test_acc:', test_acc)"
      ],
      "metadata": {
        "id": "H7jGAGCYvQqD"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "acc = history.history['accuracy']\n",
        "val_acc = history.history['val_accuracy']\n",
        "loss = history.history['loss']\n",
        "val_loss = history.history['val_loss']\n",
        "\n",
        "epochs = range(1, len(acc) + 1)\n",
        "\n",
        "# \"bo\" es para \"punto azun\"\n",
        "plt.plot(epochs, loss, 'bo', label='Pérdida de Entrenamiento')\n",
        "# \"b\" es para \"linea continua azul\"\n",
        "plt.plot(epochs, val_loss, 'b', label='Pérdida de Validación')\n",
        "plt.title('Pérdida de Entrenamiento y Validación')\n",
        "plt.xlabel('Epochs')\n",
        "plt.ylabel('Loss')\n",
        "plt.legend()\n",
        "\n",
        "plt.show()"
      ],
      "metadata": {
        "id": "lvQrKDE7zDco"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Predecir valores"
      ],
      "metadata": {
        "id": "bFdZ-N49rd3L"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Imprimir la matriz de confusión, que muestra los valores predichos frente a los reales."
      ],
      "metadata": {
        "id": "DSbIO9vLrd3L"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Note, this code is taken straight from the SKLEARN website, an nice way of \n",
        "# viewing confusion matrix.\n",
        "def plot_confusion_matrix(cm, classes,\n",
        "                          normalize = False,\n",
        "                          title = 'Confusion matrix',\n",
        "                          cmap = plt.cm.Blues):\n",
        "    '''\n",
        "    This function prints and plots the confusion matrix.\n",
        "    Normalization can be applied by setting `normalize = True`.\n",
        "    '''\n",
        "    plt.imshow(cm, interpolation = 'nearest', cmap = cmap)\n",
        "    plt.title(title)\n",
        "    plt.colorbar()\n",
        "    tick_marks = np.arange(len(classes))\n",
        "    plt.xticks(tick_marks, classes, rotation = 45)\n",
        "    plt.yticks(tick_marks, classes)\n",
        "\n",
        "    if normalize:\n",
        "        cm = cm.astype('float') / cm.sum(axis = 1)[:, np.newaxis]\n",
        "\n",
        "    thresh = cm.max() / 2.\n",
        "    for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):\n",
        "        plt.text(j, i, cm[i, j],\n",
        "                 horizontalalignment = 'center',\n",
        "                 color = 'white' if cm[i, j] > thresh else 'black')\n",
        "\n",
        "    plt.tight_layout()\n",
        "    plt.ylabel('Observación')\n",
        "    plt.xlabel('Predicción')"
      ],
      "metadata": {
        "id": "Uha1SB_Frd3L"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "Y_pred = model3b.predict(test_data)\n",
        "print('Valores de predicción del modelo')\n",
        "# Convertir en predicciones de clases desde los vectores one-hot encoding \n",
        "Y_pred_classes = np.argmax(Y_pred, axis = 1) \n",
        "# Convertir en clases de observación desde los vectores one-hot encoding\n",
        "Y_true = np.argmax(test_labels, axis = 1) \n",
        "# Computar la matriz de confusión\n",
        "confusion_mtx = confusion_matrix(Y_true, Y_pred_classes) \n",
        "# Dibujar la matriz de confusión\n",
        "print('Matriz de confusión')\n",
        "plot_confusion_matrix(confusion_mtx, classes = range(10))"
      ],
      "metadata": {
        "id": "Dvz0RQ4mrd3M"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "Se grafica el elemento 6 (```test_data[5]```) para comprobar las soluciones. "
      ],
      "metadata": {
        "id": "bwAm3z1trd3M"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "digit = test_data.reshape(10000,28, 28)[5]\n",
        "plt.imshow(digit, cmap = plt.cm.binary)\n",
        "plt.show()\n",
        "print(test_labels[5])"
      ],
      "metadata": {
        "id": "uA4P_aigrd3N"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "La clase del elemento 6 es la clase 1 (pantalones)."
      ],
      "metadata": {
        "id": "lEv3tdyMrd3N"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "predictions = model3b.predict(test_data)\n",
        "print('Predicción del elemento 6: clase', np.argmax(predictions[5]))\n",
        "print(predictions[5])\n",
        "print('Suma de las predicciones del elemento 6:', np.sum(predictions[5]))"
      ],
      "metadata": {
        "id": "MHflWlgOrd3N"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "def plot_image(i, predictions_array, true_label, img):\n",
        "  predictions_array, true_label, img = predictions_array, true_label[i], img[i]\n",
        "  plt.grid(False)\n",
        "  plt.xticks([])\n",
        "  plt.yticks([])\n",
        "\n",
        "  plt.imshow(img, cmap = plt.cm.binary)    \n",
        "\n",
        "  predicted_label = np.argmax(predictions_array)\n",
        "  if predicted_label.all() == true_label.all():\n",
        "    color = 'blue'\n",
        "  else:\n",
        "    color = 'red'\n",
        "\n",
        "  plt.xlabel('{} {:2.0f}% ({})'.format(predicted_label,\n",
        "                                100*np.max(predictions_array),\n",
        "                                true_label),\n",
        "                                color = color)\n",
        "\n",
        "def plot_value_array(i, predictions_array, true_label):\n",
        "  predictions_array, true_label = predictions_array, true_label[i]\n",
        "  plt.grid(False)\n",
        "  plt.xticks(range(10))\n",
        "  plt.yticks([])\n",
        "  thisplot = plt.bar(range(10), predictions_array, color = '#00FF00')\n",
        "  plt.ylim([0, 1])\n",
        "  predicted_label = np.argmax(predictions_array)\n",
        "\n",
        "  thisplot[predicted_label].set_color('red')\n",
        "  thisplot[np.argmax(true_label)].set_color('black')"
      ],
      "metadata": {
        "id": "RbqtOApSrd3O"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "plt.figure(figsize = (6,3))\n",
        "plt.subplot(1,2,1)\n",
        "plot_image(5, predictions[5], test_labels, test_data.reshape(10000,28,28))\n",
        "plt.subplot(1,2,2)\n",
        "plot_value_array(5, predictions[5],  test_labels)\n",
        "plt.show()"
      ],
      "metadata": {
        "id": "vAaiHM5ard3O"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Caso 4"
      ],
      "metadata": {
        "id": "pG0RpOcQnL_Y"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Crear la red de neuronas"
      ],
      "metadata": {
        "id": "qzKPBAMxnL_Y"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Se crea la red neuronal con una capa de entrada con 10 neuronas y función de activación ```sigmoid```, y una capa de salida con función de activación ```softmax```."
      ],
      "metadata": {
        "id": "pAroJtSYnL_Y"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model4 = models.Sequential()\n",
        "model4.add(layers.Dense(10, activation ='sigmoid', name = 'Capa4', input_shape = (28*28,)))\n",
        "model4.add(layers.Dense(10, name = 'CapaSalida4', activation = 'softmax'))"
      ],
      "metadata": {
        "id": "4LBCts9cnL_Y"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Explicar model.summary()\n"
      ],
      "metadata": {
        "id": "bJGnvIcWnL_Y"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "El resumen del modelo indica el número de parámetros de salida de cada capa. Este número es igual al tamaño de los datos de entrada multiplicado por el número de neuronas, y sumando el número de umbrales de activación de la capa.\n",
        "\n",
        "\n",
        "En el caso de la primera capa: 7850 = 784 (tamaño de los datos) x 10 (número de neuronas) + 10 (número de umbrales de activación).\n",
        "\n",
        "En el caso de la segunda capa: 110 = 10 (tamaño de los datos) x 10 (número de neuronas) + 10 (número de umbrales de activación)."
      ],
      "metadata": {
        "id": "_YaXFlUcnL_Z"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model4.summary()"
      ],
      "metadata": {
        "id": "V3uzK2MmnL_Z",
        "outputId": "4753fab8-523a-4329-fb49-9f3d38efbfbd",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential_11\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " Capa4 (Dense)               (None, 10)                7850      \n",
            "                                                                 \n",
            " CapaSalida4 (Dense)         (None, 10)                110       \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 7,960\n",
            "Trainable params: 7,960\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Se definen las funciones del optimizador, la pérdida y la métrica."
      ],
      "metadata": {
        "id": "OnCHy0j4nL_Z"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model4.compile(optimizer = 'rmsprop',\n",
        "              loss = 'categorical_crossentropy',\n",
        "              metrics = ['accuracy'])"
      ],
      "metadata": {
        "id": "8eC24vkgnL_a"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Entrenar la red"
      ],
      "metadata": {
        "id": "V2ncKrWKnL_a"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Para entrenar la red se realizan ```5``` ciclos con lotes de ```128``` imágenes. Se compara con los datos de validación para controlar el sobreentrenamiento. "
      ],
      "metadata": {
        "id": "JhSrEon5nL_a"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "history = model4.fit(train_val_data, train_val_labels,\n",
        "                    epochs = 5,\n",
        "                    batch_size = 128,\n",
        "                    validation_data = (val_data, val_labels))"
      ],
      "metadata": {
        "id": "h9d66JO7nL_a",
        "outputId": "9343bb10-e99a-49c7-c872-858477d51a3d",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/5\n",
            "391/391 [==============================] - 2s 3ms/step - loss: 1.5296 - accuracy: 0.6506 - val_loss: 1.1364 - val_accuracy: 0.7660\n",
            "Epoch 2/5\n",
            "391/391 [==============================] - 1s 3ms/step - loss: 0.9143 - accuracy: 0.7847 - val_loss: 0.7672 - val_accuracy: 0.7916\n",
            "Epoch 3/5\n",
            "391/391 [==============================] - 1s 3ms/step - loss: 0.6740 - accuracy: 0.8089 - val_loss: 0.6201 - val_accuracy: 0.8086\n",
            "Epoch 4/5\n",
            "391/391 [==============================] - 1s 3ms/step - loss: 0.5695 - accuracy: 0.8235 - val_loss: 0.5505 - val_accuracy: 0.8206\n",
            "Epoch 5/5\n",
            "391/391 [==============================] - 1s 3ms/step - loss: 0.5160 - accuracy: 0.8321 - val_loss: 0.5105 - val_accuracy: 0.8281\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "La pérdida indica el porcentaje de errores entre las etiquetas y las predicciones del modelo.\n",
        "\n",
        "La precisión indica el porcentaje de aciertos entre las predicciones y las etiquetas."
      ],
      "metadata": {
        "id": "QdKSZ3jonL_a"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "test_loss, test_acc = model4.evaluate(test_data, test_labels)"
      ],
      "metadata": {
        "id": "FaoTK6dbnL_b",
        "outputId": "066a929c-d8fe-4a52-edd4-70f8da781a42",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 1s 2ms/step - loss: 0.5245 - accuracy: 0.8212\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "acc = history.history['accuracy']\n",
        "val_acc = history.history['val_accuracy']\n",
        "loss = history.history['loss']\n",
        "val_loss = history.history['val_loss']\n",
        "\n",
        "epochs = range(1, len(acc) + 1)\n",
        "\n",
        "# \"bo\" es para \"punto azun\"\n",
        "plt.plot(epochs, loss, 'bo', label='Pérdida de Entrenamiento')\n",
        "# \"b\" es para \"linea continua azul\"\n",
        "plt.plot(epochs, val_loss, 'b', label='Pérdida de Validación')\n",
        "plt.title('Pérdida de Entrenamiento y Validación')\n",
        "plt.xlabel('Epochs')\n",
        "plt.ylabel('Loss')\n",
        "plt.legend()\n",
        "\n",
        "plt.show()"
      ],
      "metadata": {
        "id": "3upJJNL3zGfw",
        "outputId": "b7528312-1c2c-4066-8ef5-2ccd434c633e",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Predecir valores"
      ],
      "metadata": {
        "id": "sp6gTseInL_b"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Imprimir la matriz de confusión, que muestra los valores predichos frente a los reales."
      ],
      "metadata": {
        "id": "0Ngih3f_nL_b"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Note, this code is taken straight from the SKLEARN website, an nice way of \n",
        "# viewing confusion matrix.\n",
        "def plot_confusion_matrix(cm, classes,\n",
        "                          normalize = False,\n",
        "                          title = 'Confusion matrix',\n",
        "                          cmap = plt.cm.Blues):\n",
        "    '''\n",
        "    This function prints and plots the confusion matrix.\n",
        "    Normalization can be applied by setting `normalize = True`.\n",
        "    '''\n",
        "    plt.imshow(cm, interpolation = 'nearest', cmap = cmap)\n",
        "    plt.title(title)\n",
        "    plt.colorbar()\n",
        "    tick_marks = np.arange(len(classes))\n",
        "    plt.xticks(tick_marks, classes, rotation = 45)\n",
        "    plt.yticks(tick_marks, classes)\n",
        "\n",
        "    if normalize:\n",
        "        cm = cm.astype('float') / cm.sum(axis = 1)[:, np.newaxis]\n",
        "\n",
        "    thresh = cm.max() / 2.\n",
        "    for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):\n",
        "        plt.text(j, i, cm[i, j],\n",
        "                 horizontalalignment = 'center',\n",
        "                 color = 'white' if cm[i, j] > thresh else 'black')\n",
        "\n",
        "    plt.tight_layout()\n",
        "    plt.ylabel('Observación')\n",
        "    plt.xlabel('Predicción')"
      ],
      "metadata": {
        "id": "F8HLRrwtnL_b"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "Y_pred = model4.predict(test_data)\n",
        "print('Valores de predicción del modelo')\n",
        "# Convertir en predicciones de clases desde los vectores one-hot encoding \n",
        "Y_pred_classes = np.argmax(Y_pred, axis = 1) \n",
        "# Convertir en clases de observación desde los vectores one-hot encoding\n",
        "Y_true = np.argmax(test_labels, axis = 1) \n",
        "# Computar la matriz de confusión\n",
        "confusion_mtx = confusion_matrix(Y_true, Y_pred_classes) \n",
        "# Dibujar la matriz de confusión\n",
        "print('Matriz de confusión')\n",
        "plot_confusion_matrix(confusion_mtx, classes = range(10))"
      ],
      "metadata": {
        "id": "EGygLIdsnL_b",
        "outputId": "0c3c1d92-ca87-4d43-9565-936895e7aca8",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 366
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 0s 1ms/step\n",
            "Valores de predicción del modelo\n",
            "Matriz de confusión\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ],
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAUsAAAEmCAYAAADr3bIaAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nOyddXgUVxeH35sEgjskuCcBAo0HCC7F3SV4oS2lpQpVpBSo8FHcaZFStLgVdwIJ7k5J0ECAECTC/f6YTQg0MtndIQm9L8887M7cOfes5OzV8xNSShQKhUKRNDap7YBCoVCkB1SwVCgUCh2oYKlQKBQ6UMFSoVAodKCCpUKhUOhABUuFQqHQgQqW/yGEEJmFEKuFEA+EEEsssNNFCPG3NX1LLYQQ1YUQZ1PbD0XaR6h1lmkPIURn4BPABQgHjgA/SCl3W2jXHxgAVJVSRlvsaBpHCCGBslLKC6ntiyL9o1qWaQwhxCfAr8BIwAEoBkwGWljBfHHg3H8hUOpBCGGX2j4o0hFSSnWkkQPICTwC2iVRxh4tmF43Hb8C9qZrtYBg4FPgNnAD6Gm6NgyIBKJMdfQGhgLz49kuAUjAzvS8B3AJrXV7GegS7/zuePdVBQ4CD0z/V413bTvwPbDHZOdvIF8iry3W/y/i+d8SaAycA+4BX8Ur7wPsA+6byk4EMpqu7TS9lgjT6+0Qz/4g4CYwL/ac6Z7Spjo8TM8LAXeAWqn93VBH6h+qZZm2qAJkApYnUeZroDLgBryFFjC+iXfdES3oFkYLiJOEELmllEPQWquLpJTZpJSzknJECJEVGA80klJmRwuIRxIolwdYayqbF/gfsFYIkTdesc5AT6AAkBH4LImqHdHeg8LAd8AMoCvgCVQHvhVClDSVjQE+BvKhvXd1gfcBpJQ1TGXeMr3eRfHs50FrZfeNX7GU8iJaIJ0vhMgC/AbMkVJuT8JfxX8EFSzTFnmBUJl0N7kLMFxKeVtKeQetxegf73qU6XqUlHIdWqvK2Ux/ngOuQojMUsobUsqTCZRpApyXUs6TUkZLKf8EzgDN4pX5TUp5Tkr5BFiMFugTIwptfDYKWIgWCMdJKcNN9Z9C+5FAShkkpdxvqvcKMA2oqeM1DZFSPjP58xJSyhnABSAAKIj246RQqGCZxrgL5EtmLK0QcDXe86umc3E2Xgm2j4FsKXVEShmB1nV9F7ghhFgrhHDR4U+sT4XjPb+ZAn/uSiljTI9jg9mteNefxN4vhHASQqwRQtwUQjxEaznnS8I2wB0p5dNkyswAXIEJUspnyZRV/EdQwTJtsQ94hjZOlxjX0bqQsRQznTOHCCBLvOeO8S9KKTdKKeujtbDOoAWR5PyJ9SnETJ9SwhQ0v8pKKXMAXwEimXuSXP4hhMiGNg48CxhqGmZQKFSwTEtIKR+gjdNNEkK0FEJkEUJkEEI0EkL8ZCr2J/CNECK/ECKfqfx8M6s8AtQQQhQTQuQEvoy9IIRwEEK0MI1dPkPrzj9PwMY6wEkI0VkIYSeE6ACUB9aY6VNKyA48BB6ZWr3vvXL9FlAqhTbHAYFSyj5oY7FTLfZS8UaggmUaQ0o5Bm2N5TdoM7HXgA+AFaYiI4BA4BhwHDhkOmdOXZuARSZbQbwc4GxMflxHmyGuyb+DEVLKu0BTtBn4u2gz2U2llKHm+JRCPkObPApHa/UueuX6UGCOEOK+EKJ9csaEEC2Ahrx4nZ8AHkKILlbzWJFuUYvSFQqFQgeqZalQKBQ6UMFSoVAodKCCpUKhUOhABUuFQqHQQZpKJGCTKbu0yZbfENuViqff5XLRMcZNwtnZJrcs0XyMswzRzw18T2yM9DyZhZ4WYpTnV69eITQ01KrmbXMUlzL6X5uoEkU+ubNRStnQmj6khLQVLLPlJ1ezkYbY3j2tgyF2AWIM/MMFuP84yjDb+bJnNMy2EMYFnbvhxm2syZvd3jDbANExCS1XtQ52tsZ0Fv18vaxuU0Y/wd452RVdcTw9Mim53VmGkqaCpUKh+C8hQKSfkUAVLBUKReogAAN7H9ZGBUuFQpF6pKOWZZr0tIxjdrYNaxB3XJ7chn71nfiihSvH/9c87ny9SgUBKJo3K9emtY07/0s3/eMr7/btRfEiDni5V4w7d+/ePZo2eptK5Z1o2uhtwsLCzHodwdeu0aRBXbzdXfHxqMjkieMBGDliGM6liuLn64GfrwcbN6wzy/6saROp5+dB3aruzJw6AYD3e3elYU0fGtb0oaqbEw1r+phlOz7nzp7F18s97nDIm5OJ43+12C5Avz69KFaoAJ5urlaxBzBj8njqVHGnblUP+vfx5+nTp+zesZWGtSrzdg0fWjWqzeVLFy2u5++NG6hUwZkKLmX4+afRFtl6r29vShZ1xMejUty57l07UtXHg6o+HlRwKkVVHw9LXQas67dlCLCx1X+kMmlqu6NdvlLy1QkeGyE4PrY5Db7fROdqpYh4FsWkDS/rSxXNm5UFA6tT/dsNidr+J5EJnt27dpI1Wzbe6dWdwMPHAfj6yy/InScPn30+mF9+Hs39sDBGjPwxUduJTfDcvHGDmzdv4ObuQXh4ODWqevPn4r/4a9kSsmXNxocff5qozfgkNMFz9vRJ+vfxZ/Wm3WTImBH/ds0YNWYiJUqVjivz/beDyJ4jBwM/TzwlY0oneGJiYihdogg7d++nWPFXkw29jJ4Jnt27dpI1azb69OpG0JETuv1IbILnxvUQWjeuw9Z9R8icOTPv9uxCnfoNmDD2J2bPX0pZZxfmzJrGkUMHGTtpZoI29EzwxMTEULG8E2vXb6JwkSJUq+zNnPl/Uq58+WTvTWiCZ/eunWTLlo2+vXtw4NCxf13/ctBn5MyRk8Fff5uk7eQmeMz128/Xi6CgQKv2mW2yOkp7V//kC5p4euCXICml9WeadJImW5bxqVHegSu3HxF897Eh9qtVr0Ge3C8vK1q7ehVdunYHoEvX7qxZtdIs244FC+LmrrUGsmfPjrOLC9evWydz2flzZ3D39CZzlizY2dlR2a8669esiLsupWTNiqW0aG3dVQDbtm6hVKnSyQZKvVSrXoM8eay7rCs6OpqnT58QHR3NkyePcXAsiBCC8PCHAIQ/fICDY0GL6jh44AClS5ehZKlSZMyYkXYdOrJmtXnfE9Deh9y5E34fpJQsX7qEth06mm0/Fmv7bRECrRuu90hlUt+DZGjlW4y/Av6Je967rhM7hjdkXC8fcmbJEHe+WP5sbB3agFWD6lC5rGVrNW/fvkXBgtofk6OjI7dv30rmjuS5evUKx44cwcvbF4DpUydRxduN9/v1Nqub7+xSgQP79xB27y5PHj9m26aN3AgJjrt+YN9u8uV3oGTpMhb7Hp8lixfSzgp/tEZRsFBh+n3wMb6VyuJRrgTZc+SgZp36/DxuCt06tMSrQmmWLVpA/48+t6ie69dDKFKkaNzzwoWLEBJiTArPPbt3UcDBgTJlylps63X6nTxCm+DRe6QyhgZLIURDIcRZIcQFIcTglN6fwdaGhm6FWXVQC5a/bTuP1xdrqDVkA7fuP2F4R3cAbj14gtunq6gzdCPfLjzMtHerkC2TdeauhBAWrxd89OgR/p3aMfrn/5EjRw76vPMuR0+dZ0/AIRwdC/L14KQkaRKmrLML7334KV3aNsW/fTPKu1bCxvbFuM7KZYtp0Ub/GjY9REZGsm7Nalq3aWdVu9bk/v0w/l6/mn2HzxB06jJPHj9m2eIFzJgygbmLVhB48iLtO3dj2DdfpLarulm6eCFt26fdHyiLUC1LEELYApOARmjJYDsJIZIf0IlHvUoFOXY1jDsPtfGpOw+f8VxKpIR5Oy7hUVLrtkRGPycsIhKAo1fDuHL7EWUcs5vte4ECDty4cQOAGzdukD9/AbNtRUVF0bVTW9p36Ezzlq01+w4O2NraYmNjQ/defQgKPGiW7Y5de7Ju6z6WrtlCzly5KFVaa3lER0ezYe1KmrVsa7bfCbFxw3rc3D1wcHCwql1rsnv7VooWK0HefPnJkCEDjZq2IDBgH6dPHMPDS5vsat66LUEH9ltUT6FChQkOvhb3PCQkmMKFCydxh3lER0ezauVy2rS1zg/f6/JbN6plCWiqgxeklJeklJFo4lMp0r5u7VuMvwJeyLs45MwU97iJZ2HOhDwAtAF5G9ObWTx/Vko5ZOPKnQizHW/ctBl/zJ8DwB/z59CkWXOz7Egp6f9uH5ydy/HBRx/Hnb9pCsQAq1euoFz5CmbZD71zG4CQ4H/YsGYlLdpq45O7d2yldFknChYuYpbdxFiyKG13wQEKFSnK4cADPHn8GCklu3duo6xzOR4+fMilC+cB2LltC2WcEpIT0o+XtzcXLpznyuXLREZGsmTRQpo0Ne97khTbtm7GycmFwkWs81m+Lr/1IdJVy9LIdZaF0bJ8xxIM+L5aSAjRF5MkqU3WF7uZsmS0pWYFRz6ZExh3bkh7N1yL5UJKuBYawadztBZZFaf8DG5VkaiY50gp+WxOIPdNLc3k6O7fmV07t3M3NJSypYryzbdD+fTzwfh37sDc32ZTtFhx5i14NQG3Pvbv3cPCBfOp4FoRP19toue7YSNYunghx48dRQhBseLFGTfBPOWCfj06EnbvHhkyZOD7n34lZ85cAKz6azHNrTyxExERwdYtm5gw2boqC926dmLXju2EhoZSukQRvv1uGD169TbbnoeXD42bt6Jh7crY2dpRodJbdOnem4KFCvNO947Y2NiQM1cuxkyYZpHfdnZ2jB03kWZNGhATE0P3Hr0oX8G8Hz2Anv6d2bVrB3dDQ3EuXYyvvhlC9569Wbp4Ee06WO+ztLbfFpHOFqUbtnRICNEWaGjSMkEI4Q/4Sik/SOyehJYOWYvElg5ZA7U3PGHU3vCESa97w62+dCh7IWnv3jf5giae7hqWqkuHjGxZhgBF4z0vwutR/FMoFOkCAbapv9hcL0YOBBwEygohSgohMgIdgVUG1qdQKNIT6WydpWEtSylltBDiA2AjYAvMllKeNKo+hUKRDklHY5aGJtKQUq5D05VWKBSKV1Ap2hQKhUIfqmWpUCgUOlAtS4VCoUiGNLIzRy8qWCoUitRDtSwVCoVCB6plaR5vFc/DnhnG7D3O7Z3oxiGLCTs40TDbgNUyKCVElIEyuxntjPtDyJPNuJ1HRmPULpv0h5oNVygUiuQRpAm5CL2oYKlQKFIJ1bJUKBQKfagxS4VCodBBOmpZph9PTVhDOrV/p1oELvmKoKVf80HnWgCMHNiSI399w4FFX7JozDvkzJb5pXuKOubmzp4xDPSvm2p+J0RMTAw1q3jRsY2WwHXG1El4VnQmT1Y77oaGmmUz+No1mjaoi4+7K74eFZlikvDt0bUj1Xw9qObrQUXnUlTzTfvSrC5lS+LtXglfL3f8Kntb1baRfhv1fYG0JIWLypRuJP7de7ByTeKSt8lRvnRBerauSnX/n/HpMIpGNVwpVTQfW/afwbPdSHw6jOL81dt83uvtl+778dPW/L3H/DwglvqdGFMnjcfJ+UXWb9/KVVm+ZiNFi5mvvmhnZ8eI0T9z4PAJNu/Yy4xpkzlz+hS/z1/I7oBD7A44RPOWrWnWopXF/sfExDDww/6sXL2ew8dOsWThn5w+dcpiu/FZv2krAYGH2bPfPPmOhDDab6O+L6/j/daNsG6mdCHEx0KIk0KIE0KIP4UQmUxZzwJMOmCLTBnQEELYm55fMF0vkZz9dBcsLZVOdSnpyMETV3jyNIqYmOfsCrpAyzpubNl/hhhTUtYDxy9T2CFX3D3NalXiSshdTl28mWp+J0RISDCbNqzDv0evuHOV3NwpVryERXaTk/CVUrJ82RKriGilKWnWFGC030Z8XyANvt9WalkKIQoDHwJeUkpXtExnHYEfgbFSyjJAGBCbhr83EGY6P9ZULknSXbC0lJMXr+PnXoY8ObOSOVMGGlarQBHH3C+V6daiChv3aL+2WTNn5NOe9flhWtpLnvTVF58w9IfR2NgY9zG+KuELsHfPLvI7OFA6HUizCiFo1rgBVX29mDVzutXspi1JWf2kNb9j1VP1HDqwAzILIeyALMANoA6w1HR9DtDS9LiF6Tmm63VFMpUYNsEjhJgNNAVumyJ9muDs5VuM+X0Tqyf35/HTSI6eDY5rUQJ80bsBMTHPWbhO67J9824TJszfSsQTfZo+r4uN69eQP38B3Nw92b1zuyF1xEr4jjJJ+MaydPFC2rZL28JlsWzetovChQtz+/ZtmjV6G2dnF6pVr5HabimIleBJ0VhkPiFEYLzn06WU0wGklCFCiF+Af4AnwN9AEHBfShltKh+Mpg0G8TTCTLl3HwB5gUQH+o2cDf8dmAjMNbAOs5izYh9zVuwDYNgHzQi5dR+Ars18aVzDlUb9xseV9XYtTqt6bvwwsCU5s2fm+XPJ08gopi7amSq+xxKwby/r165m08b1PHv6lPDwh/Tr1Y1ps63zdkdFReH/ioQvaNKsq1cuZ8ce64z/GS3NGmurQIECNGvRksCDB6wSLNOcpKxO0pTfQiBsUhQsQxPT4BFC5EZrLZYE7gNLgIYW+xgPw/pvUsqdwD2j7FtC/tzZAG2Gu0Wdt1i0PpD6VcvxSY96tB04jSdPXwiE1ev9Ky5NhuDSZAgT/9jOz7P+TvVACfDd8JGcPH+Vo6cvMnPOH1SvWdtqgVJKyQcJSPgCbE9H0qwRERGEh4fHPd6yeRPlK1ink5O2JGX1k9b8tmI3vB5wWUp5R0oZBfwF+AG5TN1yeFkHLE4jzHQ9J3A3qQpSfcxSCNFXCBEohAi8E3on2fLdunaiVvUqnDt7ltIlivD77FkprvPPX/pwaNnXLB3Xj4GjF/Pg0RPGDmpP9iyZWDPlA/YvHMz4r63bzbSG33qYNnkCFcoW53pIMNV93fnwff3qebHESvju3LEtbqnQ3xu0MdtlSxbRpr0x0qxuFcvRpl17q0mz3r51i3q1quPr6UaNqr40bNSYtxtYp7FhpN9g3PfFaL9TihWD5T9AZSFEFtPYY13gFLANaGsq0x2Inc1aZXqO6fpWmYzUrWFSuACm6fg1escsPT295J6AwOQLmkF6TqTxJDLGMNu2KesGpYiMdsb9Fhv8vTXMdnrFCClc2zwlZbYGw3WXf7iwW5JSuEKIYUAHIBo4DPRBG5tcCOQxnesqpXwmhMgEzAPc0XrAHaWUl5KqX+3gUSgUqYMwHVZCSjkEGPLK6UuATwJlnwLtUmJfBUuFQpEqCHQvCUoTGNZPEkL8CewDnIUQwUKI3sndo1Ao/ltYeZ2loRipG97JKNsKheLNIC0EQb2obrhCoUg1VLBUKBSK5LDyBI/RqGCpUChSBYEwNK+BtVHBUqFQpBqqG65QKBR6SD+xMm0FSwlEx8sAZE2M3GVTd6yxe8X//qi6YbZ3nTcvm7oeajnnN8x2WERU8oXMxGiZ3Zjnxu0+MnJHltURqmWpUCgUulDBUqFQKHSggqVCoVAkQ3rb7qiCpUKhSD3ST6xM/XyWenivb29KFnXEx6PSS+enTp6IR6XyeLtX5JuvBlmlLmvIhGazt2VE83Is6OXFH728qFAoO9kz2fFru4os7OPNr+0qkt1e+51yL5qTjR9W5ffuHvze3YOeVYrpquO9vr0oUcQBb/eKcef+WrYELzdXsmey5VBQylLd3bkRwqCerejbvBr9WlRnxTxNr2bUp+/Qv01t+repTfe3PenfpjYAh/ZuZ0D7erzXqiYD2tfjSMCuFNUXixGSr9Mnj6N2FTfqVHHn/d7+PH36lN+mT8bPoxyFc9tz7651JrWsKSmb0OcZy/ixY8hmb0OomdLGr5JmpHBF+tobni6CZRf/7ixf9bJg2M7t21i7ehX7Dh7m4OHjfDTwU4vrsZZM6MA6ZQi4HEbn2YF0/z2Iq3cf4+9blMCr9+k48yCBV+/T1feFaNTR4Af0mHOIHnMO8du+f3TV0cW/BytWr3/pXPnyrixYtAw/M2QTbO3seOfzYUxftZuxC9azZuFsrl48y5djZjBp2TYmLdtGtfpNqFqvCQA5cudl6MT5TFm+g09/mMAvX/ZPcZ1gfcnXG9dDmD1tEuu27mPrvsPEPI9h5V+L8a5clYUr1lOkqPkSwfGxtqRsQp8naBruWzZvomgxfT+iyZGmpHBRwdLqVKteg9y5X5YFnTljKp989gX29vYA5C9QwOJ6rCETmjWjLW8Vycnq45psbvRzyaNnMVQvk5f1J28BsP7kLWqUzWuRrwm9Jy7lyuHk7GyWvTz5HShTXmu5Z8majaKlnLh760bcdSklOzesolZjTY+nTLmK5C3gCEDxMi48e/qUyMhnZr0Oa0u+RkfH8PTpE6Kjo3ny+DGOjgVxreRG0WIlrFaHtSVlE/o8AQZ9/gkjRv1otWCR1qRwhY3QfaQ26SJYJsSF8+fZu2c3tatXoWG92gQFWi6gZQ2Z0EK5MnH/SSRfN3Lit24eDG5QlkwZbMidJSN3IzSFyLsRkeTO8mItn2uhHPze3YNf2rhSMm8Wi1+HpdwK+YeLp4/jXMkz7tyJoP3kzpufwsVL/av87k1rKFO+Ihkz2r9ONxOkYKHCvDtgID4Vy+DuUpwcOXJSs059q9fzOiRl16xaSaFChahY6S2r2XzDpXANxch8lkWFENuEEKeEECeFEB9Z0350dDRhYffYunMvI0b9SPcuHQ2VGtCLrRA4OWRn+ZEb9Jx7iCdRz/H3KfqvchLN17O3HtFmWgA95hxi2aEQRrVKPT0UgCePHzHi4170G/Q9WbNljzu/fd1f1Gzc6l/lr144w+z/DWfAd7+8TjcT5f79MDauW8P+I2c5dPoKjx9HsGzRgtR2K8U8fvyYX34axTdD9MsupDdSEijf6GCJpoPxqZSyPFAZ6C+EKG8t44ULF6Z5i1YIIfDy9sHGxvIBcGvIhN5+9Iw74c84dUNTFdx+9g5ODtkIexxJ3qxaazJv1ozcf6ztQHkcGcOTKG3X0r7LYdjZCHJmTp1FCtFRUYwY2IvaTdrgV79p3PmY6Gj2bl5LjYYtXyp/5+Z1vv+oB5+NnEihYiVft7sJsmv7VooVL0HefPnJkCEDjZq1JPDAPqvXY7Sk7KVLF7ly5TJVvN0o71SSkOBgqlX25NbNmxbZTVNSuKiWJQBSyhtSykOmx+HAaV4InFtM0+Yt2LljOwDnz58jMjKSfPnyWWTTGjKh9yKiuB3+jGK5MwPgWTw3V+4+ZveFuzSq4ABAowoO7LqgqW7myZoh7t5yjtkRAh48if63YYORUvLrdwMpWsqJ1t3fe+na4f07KVKqLPkdC8Wde/TwAUPe70zPgd9QwcP3dbubKIWLFOVQYABPHj9GSsnuHdso6+xi9XqMlpR1da3IleBbnDp3mVPnLlO4SBF27w/CwdHRIrtvsBSu4byWMUuTyqM7EJDAtTgp3NA7CUvh9vTvTN1afpw/dxbn0sWY89ss/Lv34srlS/h4VKKnf2emzfzN4jfUWjKhY7dcYEhTF+b08KBsgazM3X+NeQHX8C6Ri4V9vPEqkYt5Adqve22n/Mzv6cnv3T34uG5phqw+o6uOHv6dqVOzKufPncWpVFHm/DaLVSuX41SqKAf276NNy6a0aKJf9vXk4QC2rF7C0YBdcUuFDuzcDMCO9cup1ejlLvjqP2dx/doVFkwdE1f+/t3kpYxfxdqSrx5ePjRp3poGtXypW9WD58+f06V7H2ZNm4hnhVLcuB5MvWpefPbhuxbVY21J2YQ+TyNIa1K4cTkt9RypjKFSuABCiGzADuAHKeVfSZX18PSSO/ceMMQPO1vjfhdUIo2EMTKRxr1HkYbZVok0/o0RUrj2DmVl4S7jdJe/PLZJklK4RmPo4JgQIgOwDPgjuUCpUCj+Y6isQxpCexdmAaellP8zqh6FQpE+EUA6ipWGjln6Af5AHSHEEdPR2MD6FApFukJgY6P/SG2MlMLdTZoYllUoFGkV1Q1XKBSK5BDpqxuugqVCoUgVBKSJ7rVeVLBUKBSphmpZKhQKhQ7UmKVCoVAkhxqzVCgUiuTR1lmmn2iZpoKllEZuBTNGjxxgw4BqhtkGyNdC/5awlHJ92QDDbBtJloy2qe2C2aSjOQ2DSRsJMvSSpoKlQqH4b5GOYqUKlgqFIpUQaumQQqFQJEt6G7NMFxo8T58+pU71yvj5elDZsxIjvx8KwJUrl6lbowrurs709O9EZGTK03YFX7tG47fr4uXmird7RSZPHA/A8mVL8HavSI7MdimWlY1vu0mDuni7u+Lj8cL2yBHDcC5VFD9fD/x8Pdi4YV0yll4woKU7QVP9CZzSlTmDGmGfwZYpA+sRMKkLByZ3YcHXTciaSUsoXKxAdtaNas2ByV3Y+GNbCufLliL/Y2JiqFnFi45ttOSwM6ZOwrOiM3my2nHXSrKs1pbCffr0KXVrVKaarwdVvCoxasRQAHZs20LNqt5Ur+xJw3o1uHTxgsV1GSkp61K2JN7ulfD1csevsrdVbacZKVy0brjeI7VJF8HS3t6eVes3syfgELv2B7Fl00YOHtjP0G++5P0BAzl84iy5cuVm3u+zU2zbzs6OkT/+TOCRE2zduZfpUydz5vQpylVw5Y9FS/GrlnJZ2fi2fxj9MwcPn2DLjr3MmKbZBug/YCB7Ag6xJ+AQDRrqyy9SKG9W3m/hht+HC/B6bz62NoJ2NZ35YvpOfPv/gc/7f3DtdjjvNdMErkb1qc4fW07j8/4fjFywn+E9/FLk/9RJ43GKl2Xct3JVlq/ZSNFi1pGTBetL4drb27Ny3WZ2Bxxi574X35VPB37A9Nlz2bU/iLbtO/HLjyMtqud1SMqu37SVgMDD7NlvuRhfLEoK13zSRbAUQpAtm9YqioqKIioqGoFg545ttGjVBoBOXf1Zuyblkp6OBQvi5u4BQPbs2XF2ceF6SAguLuVwcjJPVjZJ29ctU9Kzs7Uhc0Y7bG0Eme3tuHHvEeGPX7SoM9nbEbuewKVYXnYc0TKy7zgaTNMq/1ZmTIyQkGA2bViHf49ececqublTrHgJi/x/FWtL4Sb4XTH9sYWHPwTg4YMHOBYsaFE9aU1SVi9pzW/VsjSAmJgYqvl6UrZ4QTwE5usAACAASURBVGrXrUvJUqXJmTMXdnbasGuhwkW4cf26RXVcvXKFY0eO4OVjfU2Zq1dNtr0129OnTqKKtxvv9+tNWFiYLhvX70bw67Igzs3tzeUF7/DwcSRbDv0DwLSP63NlwTs4F8nN5FVHADh+6Q4t/MoA0KJqaXJksSdP9ky66vrqi08Y+sNobGzSzVckjpiYGKpX9sSpREFq1amLl7cv4yZNo33rZlQoW5zFC/9g4KeDLKrDaElZIQTNGjegqq8Xs2ZOt5rdNCWFK1TLEgAhRCYhxAEhxFGTFO4wS+zZ2tqyOyCIk+evEhR4kHPn9GnV6OXRo0d07dSO0b/8jxw5cljdtn+ndoz+WbPd5513OXrqPHsCDuHoWJCvB3+my06ubPY0rVyacj1/o1SXmWS1z0DH2lo3ud/YTZTqOpMz1+7RtoYTAF/O3EX1ikXYN7Ez1SsWISQ0XNc61o3r15A/fwHc3D2TLZsWsbW1Zdf+IE6eu8qhoIOcOnmCKRPHsfiv1Zw8f5XOXbvzjc73PLXYvG0X+w4EsWL1OqZPmczuXcZKl6QGscl/VcsSngF1pJRvAW5AQyFEZUuN5sqVi+o1anEwYD8PHtwnOlpTQrweEkzBQoWSuTthoqKi6NqxLe07dqZFy9aWuvhv253a0r5DZ5qbbBdwcMDW1hYbGxu69+pDUKC+Mak6bsW4cusBoQ+eEB3znBV7L1C5/Ivu5PPnkiU7ztHS1Jq8cS+CjiPWUOWDBQyZsxeABxHPkq0nYN9e1q9dzVvlStOnexd27dhGv17dUvrSU52cpu/K5r83cOL4sbhWfau27TkQYJk8rtGSsrG2ChQoQLMWLQk8aB1tqrQlhWtd3XAhRC4hxFIhxBkhxGkhRBUhRB4hxCYhxHnT/7lNZYUQYrwQ4oIQ4pgQwiM5+0ZK4Uop5SPT0wymw6ztOaF37nD//n0Anjx5wvatm3FydqF6jVqsXL4MgD/nz6Nxk5RLekop6d+vD84u5Rjw0cfmuJe07Xf74Oxcjg/i2b5540bc49UrV1CuvD51vWt3wvFxKUhme23oobZbUc5eu0epgjnjyjStXIpzwVq3Pm+OTHG/yJ938GbO3yd11fPd8JGcPH+Vo6cvMnPOH1SvWZtps+fquje1Cb1zhwfxvivbtm7GycWFhw8fcOH8OYC4748lGCkpGxERQXh4eNzjLZs3Ub6CdVYLpD0pXKu2LMcBG6SULsBbaPLbg4EtUsqywBbTc4BGQFnT0ReYkpxxowXLbIEgoAwwSUr5LylcPdy8eYP33ulFzPMY5PPntGzdloaNm+JSrjy9unVmxLDvqPSW20uTEXrZt3cPfy6YTwXXilT10X5chgwfwbNnz/j8k48IvXOHtq2aUanSW6xI4azt/r17WGiy7eer2f5u2AiWLl7I8WNHEUJQrHhxxk2YqsvewbM3Wb77PPsmdCY65jlHL95h1voTbBjVhuxZMiIEHL8cyocTtwJQo1IRhvfwQ0rYfSKEgZO3pcj/V5k2eQLjx/7C7Vs3qe7rTr0GjRg/2bLxtG5dO7Frx3ZCQ0MpXaII3343jB69eptt7+bNG7zftxcxMTE8f/6cVm3a0rBRU8ZNnEa3zu2xsbEhV+5cTJwy0yK/40vKxsTE0L1HL6tJyt6+dYuO7bReSHR0NO07duLtBvpljZPCSL9TjBUXpQshcgI1gB4AUspIIFII0QKoZSo2B9gODAJaAHOlJm+739QqLSilvEEiGC6FC1rzGFgODJBSnnjlWl+0yE7RosU8j5+9ZIgPRkmEgran3UgKtBpvmG0j94ZnNnD/9tPIGMNsZzJ437mRf3NGTYQYIYWbvaiLdBuo/0dr92fVrwLxF/hOl1JOBxBCuAHTgVNorcog4CMgREqZy1RGAGFSylxCiDXAaJP8DUKILcAgKWWii6pfy1SnlPI+sA3418+jlHK6lNJLSumVN59xOtMKhSLtkcIxy9DYWGE64ndr7AAPYIqU0h2I4EWXG9CGBjFzKBCMnQ3Pb2pRIoTIDNQHrDuFrVAo0jVWHLMMBoLjDfUtRQuet4QQBbW6REHgtul6CFA03v1FTOcSxciWZUFgmxDiGHAQ2CSlXGNgfQqFIp1hrdlwKeVN4JoQInYnSV20LvkqoLvpXHcgdgX+KqCbaVa8MvAgqfFKMFYK9xjgbpR9hUKRzrH++skBwB9CiIzAJaAnWoNwsRCiN3AVaG8quw5oDFwAHpvKJonKOqRQKFIFYeXkv1LKI4BXApfqJlBWAv1TYl93sBRCVAVKxL9HSpk+Ft8pFIo0SVrYmaMXXcFSCDEPKA0cAWLXbEhABUuFQmE2NukoWuptWXoB5eXrWJSpUCj+M6SjWKk7WJ4AHIEkZ4sUCoVCL0IYu1nE2ugNlvmAU0KIA2gJMgCQUqbeplKFQpHuSQup1/SiN1gONdKJWGwE2GcwZquZcRK7EBaRcjmLlHDzL+O2JBb0n2OY7XsLU75XXy9Gfp6K10c6ipX6gqWUcocQwgGIFQM5IKW8ndQ9CoVCkRQCbflQeiHRHTxCiGLxHrcHDgDt0BZ1Bggh2hrvnkKheJOxEfqP1CaplqWvEKKdlHIM8DXgHduaFELkBzaj7b9UKBSKlJNG5CL0kmjLUkq5BLgZW+6VbvfdpO5VKBQKPbwxshJSyj9MDzcIITYKIXoIIXoAa9H2Vr52rK0z/V7fXpQo4oC3e8W4c8OHfouv51tU8XaneeMGFgmhzZo2kXp+HtSt6s7MqRPizv82fTK1fStRt6o7Pwz9KsV2g4Ov0bRhXXw9KlLZsxJTJmk5L48fO0r9Wn5U9XajQ5sWPHz4ULfND5pWIHBsKw7+rxW/D6yFfQZbihfIxo5RzTg+oS1zP65FBjvtK+NXzoG9PzXn4aIetKxcIsX+x8faOtbuFcpQ3deNWlU9qVtDk5IY9f0QalR2p1ZVT9q2aMSNG5aJ24Fx+tvnzp7F18s97nDIm5OJ43+1mv20ohsu0Bal6z1SG12tQynl52iJNSuZjulSSsvk8czE2jrTXfx7sGL1+pfODfzkcwKCjrLv4GEaNm7CqB+Gm2X77OmT/Dl3Nqs37WbjzoNs2biOK5cusnfXdv5ev5oNOw+yZe9h+vUfmGLbdrZ2jBj1MwGHjrNp+x5mTpvCmdOn+PD9fgz5fiR7Dx6hafOWjB/7iy57hfJk4f1G5ak2aBXenyzXNMn9SjKiqzcT1pyg4oCl3I+IpEcdTQztWmgEfSftYtFuy5I1G6VjvWLtZrbvDWLLTi1j1wcffcrO/YfZvjeItxs25pfRI9Kk3wBOzs4EBB4mIPAwewMCyZwlC81btLKK7bSnG/6GtCzjI6VcJqX8xHQsN9KppLC2znS16jXInftle/HVHR8/jjB7XOX8uTO4e3qTOUsW7OzsqOxXnfVrVjDvtxm8/9Fn2NvbA5Avf4EU235Vk9zJ2YUb10O4eOEcftVqAFC7bj1Wr9T/UdnZCjJntMXWRpDF3pabYU+o6VqQ5fuuADB/+3ma+mjzfv/cecSJq2E8t3AJz+vSsc4e/zONeGzxWNnr8nvb1i2UKlWaYsWLW8Ve2tMNf0OkcIUQsSnXw4UQD+Md4UII/f27dMjQ777GuXQxFv25gG+GmNeydHapwIH9ewi7d5cnjx+zbdNGboQEc/nieQ7s30Pz+tVp16weRw8lmsleF1evXuH40SN4evviUq48a1evAmDFX0sJiafklxTX7z3m11UnODulA5dmdOTB4ygOXwrlQURk3JrGkLuPKZQnq0W+/qteA3SshRC0bdmIOtV9mDN7Rtz5H4Z9SyWXkixd/CeDvx5qUR2vS397yeKFtOvQ0Wr20pJueOwOHr1HapPcmGU10//ZpZQ54h3ZpZS6xLWFELZCiMMmzYt0w9DhP3D24j906NSZaVMmmmWjrLML7334KV3aNsW/fTPKu1bCxtaW6OhoHoSFsfLvnXw9dBTv9+5iti7Lo0eP6NapPSN/0jTJJ06dyawZU6hZ1YdH4eFkyJhRl51cWTPS1LsY5fsvoXTfhWS1t6O+WxGzfEpt1v69nW27D7LorzXMnjGFvbt3AfD1kO85duYybdt3Yub0yansZfJERkaybs1qWrdpl9quGIZIwZHa6OqGCyEqCyGyx3ueXQjhq7OOj9AkKdMlHTp2YeXyv8y+v2PXnqzbuo+la7aQM1cuSpUuS8FChWnYtAVCCNw8vRE2Nty7G5q8sVeIioqiW+d2tOvYieYttTEtJ2cXlq/ewI69B2jbviMlS5bSZat2pUJcvf2I0IdPiY6RrAy4ShWXAuTMmjHuV71w3ixcvxeRYj+Twggd64KFtPvz5y9A42YtORT0si572w6dWJOC4YmEeB362xs3rMfN3QMHBwer2UxbuuFvUDc8HlOAR/GeR6BDZ1cIUQRoAlimO/qauXD+fNzjNatXWqQxHXpHW3EVEvwPG9aspEXbDrzduDn7du8A4NKF80RFRpInb74U2ZVS8sF77+DkXI4PPnyhSX7ntlbf8+fP+fnHkfTs00+XveDQCLyd8scpMtaqWJDTwffZefIGraqUAKBrrbKsPfhPivxMDmvrWL+qub19yybKla/AxQsvPtP1a1dR1sk5MROp4ndCLFlk3S44pC3dcG02/M1YlB4fET89m5TyuRBCz72/Al8A2RMr8JIUbrFiiRWLw9o60z38O7Nr53buhobiVKooX387lI0b1nP+3FlsbGwoVqw44yYm+7uQKP16dCTs3j0yZMjA9z/9Ss6cuejQpTufD+hLPT8PMmbMyP8mzUzxL+f+fXtYtGA+5V0rUs3XE4Dvhn3PxYsXmDlN87dZi5Z07dZDl72D5++wYt8V9v7cgugYydHLd5m96SwbgoKZ+3EthnT05OiVu/y+5RwAnqXzsfCLuuTKmpHGXkX5poM7Xh+nvLVmbR3rO7dv0b2ztrksOjqGNu07Urd+A3p0ac+F8+ewsREUKVqcMeMmmV2HEX6/SkREBFu3bGLCZH2a8npJW7rhaaPFqBdduuFCiL/QxMljo8b7QG0pZcsk7mkKNJZSvi+EqAV8JqVsmlQ9np5eck+AZZMdiWFk4oV7j4xNpJEjs3HqH+k1kUbE02jDbGfNZKzaitIN18hbqoJs/P0C3eXnd3ULklImJBvxWtDbDX8XqIomFRkM+GJqDSaBH9BcCHEFWAjUEULMN9NPhULxBpKexiz1Zh26DaRo8ERK+SXwJUC8lmXXlDqoUCjeTGLHLNMLejV4MgG9gQpAptjzUkrj+lkKheKNJy20GPWitxs+D01WogGwAygChOutREq5PbnxSoVC8d9CCLAVQveR2ugNlmWklN8CEVLKOWjLgfSus1QoFIoESU97w/VO+0WZ/r8vhHBFS92W8g3NCoVCEY/01A3XGyynCyFyA98Cq4BspscKhUJhNukoVuoOlr9JKWPQxiv17Z9TKBSKJBCkjTyVetE7ZnlZCDFdCFFXpKd2s0KhSLukYLwyLUQdvS1LF6Ap0B+YLYRYDSyUUu42zDMrY2SKp/w57A2zbTRG7rJx6DbPMNu35vobZvvm/aeG2QYoYOD3JTT8mSF2o2KM2XWUntpeejOlP5ZSLpZStgbcgBxoXXKFQqEwG5sUHKmNbh+EEDWFEJOBILSF6e0N80qhULzxCN7A7Y6m/d2HgcXA51JK6yY1VCgU/0neqO2OQghbYLaU0jxtBYVCoUiAWFmJ9EKy3XDTkqE0tVXRSCnP9Gj72rVrNKhXG/dK5fF4qwITx4+zmm2w3O8yBXOwa2STuOPazA6819CFisVzs3lYQ3aNbML2EY3xKJ0XgHZ+Jdkzuil7Rzfl76ENcC2W2yy/rSGb/MWH/fAqV4wG1T3jzq1duYy3q3lQqkAWjh0Jijsfdu8unVo2oELxfHw3KOWKnfGZNGEcXu4V8XJztYoMbkKSzP/78Xu8K5SiYU0fGtb0Yesm66mm6iU9Jf/VO2a5RwgxUQhRXQjhEXsY6lkiGCnlmV5t29nZMfqnMRw+doodu/czbeqkNOX3hRsPqf7VWqp/tZaaX6/jSWQMawKvMbyTB6P/Okb1r9byw9IjDO+kfaWu3n5Ek+//purgNfy0/Djj+lQ2y3dryCa36ejP7wtfVj90LleBKb8vxKdKtZfO29tn4pPB3/HVsFEW1Xny5Al+mz2TnXsC2B94hPXr1nLxwgWz7SUmyQzQ570BbNhxgA07DlCnfkOL/DaH9LR0SG+wdEPLODQcGGM69AlSWxkjpTzTq+2CBQvi7vFCFtfFpRzXr1tHsc/aftdydeTyrXCuhUYggRyZMwCQI3NGboY9AeDA+Tvcj9ASKgdeCKVQnixm1WUN2WTfqtXI9YpUchknF0qXcfpX2SxZs+Jd2Q97+0z/upYSzp45jbePD1lMEsrVa9Rg5QrzdaASk2RObbQUbUL3kdroXTpUO4GjjtHOJYSRUp7p1XZ8rl65wpEjh/H2sU6eE2v73bpKCZbuuwLA4LkHGd7Zk5MTWjOiiyfDFh3+V3n/WmXYfDR1pFpTi/LlXdm7ezd3797l8ePHbNywXrekcUIkJskMMGfmFN6u7sVnA/py/36YtV6Cbt64pUNCCAchxCwhxHrT8/JCiGSFb4QQV4QQx4UQR4QQxuhFKOJ49OgRndq34ecxv5Ijhy6l4tdKBlsbGnsWYcX+qwD0rufMV/MCqTDgL76aF8jEvlVeKl+9vAP+tcrw3Z+HUsPdVMOlXDk++ewLmjdpQMtmjahU6S1sbG3NtpeYJLN/z77sCjrNhh0HKODgyIhvB1nxVejjTeyG/w5sBAqZnp8D9I5g15ZSullLO8NIKc/0ahs0WdxO7dvQoVMXWrZqbTW71vS7vlshjl6+x52H2g6ZTjVKscqkFrk84CoepfLGla1QNBcT3qlCpzHbCDNY4ygt0r1nb/bsD+TvLTvIlTs3Zcv+u9ufEhKSZM5fwAFbW1tsbGzo1K0XRw693vaMSEEXPN10w4F8UsrFwHMAKWU0EGOYV0lgpJRnerUtpeTdd3rj7FKOjz7+xCo2Y7Gm322rlozrggPcDHtCtXKaJnbNCo5cuqXlky6SNwvzP65J38l7uHhTd47pN4rbJknja//8w6oVy2nfsbNF9hKSZL5180bc9Y1rV+Fc7vWrPKanlqXeveERQoi8gAQQQlQGHui4TwJ/CyEkME1KOf3VAimVwjVSyjO92t67Zw8L/piHq2tFfD3dABg2YiQNGzW22La1/M5ib0dt14IMnLk/7tyHM/fxYzdvbG0Ez6Ke85Hp2qDWlciT3Z4xPX0ATZmz1jfrUlynNWSTP+zbjf17dhF2L5QqlUoz8ItvyZU7N0O//IR7d0Pp1bk15StUYu6S1QBU83DmUXg4UZGRbFq/mrlL1lDWuVyKfe/SsS337t7FLkMG/jduIrly5UqxjfgkJMn83aCPOXXiGEIIihQrzqgxEy2qwxzSwpIgveiVwvUAJgCuwAkgP9BWSnksmfsKSylDhBAFgE3AACnlzsTKGymFq3j9qEQaCWNkIo27Bg1ZNKlTlWNHgqwa2go7VZTvTtavNf9d/bLJSuGaNtEEAiFSyqZCiJJo6rJ50bZq+0spI4UQ9sBcwBO4C3SQUl5Jyrbe2fBDQE00Odx+QIXkAqXpvhDT/7eB5YCPnvoUCsV/gBQsSE9BC/Qj4HS85z8CY6WUZYAwNOFFTP+Hmc6PNZVLEr2z4e2AzFLKk0BLYFFyi9KFEFmFENljHwNvo7VKFQqFAtASAOv9l6wtIYqg6YPNND0XQB1gqanIHLT4BdDC9BzT9WRz9eqd4PlWShkuhKgG1AVmAVOSuccB2C2EOAocANZKKV//fiqFQpEmidUNT0HLMp8QIjDe0fcVk78CX2CaiEbret83TUgDBAOxSzkKA9cgbsL6gal8ouid4Imd+W4CzJBSrhVCjEjqBinlJeAtnfYVCsV/kBRO8IQmNmYphGgK3JZSBgkhalnBtX+hN1iGCCGmAfWBH02Do2lhUb1CoUjHWDFPpR/QXAjRGC3fbg5gHJBLCGFnaj0WAWK3g4UARYFgIYQdkBNtoidR9Aa89miL0htIKe8DeYDPU/hiFAqFIg4zuuGJIqX8UkpZREpZAugIbJVSdgG2AW1NxboDsYkNVpmeY7q+VSazNEi3rARwBWgkhBgAFJRS/q3nXoVCoUiQ1yNYNgj4RAhxAW1Mcpbp/Cwgr+n8J8Dg5AzpzZT+HdAOiE198psQYomUMslxS4VCoUgKI7YxSim3A9tNjy+RwJJFKeVTtJimG71jll2At0wVIIQYDRwBVLBUKBRmEdsNTy/oDZbX0QZNY7c22PNioNRqPImK4XTIQ2ubBcClUHZD7AJcDzN2x0eOzHo/ppRzNfSxYbaN3GUzM+CyYbb7+JY0zDbAk0jj0ioYJcucwdaIqCawTQubvnWS5F+hEGIC2v7uB8BJIcQm06V6aGsnFQqFwiw0dcfU9kI/yTVZYjdqnwK2oAXOaLQZJoVCoTCfNKKto5fkguUC4AegF3AV7cegGPAb8JWxrikUijedtJCnUi/JLR36CcgNlJRSekopPYBSaAs4fzbaOYVC8eYS2w1PL/kskwuWTYG+Usq4DKxSyofAe2hbHw1j6Of9qetZmnZvv1D2O3fqON1b1aN9gyp81LsDj8K1yaD9u7bSuWkN2jeoQuemNTiwd4dFdcfExFDZ24PWLZul+N7BH/XDp3xxGtV4sStr9LCveNvPjSa1fHivRwcePrj/0j3Xg69RqWR+Zk5OmeSpp2tZalZ2p7afF/Vrau/T8WNHaFSnWty5Q4EHddm6eT2Ydzs3pf3bvrRvUJk/f9O2/j+4H0Z//5a0ru1Bf/+Wcb4/fHCfz9/tQqdGVenesg4XzpqvJmmp1G7Us2f8/E5LRnVvzA9dG7B21lgAxr7fntE9mjC6RxO+blGZ6V/2A7RkyUt/HcawDrUZ1b0R186al9/FCGnjmJgYalbxomMbLcHygPfeobqvB9V83OnepT2PHj2yuA4j5Z5TypuUKV0mtKrdpCWefCJMC2jWtjMT5yx76dzwwQP4cNBQFm/cR+0GTZk7fTwAuXLnZdysRSzeuI/hY6by7cf9LKp70oRxuLikPGErQOuO/sxe+LJynl/NOqzbEcja7QcoWbosU8e/LIz5w5BB1Kj7tln1/bV2E9v2BLJph5Y4d/i3X/HZ4G/YtieQQV8NYfh3X+qyY2dnx8CvRrD47wB+W7aJpfNmcun8GeZMHYt31Zr8te0Q3lVrMmeKFoh+mzwGp3IV+XP9XoaNmcqY4cmu6U0Qa0jt2mXMyIfj/uDLOesY/PsaTu/fyeUTh/l48mIG/76Wwb+vpaSrO2/VaADAqf3buX3tCt8t3ErHz0ey6JdvU8XvhJg6aTxOzi5xz3/4cQy7Ag6x+8BhihQpysypkyyyb6Qkszm8SS3LU0KIbq+eFEJ0Bc4Y45KGp68fOXPmfuncP5cv4uHrB0DlarXZsn4VAC6ub5HfoSAApZ3K8ezpEyKfPTOr3uDgYDasX5fijNqx+FSpRq5cL0unVq9VDzs7bXjYzdObm/FkajetW0XRYiXMyqadEEIIwk0t7ocPH+DoWFDXffkKOOLiqmVZz5otOyXKOHHn5g12bFpH0zadAGjaphPbN60F4PL5s3hVqQFAidJO3Aj5h7sm6YKUYA2pXSEE9lmyAhATHU1MTPRLe46fRIRzLmgflWrUB+D4rs34NGyFEIKSru48efSQB6Ep890IaeOQkGA2bViHf49ecediheeklDx9+tTivdRGSjKnFMGbpe7YH+gvhNguhBhjOnYAH6J1xV8rpcq6sP1v7Y9187oV3Lrx76WeW9avxMX1LTLam7fe7ItPP2bEqB+xsTHm41myYG5cKzIi4hHTJv6PAZ+ZN1cmhKB9y8bUq+HL3N9mAjDix18Y9u2XuJUrxdBvBvP10JTvG7gefJWzJ49Twc2Te6G3yVfAEYC8+R24ZwoqZcu5sm2jJqVw8mgQN0Oucfvm9ZTXZSWp3ecxMYzu0YQvm3nj4uVHiQpucdeO7dyEs1dVMmfV1treD71J7gIvfkRyFXDkQejNVPE7Pl998QlDfxj9r+9e/369cSlZmPPnzvDOex9YVMfrkmTWhdC+w3qP1CbJiCClDJFS+gLD0faGXwGGSyl9YrOgJ4UQIpcQYqkQ4owQ4rQQokpy9yTFkJ8msWT+TDo3rUHEo0dkyJDhpesXz51m/OghfD0yZWN/saxbu4b8BfLj4eFpiZuJMnnsj9jZ2dGiTUcAxv/8Az37DSBr1mxm2Vu9cRtbdh3gz2WrmT1jCvv27OL3mdMZPupnjpy+xPejfmbgBykbkngc8YhB73fjk29Hki37y3K68b+03d8dSPjDB3RuUo1Fc6bhVL4SNrap9/tvY2vL4N/X8v1fe7l6+hjXL52Nuxa0eTWe9VI+/vw62bh+DfnzF8DN/d/fvUnTZnHq4jWcnMuxfOniVPDOOEQKjtRG19YQKeVWYKsZ9scBG6SUbYUQGYEsZtiIo2QZJybP08YDr166wO5tG+Ou3boRwqf9ujD8f9MoWryUWfb3793D2jWr2bhhPU+fPiX84UN6dfdn9hzLtWSWLZzH1k3rmbd0XVzAOXroIBvWLOen77/m4YMH2NjYkNHenm699TXaCxbS8pjmz1+Axk1bcCjoIIv+nMcPP/0PgOat2vLxgHd1+xgdFcWg97vRsHk76jTUJhjy5CtA6O2b5CvgSOjtm+TOmx+AbNlzMOTnyYDWRWxRoxKFi5bQXVcs1pYIzpI9B2U9KnN6/04KlXLm0f17XD19lHdGTo0rkyufI2G3Xygb3r99k5z5HFPV74B9e1m/djWbNq7n2dOnhIc/pF+vbkybPRcAW1tbWrdtz/ixv9ClWw+z6zFakjklCEhXO3gMawoIIXICNTBl+ZBSFdK/kgAAIABJREFURprSu5nNvdA7ADx//pyZE3+mTRdtbCf8wX0+7NmeAYOG4uZVOSkTSTL8h1FcuHyNM+cvM3f+n9SsXccqgXLH1r+ZPmks0+YuIXOWF78XC1dtZkfgGXYEnqFH3/6899HnugNlREQEj8LD4x5v37qZcuUq4OhYkL27NU24XTu2Uap0GV32pJR8P/gDSpR2okufF129GvUasWbZnwCsWfYnNetripHhD+8TFamJY61YNBd3n6r/aonqwRpSu+Fhd3lsGqeNfPaUMwd342D6wTyyfT2uVeuQId6wjGu1uhzYsBwpJZdPHCZTtuzkzFfgtfsdn++Gj+Tk+ascPX2RmXP+oHrN2kydNYdLFy8A2uezfu1qyjo5m12HEX5bSnqa4DFu0zGUBO6gZSh6C01Z7SMpZUT8QvGlcB0LvxhL+XJAL4L27+Z+2F0aVi7Hux9/yeOICBbPmwFAnQbNaNGuKwCL5s7g2tVLzBj3EzPG/QTA5HnLyZMvv4EvL2EG9utOwN6dhN27i59bGT76/Bumjv+FyMhn9GjfFAA3Tx++/3mCRfXcuX2LHl20pCkx0dG0bteROvUbkCVbNr4Z9AnR0dFkss/EmHHJqX9oHA3cz7rliyjjXJ7OTaoB0P+z7+j+7sd8+UEPVi2eh2Phooya+DsAly+cY9hn74EQlCrrwrc/miejag2p3Yd3bzP/h895/jwG+VziXqcxrn51AQjavIb6XV9uXVeoUptT+7YzvENtMmTKRNevfkoVv5NDSsn7fXsS/jAcKSWuFSvxyzjLZsNfh9/6SRtjkXrRJYVrlmEhvID9gJ+UMkAIMQ54KKVMdJ1G+Uru8o/Vlq2RTAyVSCNhjEyk4Vo0p2G2VSKNhMmc0dYQu36+XgQFBVo1spUu/5Yc+Yd+PfiOHkWSlcI1EiNH5IOBYCllgOn5UiBJRUiFQvHf4o2ZDbcEKeVN4JoQInaQpS5aQg6FQqEA3sDZcAsYAPxhmgm/BPQ0uD6FQpFeEFYVLDMcQ4OllPIIkGpjDAqFIu0Su4MnvWB0y1KhUCgSRbUsFQqFQgdvUvJfhUKhMAStG55+oqUKlgqFItVIR71wFSwVCkVqIRCqZWkemTPYUq5wyvcXpzaF82RObRfM5v/tnXd8FNX6h5+XhB4UkCoBBAKEUFMgdJAOFyQK0pQiCHrBXrle67WhCCo/RVFREJQiTUBpgiLgBUJCR1G8gJRQonRBks37+2MmISBlk53JZuE8fPaTyezs97xsZt89Z+ac9+vmKpszKe6tVHFzlU2xlk+7pg1w5Lusl83zWvvUWVd0U9PcWunniqwr5KpkaTAYrh3MNUuDwWDwhlxSTchbTLI0GAx+wyRLg8Fg8IJAusETSKuNMnDTytNou6t95swZWjdrSNPYKBpF1+HVF58HYOiQgdStEUaz2GiaxUazeeMGByJ3Jvb7ezQmYdL9rPv0fiY+34P8+YJpGV2ZH8YPZfUnw1g6djCVy50zqevWqhaJkx4gYdL9THjudr/Fnc6OX7bTpmn9jEe18iX4cOwYtm7eRJe2zWnVOIp+PW/lxPHjPrWTVQRrUrq3D38TcD3LdCvPrxYsoVxoKE0b1qdz51uoERFhtANAO3/+/Hy54BtCQkJISUmhY+vmtGnfAYD/vPIaXW/t5nPMTsZ+Y4kiDO3eiMg73+bM2VQm/6cnt7euzRP9WnD78M/YvvswQ25twPD+LRnyyiyqhN7AY3c2p9XQDzh64gwlixb2S9yZCatanW9WxmdoR9WoRMfOXRncvzfPvjiCRk2bM2XSBN4bM5onnn4+W21kl9zgB+4tAdezdNPK02i7ry0ihIRYBm0pKSmkpKS6NhRzKvbgoDwUzJ+XIPtnUrJVufy6wpZVxXWFC5CUbFl8DOwSw7hZazh6wioIffjoqUvquh33xVixfBkVK1UmtEJF/vfrLzRs0gyA5je35qt5sx1pIytIFv75m4BLlm5aeRrtnNH2eDw0i42mWsWytGzdmpgGsQC89PwzNGkQyVNPPMJf2fR9z4wTse9PPsFbU1fy88zH2DnnSY6f+oul8TsYOmIOs0f2Y8esx+nTvh5vTLZ8j6qWv4Gq5UuwbOxglo+7h7axVf0S96X4cuYXxHXrAUC18AgWfjUXgPlzZrJ/315H2vCWQBuGu2lYVl1ENmR6HBeRh9xqzxA4BAUFsWJNAlt/2U3iuni2bd3Csy+8zNoNW1m2YjVHjhzh7VFZ98Vxg6JFCtC5aQ1q9BhF5bjXKFwgL73a1eX+no259fFPCbttJJO+TuS1+zsCEBSUh7DyN9Du/vH0e346Y5+I4/qQAn7+X1icPXuWxQvm0yXOutQx+p1xTBw/jvYtGnLy5Eny5c2XwxFlpV/p/2zpZqX07apaT1XrAdHAn4DP/Xw3rTyNds5qX1+0KM2at2TpkkWUKVsWESF//vzc0bc/CevifdZ3IvZWMVXYlXSE5KN/kupJY87322hUuwK1w8oSv83qic1YtpmGtSpYbRw+zvyVP5HqSWN30hF+2ZNMWOgNOR73xVi2ZCG169ajZKnSAFStFs7U2V+zaPlq4rr3oGKl7FlIZ5ssODvmhkubOTUMbw38qqq7fRVy08rTaLuvnXz4MMeOWo7Ip0+f5ttl31C1WnUOJFk+3qrKV/PmUsMBx0EnYt9z8BgNaoZSMH9eAG6OrsJPuw5zXeH8hJW3kmCrmDC277Zsmuet+JHmkdZSzBuuL0TV8iXYuf+PHI/7YsyZOZ24bj0zfk8+fAiwrKXfHjmCvncN9rmNrGJsJf5OL2DKxZ7IbIVbvkKFKwq5aeVptN3XPnAgiaGDB+JJ85CWlsatt3WnQ6fO3NKxDcnJyagqtevUZfSYsbki9vhte5n97Vb++/FQUj1pbPw5ifFz49l3+BhTXupNmipHT5zhnldnAbBkzS+0qR9G4qQH8KSl8dTYhfxx/HSOx30hf546xYpvl/L6m+esdOfMmMaEj94HoGOXOHrd2d+nNrKKdc0yN6RB73DNCjejAct/Zz9QU1UPXu7Y6OgYXbVmnavxGHIONwtpFMjrjuUrmEIaF6NDy0ZsXJ/gaGarUTtSP5n9rdfHN6pazK9WuDnRs+wIJF4pURoMhmuQwOlY5kiy7M0lhuAGg+HaJpCG4a7e4BGRwkBbYJab7RgMhsAkkG7wuJosVfWUqt6gqsfcbMdgMAQoDmVLESkvIt+KyDYR2SoiD9r7i4vIEhH5xf5ZzN4vIjJGRHaIyCYRibpSqAG3gsdgMFwdWDnQsUnpqcCjqhoBNASGiUgEMBxYqqpVgaX272DdS6lqP4YA712pAZMsDQaDf3BwUrqqJqlqor19AvgRKAd0BSbah00E4uztrsCnarEaKCoiZS/XhkmWBoPBb2RxFF5CRNZlegy5qKbITUAksAYorapJ9lMHgNL2djlgT6aX7bX3XZKAK9FmMBiuIrJ25yb5SvMsRSQEmAk8pKrHJVOXVFVVRLI9sdz0LA0Gg59wtpCGiOTFSpSfqWr6DJyD6cNr++che/8+oHyml4fa+y6JSZYGg8FvOHXNUqwu5HjgR1UdnempuUD6Os7+wJeZ9vez74o3BI5lGq5flFw1DFesQgpuIC5Ofk31pLmmDe7GHuRiocB8Qe59F59Nde89/+PbF13TBigW+6Br2n+sfssVXTfOE4fnTzYB+gKbRSTdk+QpYAQwXUQGAbuBHvZzXwOdgB1YFdHuulIDuSpZGgyGawunOgKqupJL597WFzlegWFZacMkS4PB4DcCaLWjSZYGg8F/BFCuDMwbPOFVK1E/sg6xMZE0aVjfMd177h5IhRtLEV2vliN6/xwyiErly9Agqk7GvldefIFqlcvTuEEUjRtEsWjh19nUHshNoaWpH1k7Y9+smV8QU68WRQoEkZjgTKm7PXv20L7NzUTWiSCqbk3eGfO2z5r3DhlIxdDSxGSK/Y8//qBzx3bUiahG547tOHLkSLa09+7ZQ+f2rWkQWYvYqNq8984YAAbc2YumsVE0jY2idvXKNI294uq2K+LEeTisdwvWTRtOwvTh3Ne7BQC3talHwvThnIp/k6ga527Y5g0OYtxzfYif9iRrpjxBs+gwv8XtCFmZZJkLsmpAJkuABUuWsWbdelat9t1+IJ2+/Qfw5fyFjund0bc/s+f+PRkOu/8hflibyA9rE2nfoVM2tQcwZ96C8/ZFRNTi82kzadKsebY0L0ZwcDAjXh/F+k3bWL5yNePef5cft23zSfPOi8Q+auQIWrZqxaZtP9OyVStGjcyeV3ZwcDAvjRjJ2vVb+Gb5D3w4biw//biNCZOnsnJNIivXJHJL3G106XqrT/+HdHw5DyOqlOWuuEY06z+KBr1fp2OzmlQOLcHWHUn0evxjVib+et7xA29tBED9nq/ReehYRjwcl+1rfm58frKD8eAJUJo2a07x4sUd1StWzDm9K2mH16hBterVHW2nbNmyREZZvbAiRYoQHl6D/ft9cxps2qw5xS+I/at5c7nDrtR9x539mT83e9avZcqWpV7kuXirh4efF6+qMnvmF3Tv0Sub0TtHeKXSxG/ZzekzKXg8aaxI3EFcqzps33WQX3Yf+vvxlcvwXfzPABw+cpJjJ04THVH+b8cFCoLx4HEdEaFLp/Y0jo1h/Ecf+DucLPPBe+/SMKYe/xwyKNvDTX+we9cuNmxYT33butZJDh06SNmy1tLcMmXKcOiQ77Wid+/exaYNG4ipfy7eH1atoGTp0lQJy7pF7YX4eh5u3ZFEk8jKFL++EAUL5KVDkwhCSxe75PGbf95H5xa1CArKQ8UbixNZI/Syx7sVt5ME0Cjc3Rs8IvIwcDfWFMrNwF2qesZX3W++XUG5cuU4dOgQXTq2o3r1cJo6OPR0k7uH3MuTTz2NiPDi88/y1JOP8d4H4/0d1hU5efIkvXt0Y+Sot7juuutcbUtEfJ5ScvLkSfr2vp1XR44+L94Z06fS/XZnepW+nofbdx1k1MSlzHt3KH+e/ouNP+/Dk3bp+aMT564hvFIZVk16lN+SjrB6467LHu9W3I6SG7Kgl7jpG14OeACIUdVaQBCWcZnPpNuClipVii5d41gXv9YJ2RyhVOnSBAUFkSdPHgYMvNsRy1e3SUlJoXePbvTsfQdxt97mShulSpUmyXZ4TEpKomTJUtnWSklJoW/v7vTo2Ydb4s7Fm5qayrwvZ3Nb9x6XebX3OHEeTvxyNU3ufIO2g/+Po8dP88tvhy95rMeTxhOjZ9Owz0h6PPoRRYsUvOhwPSfidgpzzfIcwUBBEQkGCmEZl/nEqVOnOHHiRMb20m+WEFHTmbvXOUG65SvAvLlzHHNhdAtV5d7Bg6geXoMHH37EtXY6de7CZ5OtSlqfTZ7IP7pkz/pVVbnv3rupXr0G9z348HnPfbfsG6pVC6dcaKjP8Tp1HpYsFgJA+TLF6NqqDtMWJFzy2IIF8lKoQD4AWsVWJ9Xj4aedWbtckds+P3nE+4e/cW0Yrqr7ROQN4DfgNLBYVRdfeFxWrXAPHTxIr9ut3kJqaio9evWmXfsOjsTc787erFj+HcnJyVS5KZRnnn2BAQMHZVvvrr59WLFiOb8nJ1O9SgWeevo5Vn6/nE2bNiIiVKhYkTHvvJ8t7QF9+7Di++/4PTmZapXL8+9nnqdY8eI89vADJB8+TLe4ztSpU48vv/Lt7v4Pq1bx+WeTqFWrNrHR9QB44aVX6NAxe3fxAfpnir1q5fI8/czzPPr4cPr26cmnn3xM+QoVmfT5tGxpr/5hFVM/n0zNWrUzpgc9+8JLtOvQiZlfTKNbj55XUPAOp87DKSMHUvz6wqSkenhoxAyOnTzNLTfXYfTj3ShRLIRZb9/Dpp/3cst971OyWBHmvXMvaarsP3SMQc9M9lvcjpELkqC3uGaFa5dvnwn0BI4CXwAzVPWSf+Go6Bh1ayqDWRt+cdxcG56W5p7NcqqL2nmD3P0EF2/4kGvabq0Nb9KwPokJ6xx9Y2rXjdJZi1d5fXy1MoX8aoXr5jC8DbBTVQ+ragqWaVljF9szGAyBhIOV0nMCN5Plb0BDESlkl09qjVXq3WAwGAAzdQgAVV0jIjOARCwzofVA4E2KNBgM7pEbsqCXuDrPUlWfA55zsw2DwRCo5I4pQd5iqg4ZDAa/kRuuRXqLSZYGg8Ev5JZrkd5ikqXBYPAfAZQtTbI0GAx+I08AjcNNsjQYDH4jcFKlSZYGg8Ff5JLJ5t6Sq5KlVQw0gN49m2AXLV8DGfcWJEK+YPfe85NnUl3TBjiyxndrjktRrP59ruj+tf03V3QDqW+Zq5KlwWC4dkivlB4omGRpMBj8RgDlSpMsDQaD/wiknmXAXWxzw5o1M4sXLaROzerUDA9j5OvZcxj0h7bTNr6ZcTrui9n4/nv440TWrkFsdF163X4bR48e9bkdcD72qJphNI+tR8vG0bRpbnn7vP7Kf6hdrSItG0fTsnE0SxYtuILK5Tlz5gxNGzWgQVRdourW5MUXsrdieFjvlqz74ikSZvyb+/q0BOCVh+LYMOtp1k77F9NGDeb6kIIA9OoYw+qpwzMepxLGUKdaOZ/+H94QSJXSXatnmR2io2N01ZrL+10nJSVxICmJyKgoTpw4QePYaKbPmEONiAif2/d4PNSOqMZXC5ZQLjSUpg3rM3HylFyvDbByxfcULhzC3QP7kbBhiyOa4FvcnkvUnFy54ntCQkIYPLA/8es3A7B0yWJa3NyK4OBgnnnqSQBefOW1S2p7U4czu7Ff7gZPVM0wlixfzQ0lSmTse/2V/1C4cAjDHvSuknxIgcsP6FSVU6dOERISQkpKCq1aNOWN0W8T27DhFbXTb/BEVCnLpyPuolnfkZxN8TD33aHc//JUKpUrwXfxP+PxpPHSA10BeHrM+U6aNcNuZProwdS85YWMfX9tn07an4cczVh1I6N10fLVXh9f9vp8V209S1dww5o1nfi1a6lSJYxKlSuTL18+bu/Zi/nzsmfJmpPa4LyNbzpuxH0xG9/WbdsRHGwlkfqxDdm3z/e/qdvvuVuICCEhlt1ESkoKqSkpWZ4lEl6pDPFbdp2z2U3YQVyreixd/RMeu1j12s07KVe66N9e26NDNF8sSvT9P+IFgVSiLeCSZWactmbdv38foaHnfJjLlQt15EPrtrab+CPuSRM+ccTqwI3YRYTb4zrSulkDPv34w4z94z8YS4uGkTzwz7s56oC9scfjITa6HhVuLEWrNm1pEJu1c3zrr/tpEhlG8esLWza7TWsSWuZ829x+XRuxaNW2v722e7sopi+8/AjPCUSsFTzePvyNq8lSRB4UkS0islVEHK2ln5PWrIac4/URLxMUHEzP3nf4O5SLMn/xdyxbGc/UWfP5+MP3+GHlCgbcfQ/xm7bz7Q8JlC5TlmefetzndoKCgliTsIEdu/ayLn4tW7dk7dLK9p0HGTVhCfPGDmPuu8PYuH1vRo8S4IlB7fF40pj69fk2LvVrVeTPMyls+zXpQkl3CKCupZtWuLWAwUADoC7QWUTCnNB2y5r1xhvLsXfvnozf9+3bm2Ebmpu13SQn45786QQWfv0VH0+c7MjiBDdiL3uj9fqSJUvRqUsc6xPiKVXqnL1x3wGDWJ/gXK+saNGitGh5M4sXZ914buKc/9LkjtdpO+gtjh7/M8M2984usXRqXosB/57wt9fc3j46R3qV6QRQrnS1Z1kDWKOqf6pqKrAc8DmzuWnNGlO/Pjt2/MKunTs5e/YsX0ybyj86Z8+SNSe13SSn4l6yaCFvjhrJtJlfUqhQIUc0nY791KlTnMxkI/vd0iWER9TkwIFzvbCv580hPMI3e+PDhw9nzAY4ffo0S79ZQvXq4VnWOd9mty7TFqyjbeMaPDKgDd0fGsfpMynnHS8idGsXxReLLm3H6zSB5MHj5jzLLcDLInIDlhVuJ8Dnryw3rFnTCQ4O5s2336HLP9rj8XjoP2CgY77ebmqD8za+6bgR98VsfEe9PoK/zv7FLZ3aAVC/QSxj3s2eTbBbsR8+dJABfboDkJrq4bYevWjdtj1DB/dni21vXL7CTbwxZqxPcR9ISmLwwP54PB7SNI1u3XvQ6R+ds6wz5Y27KV403WZ3OsdOnubNJ3uQP18w89+z7pqv3byLB16eCkDTqDD2HjjCrn2/+xS/9+SOKUHe4urUIREZBAwFTgFbgb9U9aELjsnsGx7986+7XYvHkLNcauqQE7hp4ev22vArTR3yBffWhjs/dSgyKkaXrVzj9fHFCwdfvVOHVHW8qkaranPgCPDzRY75QFVjVDWmZImSboZjMBgM2cbV5Y4iUkpVD4lIBazrlVeeVWswGK4ZcsO1SG9xe234TPuaZQowTFWdWcNmMBiuCgLpmqXbVrjN3NQ3GAyBizUp3d9ReI+pOmQwGPyHSZYGg8FwZcww3GAwGLwgkG7wBHQhDYPBENg4udxRRDqIyHYR2SEiw52O1SRLg8HgPxzKliISBLwLdAQigN4i4kyxWBuTLA0Gg99wsFJ6A2CHqv5PVc8CU4GuTsaaq65ZJiYmJBfMK96udywBJLsUipvabusb7atH2239rGhXdLrx9YkJiwrlkxJXPjKDAiKSub7EB6r6gb1dDtiT6bm9gDOFbm1yVbJUVa/XO4rIOrfWibqp7ba+0b56tN3Wdzv2K6Gqvld4zkHMMNxgMFwN7APKZ/o91N7nGCZZGgyGq4F4oKqIVBKRfEAvYK6TDeSqYXgW+eDKh+RKbbf1jfbVo+22vtux5xiqmioi9wGLgCDgY1Xd6mQbucoK12AwGHIrZhhuMBgMXmCSpcFgMHiBSZYGrxAn7BZzGBEp7KJ2mUB8TwzZJ6CSpYhUF5FGIpLXXt7ktL7jmrZumIjEiEh+F7RrikgLu8iy09pNRaQvgKqq08lBRLqIyINOambS7gq8JiKlXNBuD8zm/KkqTmk3FJG+9s98DmtXtc/DPG6d61czAZMsReQ24EvgJWA8MExErnNIuxqAqnqcPolEpDMwCxgJTEhvyyHtjsAU4GHgUxEp45BuHhEJAcYB/xKReyEjYTpyzohIO+BFYJsTehdotwBeA75U1UMOa7eztcsCjzqsfQvWHeo2wGM4uGpGROKAGcC/gNHAPW72vK9GAiJZikheoCcwSFVbYyXN8sCTviZMO5ltEJHPwdmEKSKNsZJkf1W9Gcu0zZFqKCLSEngbuFtV44CzQC0ntFU1TVVPAhOxvpgai8jD6c/5qm+/L5OAIaq6RESuF5GKIuKMYThEAx/Z2jeKSFsRiRWR630RFZE2wFjgDqAqUENEmjsQL/bIYBjQR1X7A8eBeiJSSkQKOKB9D9BbVbsBm4C7gEdEpIiPoV8zBESytLkO6wQFawg0H8gL9Mnu8ND+Zr0PeAg4KyKTwfEe5muqut7efg4o7tBw/CBwj6qutXuUscB9IjJORLo7NGROxfpSmgg0EJHRIvKqWPhy7vyO5ctU1v4gzwHew+p5OxF7Zi/bGcBArL/zuyJSzAfdIKCfPX+vMLAdqAmOXNNNBQoC4XYHoCXQD3gLeNrHXmAqEAKUAVDVj4FdWGvDs25Ifq2iqgHxANpizchvZv8eBPQBJmPPF82m7o1YJ1IJrA/WZAdjDgKuy7QdCqwHStr7bnConX8DT9vbA7AqrpR0QLcKMNzefhT4E3jXoZjrAv/DKngwGOuLeyDWZYXiPmrXxkpkU4G77H2VgfeB9g7Ensf+2QE4ANR26D3pDiQAq4Fn7H2tgAlAXR+177U/K32Bl+3te4DxTsR+LTwCqWe5AlgM9BWR5qrqUdXPsZJd3eyKqup+VT2pqslYJ0/B9B6miESJSLgP2h5VPW7/KsBR4A9VPSwidwAviUjB7OpnaudlVX3J3p6A1Qt34ubDaaC6iAzG+rCNACqIyD2+CqvqRqxezQhV/VCtof/HQDGggo/am7Gu+cUClex9/8P6wvLZnF7tSxGquhDrGmNnB3rbqOoMrOuVK7C+VFHVZUARfL9+OQVYANwMFFTVO1V1HFDaqWv/VzsBs9xRVc+IyGeAYt10CAf+AkoDSQ618budCEaKyE9YH66bHdJOBU6KyB4ReRVoBwxQ1dO+6IqIqN11sH/vhvWe7PcpYKwvEhHZAzyDZWU8T0RuBnb4qm3rbyPTDR479pI48/dcgHXZ43mRjLJ/kVgJ30k2Yt1ge11VPb6KqeoREVkG9BCRs0ABrIS/yUfdY8BnIjIlPdmLSD+gOOBz3NcE/u7aZvUB5MNKYFOxhieRLrTxMA4Or2xNsWP/FfgNqOpwzPmBQcBWoJaDuuWB6Ey/53Hh/RasIfg2oKbD2lHAK8AoJ/+eF7QxHbjJQb2iwAPAcqy1zj4NwS/RRvr77cp7cjU+AnZtuH0DRtWBu7MX6BbDOvkfVVWfvs0voT8AiFenF/lbMwbaAr+q6nYntW3983qwTmsDLYADqvqTG224gZvvia1fBOt6/PErHpx17YpAXlV1ZJRwLRCwydJNRKSAqp5xSdvVD5jBYHAHkywNBoPBCwLpbrjBYDD4DZMsDQaDwQtMsjQYDAYvMMnSkGsRkfYiUs/fcRgMYJLlNYuIeERkg4hsEZEvfCliISITRKS7vf2RiERkQ+NrESma6fdWQHusSd8Gg98xd8OvUUTkpKqG2NufAQmqOjrT88FqrTryRmsCMF+t5XoGw1WJ6VkawFqLHCYiLUVkhYjMBbaJSJCIjBSReBHZlL4m3F4H/Y6IbBeRb4CMArsi8p2IxNjbHUQkUUQ2ishSe1+IiHwiIpttzW72/l0iUsLefsTu8W4RkYfsfTeJyI8i8qGIbBWRxU6sqzcYvCVg1oYb3EFEgoGOwEJ7VxTWcsmdIjIEOKaq9e2ycqtEZDHWGuvqQATWOvRtwMcX6JYEPgSa21rF7aeesTVr28cVu+B10Vi1FmOxlkGuEZHlWLVAq2K7Nv3pAAABWUlEQVTVZBwsItOBbljVcwwG1zE9y2uXgiKyAViHtVZ9vL1/rarutLfbAf3s49YAN2AlrObAFLWqKu0Hll1EvyHwfbqWqv5h728DvJt+kKoeueB1TYHZqnpKrQLEs4Bm9nM7VXWDvZ0A3JT1/7bBkD1Mz/La5bSqnnen2a5feyrzLuB+VV10wXGd3A/vovyVaduDVSzXYMgRTM/ScDkWAf+0i3QgItXsit3fAz3ta5pluXgZu9VAcxGpZL82fRi+BMs+AXv/hZXLVwBxIlLIbutWe5/B4FdMsjRcjo+wrkcmisgWLAOzYCxbj1/s5z4F/nvhC1X1MDAEmCUiG4Fp9lMvAcXsmzcbuSDRqmoiVum9tVhD/4/0nC2HweA3zNQhg8Fg8ALTszQYDAYvMMnSYDAYvMAkS4PBYPACkywNBoPBC0yyNBgMBi8wydJgMBi8wCRLg8Fg8IL/B+14fCJq4nN6AAAAAElFTkSuQmCC\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Se grafica el elemento 6 (```test_data[5]```) para comprobar las soluciones. "
      ],
      "metadata": {
        "id": "HU9RhG_xnL_b"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "digit = test_data.reshape(10000,28, 28)[5]\n",
        "plt.imshow(digit, cmap = plt.cm.binary)\n",
        "plt.show()\n",
        "print(test_labels[5])"
      ],
      "metadata": {
        "id": "vOyRw8bXnL_b",
        "outputId": "2848e487-c4f9-4c0e-a956-e994b07d2f4c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 283
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAPsAAAD4CAYAAAAq5pAIAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAQYklEQVR4nO3dXWxVZboH8P/fUgRpxWIrEEALExKDx6hjJSaQkRMiUW50bgQvJhrNMBeSDMlcHKMX46U5Oc5kTE7GMIrDmNHJRMboBTlnODqJmRuwmKqgooAQCvSDLwEp9Os5F11OCnY9b9lrf9Hn/0ua7q5nr+7HZf+svfe73/XSzCAiU991tW5ARKpDYRcJQmEXCUJhFwlCYRcJYlo1H6y1tdXa29ur+ZBTwr59+9w6yZJqAJAajZk+fXqh/YeGhnJr111X7FyTeuylS5cW+v3XokOHDuHEiRMT/k8vFHaSDwH4HYAGAK+a2Yve/dvb29HZ2VnkIUNatWqVW29oaMitXX/99e6+Fy9edOupf5xT+/f29ubWmpub3X1HRkYK1bdv3+7Wp6KOjo7cWsn/tJJsAPDfAB4GsAzA4ySXlfr7RKSyijyPWg5gv5kdNLNBAH8B8Eh52hKRcisS9gUAjoz7uTvbdhmSG0h2kuzs7+8v8HAiUkTF3403s81m1mFmHW1tbZV+OBHJUSTsRwEsGvfzwmybiNShImH/CMBSkotJTgewHsB75WlLRMqt5KE3MxsmuRHA/2Js6G2Lme0tW2eBnD171q3v3esf1iIvjwYGBtz6gQMH3PqMGTPcujeWfsMNN7j7Dg4OunW9LLw6hcbZzWw7gHiDmSLXIH1cViQIhV0kCIVdJAiFXSQIhV0kCIVdJIiqzmeXiaWmiabmpHtTPVPz0VP1lpaWkh8b8D9DkJrPnppeO3PmTLcul9OZXSQIhV0kCIVdJAiFXSQIhV0kCIVdJAgNvdWBbdu2ufWTJ0+69YULF+bWUkNjo6Ojbj11ddrU/t4U2uHhYXffb7/91q0fO3bMre/evTu3du+997r7TkU6s4sEobCLBKGwiwShsIsEobCLBKGwiwShsIsEoXH2OvDqq6+69fnz57v1W265JbfmraIKANOm+X8CR44cceupy0F7K8ymLkOd6q2vr8+t79q1K7emcXYRmbIUdpEgFHaRIBR2kSAUdpEgFHaRIBR2kSA0zl4H9u3b59Y7OjrcujdnfGhoyN03NR+9qanJraeWVfbmrM+ePdvdN1VPXYo6Nd89mkJhJ3kIwDkAIwCGzcz/qxSRminHmf3fzexEGX6PiFSQXrOLBFE07Abg7yR3k9ww0R1IbiDZSbKzv7+/4MOJSKmKhn2lmf0YwMMAniH5kyvvYGabzazDzDra2toKPpyIlKpQ2M3saPa9D8A7AJaXoykRKb+Sw05yFsnm728DWANgT7kaE5HyKvJu/FwA72TLCU8D8KaZ/U9Zuppijh8/7tZT13b35qsD/rzu1Fh0asnm1Hz21Jx0b5w+9RmA1HXlU4+duuZ9NCWH3cwOArirjL2ISAVp6E0kCIVdJAiFXSQIhV0kCIVdJAhNca2Cnp4et566HHOKmeXWZs6c6e574oQ/hyk1vXbPHv+jFefPn8+tpabPpoYkvctUA+mhuWh0ZhcJQmEXCUJhFwlCYRcJQmEXCUJhFwlCYRcJQuPsVfDVV1+59cbGRrc+a9askh87m4KcKzX99sCBA279nnvuceveZbJvu+02d9/U9NvUks6a4no5ndlFglDYRYJQ2EWCUNhFglDYRYJQ2EWCUNhFgtA4exV8+eWXbj01n/27775z69687jNnzrj7Fl2l5/7773frXV1dubXUZwAuXbrk1lP7p8bpo9GZXSQIhV0kCIVdJAiFXSQIhV0kCIVdJAiFXSQIjbNXwf79+9367Nmz3frg4KBb9+bDHzt2zN33ySefdOspTz31lFt/5ZVXcmujo6OFHjt13fhUPZrkmZ3kFpJ9JPeM2zaH5A6SX2ffWyrbpogUNZmn8X8E8NAV254F8L6ZLQXwfvaziNSxZNjN7EMAp67Y/AiArdntrQAeLXNfIlJmpb5BN9fMvr94WQ+AuXl3JLmBZCfJzv7+/hIfTkSKKvxuvI2tKpi7sqCZbTazDjPrKDrpQkRKV2rYe0nOB4Dse1/5WhKRSig17O8BeCK7/QSAd8vTjohUSnKcneRbAFYBaCXZDeDXAF4E8FeSTwM4DOCxSjZ5rTt79qxbT62hnpq3PTQ0VFINADZt2uTWU+677z637vWeGmdPjZOnrguvcfbLJcNuZo/nlFaXuRcRqSB9XFYkCIVdJAiFXSQIhV0kCIVdJAhNca2C1NLCTU1Nbj019DYwMJBbmzdvnrvvkiVL3HpRra2tubXU0NucOXPc+smTJ926d1wi0pldJAiFXSQIhV0kCIVdJAiFXSQIhV0kCIVdJAiNs1fBzTff7NaHh4cL/f7z58/n1h566MprhVaXN86fmoLqjdEDwKlTV14a8XJFL1U91ejMLhKEwi4ShMIuEoTCLhKEwi4ShMIuEoTCLhKExtmrIDVf/fTp0249NQ7vLQn90ksvufumpMaqr7vOP18sXrw4t9bd3e3um1pBaGRkxK2nfn80OrOLBKGwiwShsIsEobCLBKGwiwShsIsEobCLBKFx9ipILS188eJFt+7NVwcAM8utLVu2zN03JTWWnRpnv+OOO3Jr33zzjbtvc3OzW+/v73frLS0tbj2a5Jmd5BaSfST3jNv2AsmjJLuyr7WVbVNEiprM0/g/Apjocie/NbO7s6/t5W1LRMotGXYz+xCAf/0fEal7Rd6g20jy0+xpfu6LI5IbSHaS7Ey9xhKRyik17L8H8CMAdwM4DiB3toWZbTazDjPrSE1sEJHKKSnsZtZrZiNmNgrgDwCWl7ctESm3ksJOcv64H38KYE/efUWkPiTH2Um+BWAVgFaS3QB+DWAVybsBGIBDAH5RwR6veXfeeadb37lzp1tPjcMvXbo0t5Zanz0lNY6esnZt/qjsyy+/7O574cIFt97T0+PWU+u7R5MMu5k9PsHm1yrQi4hUkD4uKxKEwi4ShMIuEoTCLhKEwi4ShKa4VsG6devc+uuvv+7Wp03z/zedPXs2t/bBBx+4+65Zs8ate9NnJ+P222/PrS1atMjdNzXsl+rt3Llzbj0andlFglDYRYJQ2EWCUNhFglDYRYJQ2EWCUNhFgtA4exU0NDS49cbGRreeupS09/vfeOMNd9/UOHtqjD+ltbU1t5aaonr48GG3njouM2bMcOvR6MwuEoTCLhKEwi4ShMIuEoTCLhKEwi4ShMIuEoTG2etAarx4YGDArXvjybt27Sqpp2pIXSJ79+7dbn1oaMitp45bNDqziwShsIsEobCLBKGwiwShsIsEobCLBKGwiwShcfY6sGLFCrf+5ptvunVvaeLp06eX1FM1tLe3u/XTp0+79UuXLrn1kZGRq21pSkue2UkuIvkPkp+T3Evyl9n2OSR3kPw6+95S+XZFpFSTeRo/DOBXZrYMwP0AniG5DMCzAN43s6UA3s9+FpE6lQy7mR03s4+z2+cAfAFgAYBHAGzN7rYVwKOValJEiruqN+hItgO4B8BOAHPN7HhW6gEwN2efDSQ7SXb29/cXaFVEiph02Ek2AdgGYJOZXbaSoI2tsDfhKntmttnMOsyso62trVCzIlK6SYWdZCPGgv5nM/tbtrmX5PysPh9AX2VaFJFySA69kSSA1wB8YWa/GVd6D8ATAF7Mvr9bkQ4D2Lhxo1t/++233bq3tPGZM2fcfQ8ePOjWlyxZ4taLaG5uduupJZdHR0fdekuLBojGm8w4+woAPwPwGcmubNtzGAv5X0k+DeAwgMcq06KIlEMy7Gb2TwDMKa8ubzsiUin6uKxIEAq7SBAKu0gQCrtIEAq7SBCa4loHFixY4NZvuukmt+5dinpwcNDdN3Wp6UqOs6em3w4PD7v11BTX1H97NDqziwShsIsEobCLBKGwiwShsIsEobCLBKGwiwShcfYqGLuQT76xSwbke/DBB936tm3bcmupsex33/UvQ7B+/Xq3XkRTU5NbP3bsmFtPHdfUfPdodGYXCUJhFwlCYRcJQmEXCUJhFwlCYRcJQmEXCULj7FWQGu9taGhw62vXrnXr3nXlZ86c6e7b3d3t1itp9uzZbj01Hz11XfhTp05ddU9Tmc7sIkEo7CJBKOwiQSjsIkEo7CJBKOwiQSjsIkFMZn32RQD+BGAuAAOw2cx+R/IFAD8H0J/d9Tkz216pRq9l3vrpk7Fy5Uq37l13PrU+e09Pj1v/5JNP3Ppdd93l1j033nijW79w4YJbb2xsdOup6+1HM5kP1QwD+JWZfUyyGcBukjuy2m/N7L8q156IlMtk1mc/DuB4dvscyS8A+EuYiEjduarnlyTbAdwDYGe2aSPJT0luITnhZxdJbiDZSbKzv79/oruISBVMOuwkmwBsA7DJzM4C+D2AHwG4G2Nn/pcm2s/MNptZh5l1tLW1laFlESnFpMJOshFjQf+zmf0NAMys18xGzGwUwB8ALK9cmyJSVDLsHLv06WsAvjCz34zbPn/c3X4KYE/52xORcpnMu/ErAPwMwGcku7JtzwF4nOTdGBuOOwTgFxXpcApIXSq6qFtvvTW31tXVlVsD0sNXO3bscOtFht7OnTvn1gcGBkr+3QDQ29tbaP+pZjLvxv8TwER/rRpTF7mG6BN0IkEo7CJBKOwiQSjsIkEo7CJBKOwiQehS0lPA888/n1ubN2+eu29qnP2BBx4oqafJWLdunVufO3euW09NYV29evVV9zSV6cwuEoTCLhKEwi4ShMIuEoTCLhKEwi4ShMIuEgTNrHoPRvYDODxuUyuAE1Vr4OrUa2/12heg3kpVzt5uM7MJr/9W1bD/4MHJTjPrqFkDjnrtrV77AtRbqarVm57GiwShsIsEUeuwb67x43vqtbd67QtQb6WqSm81fc0uItVT6zO7iFSJwi4SRE3CTvIhkvtI7if5bC16yEPyEMnPSHaR7KxxL1tI9pHcM27bHJI7SH6dfZ9wjb0a9fYCyaPZsesiubZGvS0i+Q+Sn5PcS/KX2faaHjunr6oct6q/ZifZAOArAA8C6AbwEYDHzezzqjaSg+QhAB1mVvMPYJD8CYDzAP5kZv+WbftPAKfM7MXsH8oWM/uPOuntBQDna72Md7Za0fzxy4wDeBTAk6jhsXP6egxVOG61OLMvB7DfzA6a2SCAvwB4pAZ91D0z+xDAqSs2PwJga3Z7K8b+WKoup7e6YGbHzezj7PY5AN8vM17TY+f0VRW1CPsCAEfG/dyN+lrv3QD8neRukhtq3cwE5prZ8ex2DwD/2k3Vl1zGu5quWGa8bo5dKcufF6U36H5opZn9GMDDAJ7Jnq7WJRt7DVZPY6eTWsa7WiZYZvxfannsSl3+vKhahP0ogEXjfl6YbasLZnY0+94H4B3U31LUvd+voJt976txP/9ST8t4T7TMOOrg2NVy+fNahP0jAEtJLiY5HcB6AO/VoI8fIDkre+MEJGcBWIP6W4r6PQBPZLefAPBuDXu5TL0s4523zDhqfOxqvvy5mVX9C8BajL0jfwDA87XoIaevJQA+yb721ro3AG9h7GndEMbe23gawM0A3gfwNYD/AzCnjnp7A8BnAD7FWLDm16i3lRh7iv4pgK7sa22tj53TV1WOmz4uKxKE3qATCUJhFwlCYRcJQmEXCUJhFwlCYRcJQmEXCeL/ASFkGZTWLx26AAAAAElFTkSuQmCC\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[0. 1. 0. 0. 0. 0. 0. 0. 0. 0.]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "La clase del elemento 6 es la clase 1 (pantalones)."
      ],
      "metadata": {
        "id": "CvCPSI9bnL_b"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "predictions = model4.predict(test_data)\n",
        "print('Predicción del elemento 6: clase', np.argmax(predictions[5]))\n",
        "print(predictions[5])\n",
        "print('Suma de las predicciones del elemento 6:', np.sum(predictions[5]))"
      ],
      "metadata": {
        "id": "-9-stjJwnL_c",
        "outputId": "7848bf9f-b47d-4760-e765-50f19421c85d",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 0s 1ms/step\n",
            "Predicción del elemento 6: clase 1\n",
            "[6.2962039e-03 9.7851682e-01 8.9907710e-04 5.4108649e-03 5.5254996e-03\n",
            " 6.8562977e-06 2.4881342e-03 2.3528957e-04 1.6495606e-04 4.5615988e-04]\n",
            "Suma de las predicciones del elemento 6: 0.9999999\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "def plot_image(i, predictions_array, true_label, img):\n",
        "  predictions_array, true_label, img = predictions_array, true_label[i], img[i]\n",
        "  plt.grid(False)\n",
        "  plt.xticks([])\n",
        "  plt.yticks([])\n",
        "\n",
        "  plt.imshow(img, cmap = plt.cm.binary)    \n",
        "\n",
        "  predicted_label = np.argmax(predictions_array)\n",
        "  if predicted_label.all() == true_label.all():\n",
        "    color = 'blue'\n",
        "  else:\n",
        "    color = 'red'\n",
        "\n",
        "  plt.xlabel('{} {:2.0f}% ({})'.format(predicted_label,\n",
        "                                100*np.max(predictions_array),\n",
        "                                true_label),\n",
        "                                color = color)\n",
        "\n",
        "def plot_value_array(i, predictions_array, true_label):\n",
        "  predictions_array, true_label = predictions_array, true_label[i]\n",
        "  plt.grid(False)\n",
        "  plt.xticks(range(10))\n",
        "  plt.yticks([])\n",
        "  thisplot = plt.bar(range(10), predictions_array, color = '#00FF00')\n",
        "  plt.ylim([0, 1])\n",
        "  predicted_label = np.argmax(predictions_array)\n",
        "\n",
        "  thisplot[predicted_label].set_color('red')\n",
        "  thisplot[np.argmax(true_label)].set_color('black')"
      ],
      "metadata": {
        "id": "46vLUQeEnL_c"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "plt.figure(figsize = (6,3))\n",
        "plt.subplot(1,2,1)\n",
        "plot_image(5, predictions[5], test_labels, test_data.reshape(10000,28,28))\n",
        "plt.subplot(1,2,2)\n",
        "plot_value_array(5, predictions[5],  test_labels)\n",
        "plt.show()"
      ],
      "metadata": {
        "id": "hf_E71v6nL_c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 211
        },
        "outputId": "3815fedf-753a-408c-aa1c-8d0abd4d9d1a"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x216 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Caso 5"
      ],
      "metadata": {
        "id": "Fby8RmBGnd2V"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Crear la red de neuronas"
      ],
      "metadata": {
        "id": "CHNZmMSPnd2V"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Se crea la red neuronal con una capa de entrada con 512 neuronas y función de activación ```relu```, y una capa de salida con función de activación ```softmax```."
      ],
      "metadata": {
        "id": "l38MSK_mnd2W"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model5 = models.Sequential()\n",
        "model5.add(layers.Dense(512, activation ='relu', name = 'Capa5', input_shape = (28*28,)))\n",
        "model5.add(layers.Dense(10, name = 'CapaSalida5', activation = 'softmax'))"
      ],
      "metadata": {
        "id": "h1P-qXZRnd2W"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Explicar model.summary()\n"
      ],
      "metadata": {
        "id": "yp-jURqhnd2W"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "El resumen del modelo indica el número de parámetros de salida de cada capa. Este número es igual al tamaño de los datos de entrada multiplicado por el número de neuronas, y sumando el número de umbrales de activación de la capa.\n",
        "\n",
        "\n",
        "En el caso de la primera capa: 401920 = 784 (tamaño de los datos) x 512 (número de neuronas) + 512 (número de umbrales de activación).\n",
        "\n",
        "En el caso de la segunda capa: 262656 = 512 (tamaño de los datos) x 512 (número de neuronas) + 512 (número de umbrales de activación)."
      ],
      "metadata": {
        "id": "MviPbsMknd2W"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model5.summary()"
      ],
      "metadata": {
        "id": "alncmfXPnd2W",
        "outputId": "ab6c4900-32da-4ed8-c05f-cc163dfcd460",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential_12\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " Capa5 (Dense)               (None, 512)               401920    \n",
            "                                                                 \n",
            " CapaSalida5 (Dense)         (None, 10)                5130      \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 407,050\n",
            "Trainable params: 407,050\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Se definen las funciones del optimizador, la pérdida y la métrica."
      ],
      "metadata": {
        "id": "pqPw06uRnd2W"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model5.compile(optimizer = 'sgd',\n",
        "              loss = 'categorical_crossentropy',\n",
        "              metrics = ['accuracy'])"
      ],
      "metadata": {
        "id": "Rz-hhHlVnd2W"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Entrenar la red"
      ],
      "metadata": {
        "id": "EqVks6j0nd2W"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Para entrenar la red se realizan ```5``` ciclos con lotes de ```128``` imágenes. Se compara con los datos de validación para controlar el sobreentrenamiento. "
      ],
      "metadata": {
        "id": "5CwoD-17nd2W"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "history = model5.fit(train_val_data, train_val_labels,\n",
        "                    epochs = 5,\n",
        "                    batch_size = 128,\n",
        "                    validation_data = (val_data, val_labels))"
      ],
      "metadata": {
        "id": "BSfT9_8Qnd2W",
        "outputId": "d12ded03-7cf0-41c0-c7e5-d0c6d97e7030",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/5\n",
            "391/391 [==============================] - 4s 10ms/step - loss: 1.0667 - accuracy: 0.6788 - val_loss: 0.7629 - val_accuracy: 0.7648\n",
            "Epoch 2/5\n",
            "391/391 [==============================] - 5s 12ms/step - loss: 0.6890 - accuracy: 0.7806 - val_loss: 0.6399 - val_accuracy: 0.7928\n",
            "Epoch 3/5\n",
            "391/391 [==============================] - 3s 9ms/step - loss: 0.6047 - accuracy: 0.8048 - val_loss: 0.5862 - val_accuracy: 0.8089\n",
            "Epoch 4/5\n",
            "391/391 [==============================] - 6s 15ms/step - loss: 0.5597 - accuracy: 0.8158 - val_loss: 0.5470 - val_accuracy: 0.8220\n",
            "Epoch 5/5\n",
            "391/391 [==============================] - 4s 11ms/step - loss: 0.5306 - accuracy: 0.8242 - val_loss: 0.5245 - val_accuracy: 0.8273\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "La pérdida indica el porcentaje de errores entre las etiquetas y las predicciones del modelo.\n",
        "\n",
        "La precisión indica el porcentaje de aciertos entre las predicciones y las etiquetas."
      ],
      "metadata": {
        "id": "n9KBvCkgnd2X"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "test_loss, test_acc = model.evaluate(test_data, test_labels)"
      ],
      "metadata": {
        "id": "mE2Kwm4gnd2X",
        "outputId": "b3e13ee4-1983-4786-b4cc-9a4d27939b8f",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 1s 2ms/step - loss: 0.6463 - accuracy: 0.7808\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "acc = history.history['accuracy']\n",
        "val_acc = history.history['val_accuracy']\n",
        "loss = history.history['loss']\n",
        "val_loss = history.history['val_loss']\n",
        "\n",
        "epochs = range(1, len(acc) + 1)\n",
        "\n",
        "# \"bo\" es para \"punto azun\"\n",
        "plt.plot(epochs, loss, 'bo', label='Pérdida de Entrenamiento')\n",
        "# \"b\" es para \"linea continua azul\"\n",
        "plt.plot(epochs, val_loss, 'b', label='Pérdida de Validación')\n",
        "plt.title('Pérdida de Entrenamiento y Validación')\n",
        "plt.xlabel('Epochs')\n",
        "plt.ylabel('Loss')\n",
        "plt.legend()\n",
        "\n",
        "plt.show()"
      ],
      "metadata": {
        "id": "_Kt_FiwkzJDQ",
        "outputId": "08d4a13c-8ee9-493d-8c2d-d72d232112ae",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Predecir valores"
      ],
      "metadata": {
        "id": "7ygO_7Uqnd2X"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Imprimir la matriz de confusión, que muestra los valores predichos frente a los reales."
      ],
      "metadata": {
        "id": "3NjWdk3Tnd2X"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Note, this code is taken straight from the SKLEARN website, an nice way of \n",
        "# viewing confusion matrix.\n",
        "def plot_confusion_matrix(cm, classes,\n",
        "                          normalize = False,\n",
        "                          title = 'Confusion matrix',\n",
        "                          cmap = plt.cm.Blues):\n",
        "    '''\n",
        "    This function prints and plots the confusion matrix.\n",
        "    Normalization can be applied by setting `normalize = True`.\n",
        "    '''\n",
        "    plt.imshow(cm, interpolation = 'nearest', cmap = cmap)\n",
        "    plt.title(title)\n",
        "    plt.colorbar()\n",
        "    tick_marks = np.arange(len(classes))\n",
        "    plt.xticks(tick_marks, classes, rotation = 45)\n",
        "    plt.yticks(tick_marks, classes)\n",
        "\n",
        "    if normalize:\n",
        "        cm = cm.astype('float') / cm.sum(axis = 1)[:, np.newaxis]\n",
        "\n",
        "    thresh = cm.max() / 2.\n",
        "    for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):\n",
        "        plt.text(j, i, cm[i, j],\n",
        "                 horizontalalignment = 'center',\n",
        "                 color = 'white' if cm[i, j] > thresh else 'black')\n",
        "\n",
        "    plt.tight_layout()\n",
        "    plt.ylabel('Observación')\n",
        "    plt.xlabel('Predicción')"
      ],
      "metadata": {
        "id": "h_jCBg67nd2X"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "Y_pred = model.predict(test_data)\n",
        "print('Valores de predicción del modelo')\n",
        "# Convertir en predicciones de clases desde los vectores one-hot encoding \n",
        "Y_pred_classes = np.argmax(Y_pred, axis = 1) \n",
        "# Convertir en clases de observación desde los vectores one-hot encoding\n",
        "Y_true = np.argmax(test_labels, axis = 1) \n",
        "# Computar la matriz de confusión\n",
        "confusion_mtx = confusion_matrix(Y_true, Y_pred_classes) \n",
        "# Dibujar la matriz de confusión\n",
        "print('Matriz de confusión')\n",
        "plot_confusion_matrix(confusion_mtx, classes = range(10))"
      ],
      "metadata": {
        "id": "IbG_zahlnd2X",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 366
        },
        "outputId": "db93236a-97bc-489d-9ffe-ebc5add7c4f2"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 0s 1ms/step\n",
            "Valores de predicción del modelo\n",
            "Matriz de confusión\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Se grafica el elemento 6 (```test_data[5]```) para comprobar las soluciones. "
      ],
      "metadata": {
        "id": "mhMLH4j0nd2X"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "digit = test_data.reshape(10000,28, 28)[5]\n",
        "plt.imshow(digit, cmap = plt.cm.binary)\n",
        "plt.show()\n",
        "print(test_labels[5])"
      ],
      "metadata": {
        "id": "fpRo7Yvznd2X",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 283
        },
        "outputId": "3860d98f-1eb2-41ff-fb32-11d53cee299d"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[0. 1. 0. 0. 0. 0. 0. 0. 0. 0.]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "La clase del elemento 6 es la clase 1 (pantalones)."
      ],
      "metadata": {
        "id": "OZyW--iOnd2X"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "predictions = model5.predict(test_data)\n",
        "print('Predicción del elemento 6: clase', np.argmax(predictions[5]))\n",
        "print(predictions[5])\n",
        "print('Suma de las predicciones del elemento 6:', np.sum(predictions[5]))"
      ],
      "metadata": {
        "id": "RhNMXTtfnd2X",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "df7ef9dc-ae50-4d8d-ad60-64fecc49d549"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 1s 3ms/step\n",
            "Predicción del elemento 6: clase 1\n",
            "[3.3270619e-03 9.8760003e-01 5.5881060e-04 1.3451112e-03 6.2468946e-03\n",
            " 7.4574364e-06 8.8897435e-04 5.3125800e-06 1.6993836e-05 3.4966970e-06]\n",
            "Suma de las predicciones del elemento 6: 1.0000001\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "def plot_image(i, predictions_array, true_label, img):\n",
        "  predictions_array, true_label, img = predictions_array, true_label[i], img[i]\n",
        "  plt.grid(False)\n",
        "  plt.xticks([])\n",
        "  plt.yticks([])\n",
        "\n",
        "  plt.imshow(img, cmap = plt.cm.binary)    \n",
        "\n",
        "  predicted_label = np.argmax(predictions_array)\n",
        "  if predicted_label.all() == true_label.all():\n",
        "    color = 'blue'\n",
        "  else:\n",
        "    color = 'red'\n",
        "\n",
        "  plt.xlabel('{} {:2.0f}% ({})'.format(predicted_label,\n",
        "                                100*np.max(predictions_array),\n",
        "                                true_label),\n",
        "                                color = color)\n",
        "\n",
        "def plot_value_array(i, predictions_array, true_label):\n",
        "  predictions_array, true_label = predictions_array, true_label[i]\n",
        "  plt.grid(False)\n",
        "  plt.xticks(range(10))\n",
        "  plt.yticks([])\n",
        "  thisplot = plt.bar(range(10), predictions_array, color = '#00FF00')\n",
        "  plt.ylim([0, 1])\n",
        "  predicted_label = np.argmax(predictions_array)\n",
        "\n",
        "  thisplot[predicted_label].set_color('red')\n",
        "  thisplot[np.argmax(true_label)].set_color('black')"
      ],
      "metadata": {
        "id": "Fvnolvvwnd2Y"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "plt.figure(figsize = (6,3))\n",
        "plt.subplot(1,2,1)\n",
        "plot_image(5, predictions[5], test_labels, test_data.reshape(10000,28,28))\n",
        "plt.subplot(1,2,2)\n",
        "plot_value_array(5, predictions[5],  test_labels)\n",
        "plt.show()"
      ],
      "metadata": {
        "id": "Zxy4UEWUnd2Y",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 211
        },
        "outputId": "a394a718-adf0-4aa2-96f7-c802587f490f"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x216 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Caso 6"
      ],
      "metadata": {
        "id": "G3lUo6NFnd2Y"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Crear la red de neuronas"
      ],
      "metadata": {
        "id": "X7dn455mnd2Y"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Se crea la red neuronal con una capa de entrada con 512 neuronas y función de activación ```relu```, y una capa de salida con función de activación ```softmax```."
      ],
      "metadata": {
        "id": "oyopFsfYnd2Y"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model6 = models.Sequential()\n",
        "model6.add(layers.Dense(512, activation ='relu', name = 'Capa6', input_shape = (28*28,)))\n",
        "model6.add(layers.Dense(10, name = 'CapaSalida6', activation = 'softmax'))"
      ],
      "metadata": {
        "id": "U6YXw4NOnd2Y"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Explicar model.summary()\n"
      ],
      "metadata": {
        "id": "YQhduq72nd2Y"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "El resumen del modelo indica el número de parámetros de salida de cada capa. Este número es igual al tamaño de los datos de entrada multiplicado por el número de neuronas, y sumando el número de umbrales de activación de la capa.\n",
        "\n",
        "\n",
        "En el caso de la primera capa: 401920 = 784 (tamaño de los datos) x 512 (número de neuronas) + 512 (número de umbrales de activación).\n",
        "\n",
        "En el caso de la segunda capa: 262656 = 512 (tamaño de los datos) x 512 (número de neuronas) + 512 (número de umbrales de activación)."
      ],
      "metadata": {
        "id": "-rLIUvZ6nd2Y"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model6.summary()"
      ],
      "metadata": {
        "id": "-s1c9nGOnd2Y",
        "outputId": "d9c2b901-243f-4264-e184-c2c423148e5d",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential_13\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " Capa6 (Dense)               (None, 512)               401920    \n",
            "                                                                 \n",
            " CapaSalida6 (Dense)         (None, 10)                5130      \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 407,050\n",
            "Trainable params: 407,050\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Se definen las funciones del optimizador, la pérdida y la métrica."
      ],
      "metadata": {
        "id": "IDS9n7w_nd2Y"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model6.compile(optimizer = 'rmsprop',\n",
        "              loss = 'categorical_crossentropy',\n",
        "              metrics = ['accuracy'])"
      ],
      "metadata": {
        "id": "JNLHIygJnd2Y"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Entrenar la red"
      ],
      "metadata": {
        "id": "lKWZ7mPDnd2Z"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Para entrenar la red se realizan ```5``` ciclos con lotes de ```128``` imágenes. Se compara con los datos de validación para controlar el sobreentrenamiento. "
      ],
      "metadata": {
        "id": "VmZbiPShnd2Z"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "history = model6.fit(train_val_data, train_val_labels,\n",
        "                    epochs = 5,\n",
        "                    batch_size = 128,\n",
        "                    validation_data = (val_data, val_labels))"
      ],
      "metadata": {
        "id": "LVVfhCk6nd2Z",
        "outputId": "44f9d97c-be79-4a2c-a73f-9d4e6a574352",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/5\n",
            "391/391 [==============================] - 5s 12ms/step - loss: 0.5745 - accuracy: 0.7952 - val_loss: 0.5298 - val_accuracy: 0.7944\n",
            "Epoch 2/5\n",
            "391/391 [==============================] - 4s 11ms/step - loss: 0.3947 - accuracy: 0.8552 - val_loss: 0.3720 - val_accuracy: 0.8655\n",
            "Epoch 3/5\n",
            "391/391 [==============================] - 5s 13ms/step - loss: 0.3480 - accuracy: 0.8718 - val_loss: 0.3439 - val_accuracy: 0.8761\n",
            "Epoch 4/5\n",
            "391/391 [==============================] - 4s 11ms/step - loss: 0.3184 - accuracy: 0.8826 - val_loss: 0.3449 - val_accuracy: 0.8745\n",
            "Epoch 5/5\n",
            "391/391 [==============================] - 4s 11ms/step - loss: 0.2979 - accuracy: 0.8912 - val_loss: 0.3384 - val_accuracy: 0.8749\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "La pérdida indica el porcentaje de errores entre las etiquetas y las predicciones del modelo.\n",
        "\n",
        "La precisión indica el porcentaje de aciertos entre las predicciones y las etiquetas."
      ],
      "metadata": {
        "id": "itleesJ3nd2Z"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "test_loss, test_acc = model.evaluate(test_data, test_labels)"
      ],
      "metadata": {
        "id": "beA62uXBnd2Z",
        "outputId": "4e38525f-0a73-41d9-d693-817b792f2547",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 1s 2ms/step - loss: 0.6463 - accuracy: 0.7808\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "acc = history.history['accuracy']\n",
        "val_acc = history.history['val_accuracy']\n",
        "loss = history.history['loss']\n",
        "val_loss = history.history['val_loss']\n",
        "\n",
        "epochs = range(1, len(acc) + 1)\n",
        "\n",
        "# \"bo\" es para \"punto azun\"\n",
        "plt.plot(epochs, loss, 'bo', label='Pérdida de Entrenamiento')\n",
        "# \"b\" es para \"linea continua azul\"\n",
        "plt.plot(epochs, val_loss, 'b', label='Pérdida de Validación')\n",
        "plt.title('Pérdida de Entrenamiento y Validación')\n",
        "plt.xlabel('Epochs')\n",
        "plt.ylabel('Loss')\n",
        "plt.legend()\n",
        "\n",
        "plt.show()"
      ],
      "metadata": {
        "id": "rewJQlO1zLYo",
        "outputId": "77feaa7e-6789-4c91-d41c-6129b0d8c230",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Predecir valores"
      ],
      "metadata": {
        "id": "jNEZyTwand2Z"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Imprimir la matriz de confusión, que muestra los valores predichos frente a los reales."
      ],
      "metadata": {
        "id": "55HJcQ9Xnd2Z"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Note, this code is taken straight from the SKLEARN website, an nice way of \n",
        "# viewing confusion matrix.\n",
        "def plot_confusion_matrix(cm, classes,\n",
        "                          normalize = False,\n",
        "                          title = 'Confusion matrix',\n",
        "                          cmap = plt.cm.Blues):\n",
        "    '''\n",
        "    This function prints and plots the confusion matrix.\n",
        "    Normalization can be applied by setting `normalize = True`.\n",
        "    '''\n",
        "    plt.imshow(cm, interpolation = 'nearest', cmap = cmap)\n",
        "    plt.title(title)\n",
        "    plt.colorbar()\n",
        "    tick_marks = np.arange(len(classes))\n",
        "    plt.xticks(tick_marks, classes, rotation = 45)\n",
        "    plt.yticks(tick_marks, classes)\n",
        "\n",
        "    if normalize:\n",
        "        cm = cm.astype('float') / cm.sum(axis = 1)[:, np.newaxis]\n",
        "\n",
        "    thresh = cm.max() / 2.\n",
        "    for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):\n",
        "        plt.text(j, i, cm[i, j],\n",
        "                 horizontalalignment = 'center',\n",
        "                 color = 'white' if cm[i, j] > thresh else 'black')\n",
        "\n",
        "    plt.tight_layout()\n",
        "    plt.ylabel('Observación')\n",
        "    plt.xlabel('Predicción')"
      ],
      "metadata": {
        "id": "CR-kxR90nd2Z"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "Y_pred = model6.predict(test_data)\n",
        "print('Valores de predicción del modelo')\n",
        "# Convertir en predicciones de clases desde los vectores one-hot encoding \n",
        "Y_pred_classes = np.argmax(Y_pred, axis = 1) \n",
        "# Convertir en clases de observación desde los vectores one-hot encoding\n",
        "Y_true = np.argmax(test_labels, axis = 1) \n",
        "# Computar la matriz de confusión\n",
        "confusion_mtx = confusion_matrix(Y_true, Y_pred_classes) \n",
        "# Dibujar la matriz de confusión\n",
        "print('Matriz de confusión')\n",
        "plot_confusion_matrix(confusion_mtx, classes = range(10))"
      ],
      "metadata": {
        "id": "b2kN3V_Xnd2Z",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 366
        },
        "outputId": "46e3cb45-d78a-4f01-dfdf-37b03044556d"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 1s 3ms/step\n",
            "Valores de predicción del modelo\n",
            "Matriz de confusión\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Se grafica el elemento 6 (```test_data[5]```) para comprobar las soluciones. "
      ],
      "metadata": {
        "id": "a_ab78Ycnd2Z"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "digit = test_data.reshape(10000,28, 28)[5]\n",
        "plt.imshow(digit, cmap = plt.cm.binary)\n",
        "plt.show()\n",
        "print(test_labels[5])"
      ],
      "metadata": {
        "id": "y-piBt2fnd2Z",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 283
        },
        "outputId": "1d0abaa2-a31b-4cbd-d5d7-28599813f235"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[0. 1. 0. 0. 0. 0. 0. 0. 0. 0.]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "La clase del elemento 6 es la clase 1 (pantalones)."
      ],
      "metadata": {
        "id": "_iWtFzYZnd2a"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "predictions = model6.predict(test_data)\n",
        "print('Predicción del elemento 6: clase', np.argmax(predictions[5]))\n",
        "print(predictions[5])\n",
        "print('Suma de las predicciones del elemento 6:', np.sum(predictions[5]))"
      ],
      "metadata": {
        "id": "c9qSVRIand2a",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "bf1ea17a-4bd1-4678-c9b3-d121cb42e28f"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 1s 3ms/step\n",
            "Predicción del elemento 6: clase 1\n",
            "[6.6349116e-06 9.9999148e-01 2.2898629e-08 5.2170105e-07 1.0920622e-06\n",
            " 9.3871785e-11 2.2505630e-07 4.8676564e-14 8.1048432e-11 7.0573301e-14]\n",
            "Suma de las predicciones del elemento 6: 0.99999994\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "def plot_image(i, predictions_array, true_label, img):\n",
        "  predictions_array, true_label, img = predictions_array, true_label[i], img[i]\n",
        "  plt.grid(False)\n",
        "  plt.xticks([])\n",
        "  plt.yticks([])\n",
        "\n",
        "  plt.imshow(img, cmap = plt.cm.binary)    \n",
        "\n",
        "  predicted_label = np.argmax(predictions_array)\n",
        "  if predicted_label.all() == true_label.all():\n",
        "    color = 'blue'\n",
        "  else:\n",
        "    color = 'red'\n",
        "\n",
        "  plt.xlabel('{} {:2.0f}% ({})'.format(predicted_label,\n",
        "                                100*np.max(predictions_array),\n",
        "                                true_label),\n",
        "                                color = color)\n",
        "\n",
        "def plot_value_array(i, predictions_array, true_label):\n",
        "  predictions_array, true_label = predictions_array, true_label[i]\n",
        "  plt.grid(False)\n",
        "  plt.xticks(range(10))\n",
        "  plt.yticks([])\n",
        "  thisplot = plt.bar(range(10), predictions_array, color = '#00FF00')\n",
        "  plt.ylim([0, 1])\n",
        "  predicted_label = np.argmax(predictions_array)\n",
        "\n",
        "  thisplot[predicted_label].set_color('red')\n",
        "  thisplot[np.argmax(true_label)].set_color('black')"
      ],
      "metadata": {
        "id": "IFpr2G1Mnd2a"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "plt.figure(figsize = (6,3))\n",
        "plt.subplot(1,2,1)\n",
        "plot_image(5, predictions[5], test_labels, test_data.reshape(10000,28,28))\n",
        "plt.subplot(1,2,2)\n",
        "plot_value_array(5, predictions[5],  test_labels)\n",
        "plt.show()"
      ],
      "metadata": {
        "id": "sHzHKQ3Lnd2a",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 211
        },
        "outputId": "1d0555a1-1608-47d3-b351-9588d3cde4b4"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x216 with 2 Axes>"
            ],
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAWoAAADCCAYAAABg4IjUAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAASlUlEQVR4nO3da7BeVXnA8f/iJCEnJISQhIvckthURRGBQNNClZFCIXXQtirptCqi09qKlXacDpZerPVDvY61dsq0QQpU8EJUHIcWUWz9UomJJhIKoQkECSEJuREgIdenH/aOTbLXPu8+h3NyFvH/m3nnvOd513rX2ufynHX2WmvvFBFIksp11Gh3QJI0MBO1JBXORC1JhTNRS1LhTNSSVDgTtSQVbsxod0B6KZk2bVrMmDFjWN9z2bJl7Nmzp1PZMWPGcPbZZw9r+yrHkiVLNkbE9EPjyXXUUndz5syJxYsXD+t7ppQGVd7f2SNXSmlJRMw5NO6pD0kqnIlakgpnopakwo3IZOJITLiUYsWKFY1Y2znGXDx3fnHcuHHZ+rmyu3fvzpY96qjuf3Nz7zt79uzO9Ufb6tWr2bhx4+BO7EovYSOSqGfMmMFwT7iU4uKLL27E+vr6smWPPvroRuyFF15oxNr+qOXKrl+/Plt20qRJjdjevXuzZXPxu+++O1u2RHPmNOZapCOapz4kqXAmakkqnBteWmzbti0bf/DBBxux6dMb69Nb7dixoxFbtWpVtuz48eMbsbZz0RMmTGjEdu3alS07mP5KGn2OqCWpcCZqSSqciVqSCmeilqTCmaglqXCu+miR22wC+d2GbRtLcjsOc7EpU6Zk6+fet201Sm41SNtGmv7+/mxcUpkcUUtS4UzUklQ4E7UkFc5ELUmFczKxxcKFC7PxTZs2NWKnnnpqtmxuMnDfvn2NWO4qe21lc1vQgew995555pls2bVr1zZiS5YsyZY977zzsnFJh48jakkqnIlakgpnopakwpmoJalwJmpJKpyrPlosWLAgGz/55JMbsRNOOCFbNnd/wzFjml/yJ554Ils/dzOAtvsz5m4ykGsLYMOGDY3YokWLsmVd9SGNPkfUklQ4E7UkFc5ELUmFM1FLUuGcTGyxYsWKbHzOnDmNWNu27t27dzdiuW3hEydOzNbP3UU8t1UcYPLkyZ1ikL92dW5buaQyOKKWpMKZqCWpcCZqSSqciVqSCmeilqTCueoDeOqppxqxtjuL57aL57ZkQ351Re4u5G1byHPbwttWiORWmLStEMm9b9vNCySNPkfUklQ4E7UkFc5ELUmFM1FLUuGcTATWrVvXiOWuBd0mIrLx/v7+Rmzjxo2NWG5bOsDy5csbseeeey5bNjfJ2DYhmrumdW6CUVIZHFFLUuFM1JJUOBO1JBXORC1JhTNRS1LhXPUBPPLII43Y2LFjs2WPOeaYzu+bUmrEctvVV61ala1/zjnnNGJtNzQ444wzGrHcdnXI353cLeRSuRxRS1LhTNSSVDgTtSQVzkQtSYVzMhF4+OGHG7G2LeTPP/98I5bbkg2wdevWRmz69Omd+zV37txGbOnSpdmyuYnLnTt3di7bNvEoafQ5opakwpmoJalwJmpJKpyJWpIKZ6KWpMK56gNYuXJlIzZ58uRs2V27djVibdvN165d24hdffXVnft1zTXXNGI33nhjtuy+ffs6v29ulUrbyhVJo88RtSQVzkQtSYUzUUtS4UzUklQ4JxOBbdu2NWK5O4hDfvv17t27s2Vz8euuu65zv84///xO7UN+MrFtgjB37WknE6VyOaKWpMKZqCWpcCZqSSqciVqSCmeilqTCueqD/F25J06cmC2bW3WxY8eObNmTTjqpEZs1a9Yge3ewadOmZeO5VR/HH398tuymTZsasbZjkDT6HFFLUuFM1JJUOBO1JBXORC1JhXMyEZg6dWojtmfPns71n3vuuWz88ssvH3Kf2uQmKCG/Bbxt4nHz5s2N2GCuZy3p8HJELUmFM1FLUuFM1JJUOBO1JBXOyUTyuxC3bNmSLZubZMzdHBfg05/+dKf22ybyjjqq+Xd05syZ2bJr1qxpxKZPn54tu3fv3k71JZXBEbUkFc5ELUmFM1FLUuFM1JJUOBO1JBXOVR/k78r9wgsvZMvmtotHRLbsmWee2an93CoMyK/6ePWrX50t+9hjjzVikyZNypZ9+umnG7EpU6YM1EVJo8gRtSQVzkQtSYUzUUtS4UzUklQ4JxOBs846qxG7//77s2Vzk4yzZ8/Olm27dvShcpOGbebNm5eNf+5zn2vEtm/fni27bt26RqztRriSRp8jakkqnIlakgpnopakwpmoJalwJmpJKpyrPoCrrrqqEbv55puzZceMaX7Jtm3bli173333NWKXXXZZI9a2BT3nla98ZTZ+2mmnNWJtq0ly7T377LOd+yDp8HJELUmFM1FLUuFM1JJUOBO1JBXOyUSgr6+vERs7dmy2bO561Ln6ALfddlsjlptMzE1Qtpk2bVo2ntsW/vjjj2fL5o5h/Pjxnfsg6fByRC1JhTNRS1LhTNSSVDgTtSQVzkQtSYVz1UeL3MoIgB07djRibSsmFi1aNKx9GkjuhgZLlizJlt29e3cjljsuSWVwRC1JhTNRS1LhTNSSVDgTtSQVzsnEFhdeeGE2fvvttzdibXfwHjdu3LD2aSAzZsxoxLZs2ZItu3PnzkZs7969w90lScPEEbUkFc5ELUmFM1FLUuFM1JJUOBO1JBXOVR8trr322mz8zjvvbMTa7va9devWRuzRRx9txGbNmjXI3jVNmjSpEWu7s/i+ffsasSlTprzoPkgaGY6oJalwJmpJKpyJWpIKZ6KWpMI5mdjilFNOycaPO+64Rqzt2tW7du1qxHLXqB6OycTcdvU9e/Zky+a2kOf6KqkMjqglqXAmakkqnIlakgpnopakwpmoJalwrvoAIqIRSylly1566aWN2MKFC7Nlcysx7rrrrkZs/vz5vbrY08SJExuxtWvXZsvmjje3rVxSGRxRS1LhTNSSVDgTtSQVzkQtSYVzMpH8RFpfX1+27Lx58xqx3DWqAfr7+xuxNWvWDLJ33UyePLkRa9sWnrv29ObNm4e9T5KGhyNqSSqciVqSCmeilqTCmaglqXAmakkqnKs+aL+LeM5FF13UiLXdZCB3F/J169Y1YsuWLcvWP/vsszv369hjj23Etm/fni07duzYRix3QwRJZXBELUmFM1FLUuFM1JJUOBO1JBXOyUTarz3d1emnn56NL126tBHLTeTde++92fqDmUx89tlnG7EdO3Z0rr9+/frOZSUdXo6oJalwJmpJKpyJWpIKZ6KWpMKZqCWpcK76GAY33HBDNn7SSSc1YrlVH294wxtedB+uuuqqRuzEE0/Mls1tF7/kkktedB8kjQxH1JJUOBO1JBXORC1JhTNRS1LhUkQM/5um9DTw+LC/sVQ5IyKmj0bDc+bMicWLFw/rew72EgYj8TurMqSUlkTEnEPjI7LqY7R+iSTpSOSpD0kqnIlakgpnopakwpmoJalwvVd9pPQF4E3ABiJe01Lm9cBngdcC84m484DX3gX8Rf3Zx4i4pY6fB/wr0A/cDXyQiCCljwNXAEuJeGdd9veAaUR8tqX9k4F/IeJNpHQ18EngLiLeW7/+YeA9wF7gj4m4p8cxvw34CPAq4AIiek/ztx3PwHUG26/jgS8DM4DVwNuJ2NKjzuXA3wN9wAIi/q5H+VSXnwdsB64m4kc96swEvgRMBZYA7yBiV486+Z+L9vJHA7cC5wGbgKuIWH1ImX7gv4EzgZcBCbiNiMsHfO9BGMKKpmnAxiE0dbjrjUab1mvKr2iKiIEf8PqAcwOWD1BmRsBrA24NeOsB8eMDHq0/TqmfT6lfWxQwNyAF/HvAFQGTA+6tX18QcFZAf8B3A8YO0P4nA95cP7864PMHvHZmwLKAowNmBqwK6OtxzK8KeEXAfwbM6fk1ajuegcsPpV+fCLi+fn59wMd7lO+r33dWwLi6vTN71JlX9z/Vx3N/h2P/SsD8+vmNAX/Yo3z7z0V7nT8KuLF+Pj/gywOUXR0wrX5+c8CFnb6HI/AAFr8U6r2U+nqk18s9ep/6iPg+sLlHmdVE/ATYd8grvw7cS8RmqpHfvcDl9Qj4WCJ+UB/RrcBb6vpj61HdBGA38CHgH4jYPUAPfhv4j5bX3gx8iYidRDwGrAQu6HE8DxGxYsAyB2o/noEMvl9Vnf0jz1s6tHEBsJKIR6lGuF+q36NXG7fWPyE/AI6rjy+v+l69Edj/X1SXfuV/Lnr3a/+x3wlc0nEB8jeA3+1QTirWSJ+jPgV44oDP19SxU+rnB8cjnqU6bfBj4CngGeCXiPhGawvVv91biNg5yD4Mp/zx9K4z2H6dSMRT9fN1QP7yeC+ujcHWmQpsJWLPCLZxcJ2qrWfqtntZDPxqh3JSscq7zGnEJ4BPAJDSAuCvSOm9wGXAT4j42CE1TgaePqx9LEFEkJJb1HrbQHW+erT880uk3mi0ab2ORnpE/SRw2gGfn1rHnqyfHxr/fymdQzUZtAJ4GxFvB15OSrMPaWMHMH4IfRhOvY9nePq1/menIaqPG0agjcHW2UR1emRMx/Ivvl9VW5PrtnsZT/UzMioiYki/rIe73mi0ab3uRjpR3wNcRkpTSGkK1aj4nvrf922kNLc+z/hO4K5D6v4t8JfAWKoVC1Cdw55wSLlHqFZBtPkmMJ+Ujq5Pk8wGFr2IY2rqdjzD0a9vAu+qn7+rQxs/BGaT0kxSGgfMr9+jVxvvJKVESnOBZw443dJUnZP/HvDWQfQr/3PRu1/7j/2twH37Z2x6+EVgeYdyUrl6zjjCHQFPBewOWBPwnkyZ8+vXng/YFPDgAa9dE7Cyfrz7gPicgOX1qoTPR71UsH7tLQEfOeDzTwU8EPDFlj5+N+AX6ucHr/qoYjfU7aw4aDUG3B3wssz7/WZ9PDsD1gfcU8dfFnB3Sx/yxwPvC3hfS53B9mtqfaz/G/CdgOMPaHtBSxvzAh6p27nhgPhHA67MlE8B/1iXfyAOXPUCS1vamBXVqpeVAV8NOLqOXxnw0ZY6bT8XCyK30gbG1++9sm5rVuv35OBVHx8K+MBwzb4P5kE1QbqCaqL4+o51vkD1n1L7Kqt8vdOo/mD+D/Ag8MGO9cZTDRCW1fX+ZpDt9lHNKX1rEHVWAw8ASxnEygjgOKqJ5IeBh4Bf7ljvFXVb+x/bgOs61v2T+uuyHLgDGN+x3gfrOg92bWvA9zscP7Aj/qgS68fq581E7ePn63Fwov5+9Fr6NwKPOoGtAmYB4+pEOPDSyKre64Fzh5CoTwbOrZ9PovpPs0t7CZhYPx8L3A/MHUS7fwrcPoREPW0IX9NbgPfWz8cBxw3x+7KOar1yr7KnAI8B/fXnXwGu7lDvNXWSnkA1D/gd9g8kh/g4MnYmRnyd6psP1fnIK+qJSP08SamflJZSJZx9pDQd+Ay9NgWNjAuAlRHxaHRfGkl0WQ6br/dU1BuTolo99RAdVjdF5bn607H1I7q0mVI6FfgNYMR/11JKk6n+iN0EEBG7ImLrEN7qEmBVRHTdtDQG6E/VvMgEYG2HOq+i2n+wPaoVSv8F/NYQ+vozR0aiBohYUH/8MhEvZ/+uRP38iNhBxOuIOIVqjfbTDLS0c2QdjmWhWSmlGcA5VKPjLuX7UvUHbgPVhrNO9ah2I/8Zzf0TvQTw7ZTSkpTS73esM5NqddfNKaUfp5QWpJSOGWS7UM3T3NGpkxFPAp8Cfkq9XDgivt2h6nLgV1NKU1NKE6h2+Z7Wo86AjpxELYmU0kRgIdV50W1d6kTE3oh4HdXqmwtSSvlLRRzczpuADRGxZAjdvCgizqW6VMT7U3UJil7GUJ0S+qeIOAd4Hrh+MI2makL9SuCrHctPofovaCbVEs9jUnU5iwFFxEPAx4FvU23EW0p1mYghM1FLI+NwLAs9SEppLFWS/mJEfG2w9etTCd+j9y5RgAuBK1NKq6lO67wxpfRvHdt5sv64Afg6vXfkQvUfyZoDRvt3UiXuwbgC+FFErO9Y/teAxyLi6ah2Rn8N+JUuFSPipog4LyJeD2yhmjMYMhO1NDJ+CMxOKc1M3ZdGDlmqloXeBDwUEZ8ZRL3pKaXj6uf9wKVUqyoGFBEfjohTI2IG1bHdFxE9R5sppWNSSpP2P6damtlz+WRErAOeSCm9og5dQrXCZTB+h46nPWo/BeamlCbUX99LqM7995RSOqH+eDrV+enbB9nXg5S3M1E6AkTEnpTStVTrw/uAL0TEg73qpZTuAC4GpqWU1gB/HRE3dWjyQuAdwAP1+WaAP4+Iu3vUOxm4JaXURzVw+0pEfKtDe0N1IvD1+jItY4DbI6LtOj2H+gDwxfoP36PAu7s2Wv9RuBT4g651IuL+lNKdwI+APVTLELtuYlmYUppKdb2i9w9x4vNnRuTmtpKk4eOpD0kqnIlakgpnopakwpmoJalwJmpJKpyJWpIKZ6KWpMKZqCWpcP8HUXtNFlbO0GgAAAAASUVORK5CYII=\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Caso 7"
      ],
      "metadata": {
        "id": "l5GNinIhnd2a"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Crear la red de neuronas"
      ],
      "metadata": {
        "id": "_dZ1tr-3nd2a"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Se crea la red neuronal con una capa de \n",
        "entrada con 512 neuronas y función de activación ```sigmoid```, y una capa de salida con función de activación ```softmax```."
      ],
      "metadata": {
        "id": "Kd8mMW02nd2a"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model7 = models.Sequential()\n",
        "model7.add(layers.Dense(512, activation ='sigmoid', name = 'Capa7', input_shape = (28*28,)))\n",
        "model7.add(layers.Dense(10, name = 'CapaSalida7', activation = 'softmax'))"
      ],
      "metadata": {
        "id": "M6A4OIP6nd2a"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Explicar model.summary()\n"
      ],
      "metadata": {
        "id": "UbGycgtInd2a"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "El resumen del modelo indica el número de parámetros de salida de cada capa. Este número es igual al tamaño de los datos de entrada multiplicado por el número de neuronas, y sumando el número de umbrales de activación de la capa.\n",
        "\n",
        "\n",
        "En el caso de la primera capa: 401920 = 784 (tamaño de los datos) x 512 (número de neuronas) + 512 (número de umbrales de activación).\n",
        "\n",
        "En el caso de la segunda capa: 262656 = 512 (tamaño de los datos) x 512 (número de neuronas) + 512 (número de umbrales de activación)."
      ],
      "metadata": {
        "id": "RDhQ1Hq6nd2a"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model7.summary()"
      ],
      "metadata": {
        "id": "u3jY7Bbend2a",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "497d948d-3498-442b-a2d4-1829423c4d35"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential_14\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " Capa7 (Dense)               (None, 512)               401920    \n",
            "                                                                 \n",
            " CapaSalida7 (Dense)         (None, 10)                5130      \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 407,050\n",
            "Trainable params: 407,050\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Se definen las funciones del optimizador, la pérdida y la métrica."
      ],
      "metadata": {
        "id": "br4bFCCend2b"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model7.compile(optimizer = 'sgd',\n",
        "              loss = 'categorical_crossentropy',\n",
        "              metrics = ['accuracy'])"
      ],
      "metadata": {
        "id": "qtp5sWA4nd2b"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Entrenar la red"
      ],
      "metadata": {
        "id": "SLOrD3Mvnd2d"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Para entrenar la red se realizan ```5``` ciclos con lotes de ```128``` imágenes. Se compara con los datos de validación para controlar el sobreentrenamiento. "
      ],
      "metadata": {
        "id": "mr9jTFmInd2d"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "history = model7.fit(train_val_data, train_val_labels,\n",
        "                    epochs = 5,\n",
        "                    batch_size = 128,\n",
        "                    validation_data = (val_data, val_labels))"
      ],
      "metadata": {
        "id": "9Y_giitWnd2d",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "d0479e68-25be-4c3d-df3f-a5a4c954b9a9"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/5\n",
            "391/391 [==============================] - 4s 10ms/step - loss: 1.7737 - accuracy: 0.5759 - val_loss: 1.3986 - val_accuracy: 0.6587\n",
            "Epoch 2/5\n",
            "391/391 [==============================] - 4s 10ms/step - loss: 1.2171 - accuracy: 0.7069 - val_loss: 1.0712 - val_accuracy: 0.7200\n",
            "Epoch 3/5\n",
            "391/391 [==============================] - 4s 9ms/step - loss: 0.9915 - accuracy: 0.7269 - val_loss: 0.9199 - val_accuracy: 0.7362\n",
            "Epoch 4/5\n",
            "391/391 [==============================] - 4s 9ms/step - loss: 0.8755 - accuracy: 0.7388 - val_loss: 0.8334 - val_accuracy: 0.7425\n",
            "Epoch 5/5\n",
            "391/391 [==============================] - 3s 9ms/step - loss: 0.8047 - accuracy: 0.7490 - val_loss: 0.7760 - val_accuracy: 0.7546\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "La pérdida indica el porcentaje de errores entre las etiquetas y las predicciones del modelo.\n",
        "\n",
        "La precisión indica el porcentaje de aciertos entre las predicciones y las etiquetas."
      ],
      "metadata": {
        "id": "w8KEH-sund2d"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "test_loss, test_acc = model7.evaluate(test_data, test_labels)"
      ],
      "metadata": {
        "id": "LvSfrIgTnd2d",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "8b220dd0-6246-4c3b-ce00-104f9a1489d0"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 1s 3ms/step - loss: 0.7901 - accuracy: 0.7477\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "acc = history.history['accuracy']\n",
        "val_acc = history.history['val_accuracy']\n",
        "loss = history.history['loss']\n",
        "val_loss = history.history['val_loss']\n",
        "\n",
        "epochs = range(1, len(acc) + 1)\n",
        "\n",
        "# \"bo\" es para \"punto azun\"\n",
        "plt.plot(epochs, loss, 'bo', label='Pérdida de Entrenamiento')\n",
        "# \"b\" es para \"linea continua azul\"\n",
        "plt.plot(epochs, val_loss, 'b', label='Pérdida de Validación')\n",
        "plt.title('Pérdida de Entrenamiento y Validación')\n",
        "plt.xlabel('Epochs')\n",
        "plt.ylabel('Loss')\n",
        "plt.legend()\n",
        "\n",
        "plt.show()"
      ],
      "metadata": {
        "id": "LoiIq3MrzOQ4",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "outputId": "08e1905c-fa5e-488d-d2a3-9cb61e8b0eef"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Predecir valores"
      ],
      "metadata": {
        "id": "JSrgwZEhnd2d"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Imprimir la matriz de confusión, que muestra los valores predichos frente a los reales."
      ],
      "metadata": {
        "id": "eLLLCUggnd2d"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Note, this code is taken straight from the SKLEARN website, an nice way of \n",
        "# viewing confusion matrix.\n",
        "def plot_confusion_matrix(cm, classes,\n",
        "                          normalize = False,\n",
        "                          title = 'Confusion matrix',\n",
        "                          cmap = plt.cm.Blues):\n",
        "    '''\n",
        "    This function prints and plots the confusion matrix.\n",
        "    Normalization can be applied by setting `normalize = True`.\n",
        "    '''\n",
        "    plt.imshow(cm, interpolation = 'nearest', cmap = cmap)\n",
        "    plt.title(title)\n",
        "    plt.colorbar()\n",
        "    tick_marks = np.arange(len(classes))\n",
        "    plt.xticks(tick_marks, classes, rotation = 45)\n",
        "    plt.yticks(tick_marks, classes)\n",
        "\n",
        "    if normalize:\n",
        "        cm = cm.astype('float') / cm.sum(axis = 1)[:, np.newaxis]\n",
        "\n",
        "    thresh = cm.max() / 2.\n",
        "    for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):\n",
        "        plt.text(j, i, cm[i, j],\n",
        "                 horizontalalignment = 'center',\n",
        "                 color = 'white' if cm[i, j] > thresh else 'black')\n",
        "\n",
        "    plt.tight_layout()\n",
        "    plt.ylabel('Observación')\n",
        "    plt.xlabel('Predicción')"
      ],
      "metadata": {
        "id": "zUF16kY_nd2d"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "Y_pred = model7.predict(test_data)\n",
        "print('Valores de predicción del modelo')\n",
        "# Convertir en predicciones de clases desde los vectores one-hot encoding \n",
        "Y_pred_classes = np.argmax(Y_pred, axis = 1) \n",
        "# Convertir en clases de observación desde los vectores one-hot encoding\n",
        "Y_true = np.argmax(test_labels, axis = 1) \n",
        "# Computar la matriz de confusión\n",
        "confusion_mtx = confusion_matrix(Y_true, Y_pred_classes) \n",
        "# Dibujar la matriz de confusión\n",
        "print('Matriz de confusión')\n",
        "plot_confusion_matrix(confusion_mtx, classes = range(10))"
      ],
      "metadata": {
        "id": "6D3SCoqind2d",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 366
        },
        "outputId": "65200fda-efc6-4085-979e-3925c419b8f2"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 1s 3ms/step\n",
            "Valores de predicción del modelo\n",
            "Matriz de confusión\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Se grafica el elemento 6 (```test_data[5]```) para comprobar las soluciones. "
      ],
      "metadata": {
        "id": "wsZUHAEOnd2e"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "digit = test_data.reshape(10000,28, 28)[5]\n",
        "plt.imshow(digit, cmap = plt.cm.binary)\n",
        "plt.show()\n",
        "print(test_labels[5])"
      ],
      "metadata": {
        "id": "vduaIFHxnd2e",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 283
        },
        "outputId": "44b7af7f-8640-4a80-bfec-7faed8fe2b71"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAPsAAAD4CAYAAAAq5pAIAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAQYklEQVR4nO3dXWxVZboH8P/fUgRpxWIrEEALExKDx6hjJSaQkRMiUW50bgQvJhrNMBeSDMlcHKMX46U5Oc5kTE7GMIrDmNHJRMboBTlnODqJmRuwmKqgooAQCvSDLwEp9Os5F11OCnY9b9lrf9Hn/0ua7q5nr+7HZf+svfe73/XSzCAiU991tW5ARKpDYRcJQmEXCUJhFwlCYRcJYlo1H6y1tdXa29ur+ZBTwr59+9w6yZJqAJAajZk+fXqh/YeGhnJr111X7FyTeuylS5cW+v3XokOHDuHEiRMT/k8vFHaSDwH4HYAGAK+a2Yve/dvb29HZ2VnkIUNatWqVW29oaMitXX/99e6+Fy9edOupf5xT+/f29ubWmpub3X1HRkYK1bdv3+7Wp6KOjo7cWsn/tJJsAPDfAB4GsAzA4ySXlfr7RKSyijyPWg5gv5kdNLNBAH8B8Eh52hKRcisS9gUAjoz7uTvbdhmSG0h2kuzs7+8v8HAiUkTF3403s81m1mFmHW1tbZV+OBHJUSTsRwEsGvfzwmybiNShImH/CMBSkotJTgewHsB75WlLRMqt5KE3MxsmuRHA/2Js6G2Lme0tW2eBnD171q3v3esf1iIvjwYGBtz6gQMH3PqMGTPcujeWfsMNN7j7Dg4OunW9LLw6hcbZzWw7gHiDmSLXIH1cViQIhV0kCIVdJAiFXSQIhV0kCIVdJIiqzmeXiaWmiabmpHtTPVPz0VP1lpaWkh8b8D9DkJrPnppeO3PmTLcul9OZXSQIhV0kCIVdJAiFXSQIhV0kCIVdJAgNvdWBbdu2ufWTJ0+69YULF+bWUkNjo6Ojbj11ddrU/t4U2uHhYXffb7/91q0fO3bMre/evTu3du+997r7TkU6s4sEobCLBKGwiwShsIsEobCLBKGwiwShsIsEoXH2OvDqq6+69fnz57v1W265JbfmraIKANOm+X8CR44cceupy0F7K8ymLkOd6q2vr8+t79q1K7emcXYRmbIUdpEgFHaRIBR2kSAUdpEgFHaRIBR2kSA0zl4H9u3b59Y7OjrcujdnfGhoyN03NR+9qanJraeWVfbmrM+ePdvdN1VPXYo6Nd89mkJhJ3kIwDkAIwCGzcz/qxSRminHmf3fzexEGX6PiFSQXrOLBFE07Abg7yR3k9ww0R1IbiDZSbKzv7+/4MOJSKmKhn2lmf0YwMMAniH5kyvvYGabzazDzDra2toKPpyIlKpQ2M3saPa9D8A7AJaXoykRKb+Sw05yFsnm728DWANgT7kaE5HyKvJu/FwA72TLCU8D8KaZ/U9Zuppijh8/7tZT13b35qsD/rzu1Fh0asnm1Hz21Jx0b5w+9RmA1HXlU4+duuZ9NCWH3cwOArirjL2ISAVp6E0kCIVdJAiFXSQIhV0kCIVdJAhNca2Cnp4et566HHOKmeXWZs6c6e574oQ/hyk1vXbPHv+jFefPn8+tpabPpoYkvctUA+mhuWh0ZhcJQmEXCUJhFwlCYRcJQmEXCUJhFwlCYRcJQuPsVfDVV1+59cbGRrc+a9askh87m4KcKzX99sCBA279nnvuceveZbJvu+02d9/U9NvUks6a4no5ndlFglDYRYJQ2EWCUNhFglDYRYJQ2EWCUNhFgtA4exV8+eWXbj01n/27775z69687jNnzrj7Fl2l5/7773frXV1dubXUZwAuXbrk1lP7p8bpo9GZXSQIhV0kCIVdJAiFXSQIhV0kCIVdJAiFXSQIjbNXwf79+9367Nmz3frg4KBb9+bDHzt2zN33ySefdOspTz31lFt/5ZVXcmujo6OFHjt13fhUPZrkmZ3kFpJ9JPeM2zaH5A6SX2ffWyrbpogUNZmn8X8E8NAV254F8L6ZLQXwfvaziNSxZNjN7EMAp67Y/AiArdntrQAeLXNfIlJmpb5BN9fMvr94WQ+AuXl3JLmBZCfJzv7+/hIfTkSKKvxuvI2tKpi7sqCZbTazDjPrKDrpQkRKV2rYe0nOB4Dse1/5WhKRSig17O8BeCK7/QSAd8vTjohUSnKcneRbAFYBaCXZDeDXAF4E8FeSTwM4DOCxSjZ5rTt79qxbT62hnpq3PTQ0VFINADZt2uTWU+677z637vWeGmdPjZOnrguvcfbLJcNuZo/nlFaXuRcRqSB9XFYkCIVdJAiFXSQIhV0kCIVdJAhNca2C1NLCTU1Nbj019DYwMJBbmzdvnrvvkiVL3HpRra2tubXU0NucOXPc+smTJ926d1wi0pldJAiFXSQIhV0kCIVdJAiFXSQIhV0kCIVdJAiNs1fBzTff7NaHh4cL/f7z58/n1h566MprhVaXN86fmoLqjdEDwKlTV14a8XJFL1U91ejMLhKEwi4ShMIuEoTCLhKEwi4ShMIuEoTCLhKExtmrIDVf/fTp0249NQ7vLQn90ksvufumpMaqr7vOP18sXrw4t9bd3e3um1pBaGRkxK2nfn80OrOLBKGwiwShsIsEobCLBKGwiwShsIsEobCLBKFx9ipILS188eJFt+7NVwcAM8utLVu2zN03JTWWnRpnv+OOO3Jr33zzjbtvc3OzW+/v73frLS0tbj2a5Jmd5BaSfST3jNv2AsmjJLuyr7WVbVNEiprM0/g/Apjocie/NbO7s6/t5W1LRMotGXYz+xCAf/0fEal7Rd6g20jy0+xpfu6LI5IbSHaS7Ey9xhKRyik17L8H8CMAdwM4DiB3toWZbTazDjPrSE1sEJHKKSnsZtZrZiNmNgrgDwCWl7ctESm3ksJOcv64H38KYE/efUWkPiTH2Um+BWAVgFaS3QB+DWAVybsBGIBDAH5RwR6veXfeeadb37lzp1tPjcMvXbo0t5Zanz0lNY6esnZt/qjsyy+/7O574cIFt97T0+PWU+u7R5MMu5k9PsHm1yrQi4hUkD4uKxKEwi4ShMIuEoTCLhKEwi4ShKa4VsG6devc+uuvv+7Wp03z/zedPXs2t/bBBx+4+65Zs8ate9NnJ+P222/PrS1atMjdNzXsl+rt3Llzbj0andlFglDYRYJQ2EWCUNhFglDYRYJQ2EWCUNhFgtA4exU0NDS49cbGRreeupS09/vfeOMNd9/UOHtqjD+ltbU1t5aaonr48GG3njouM2bMcOvR6MwuEoTCLhKEwi4ShMIuEoTCLhKEwi4ShMIuEoTG2etAarx4YGDArXvjybt27Sqpp2pIXSJ79+7dbn1oaMitp45bNDqziwShsIsEobCLBKGwiwShsIsEobCLBKGwiwShcfY6sGLFCrf+5ptvunVvaeLp06eX1FM1tLe3u/XTp0+79UuXLrn1kZGRq21pSkue2UkuIvkPkp+T3Evyl9n2OSR3kPw6+95S+XZFpFSTeRo/DOBXZrYMwP0AniG5DMCzAN43s6UA3s9+FpE6lQy7mR03s4+z2+cAfAFgAYBHAGzN7rYVwKOValJEiruqN+hItgO4B8BOAHPN7HhW6gEwN2efDSQ7SXb29/cXaFVEiph02Ek2AdgGYJOZXbaSoI2tsDfhKntmttnMOsyso62trVCzIlK6SYWdZCPGgv5nM/tbtrmX5PysPh9AX2VaFJFySA69kSSA1wB8YWa/GVd6D8ATAF7Mvr9bkQ4D2Lhxo1t/++233bq3tPGZM2fcfQ8ePOjWlyxZ4taLaG5uduupJZdHR0fdekuLBojGm8w4+woAPwPwGcmubNtzGAv5X0k+DeAwgMcq06KIlEMy7Gb2TwDMKa8ubzsiUin6uKxIEAq7SBAKu0gQCrtIEAq7SBCa4loHFixY4NZvuukmt+5dinpwcNDdN3Wp6UqOs6em3w4PD7v11BTX1H97NDqziwShsIsEobCLBKGwiwShsIsEobCLBKGwiwShcfYqGLuQT76xSwbke/DBB936tm3bcmupsex33/UvQ7B+/Xq3XkRTU5NbP3bsmFtPHdfUfPdodGYXCUJhFwlCYRcJQmEXCUJhFwlCYRcJQmEXCULj7FWQGu9taGhw62vXrnXr3nXlZ86c6e7b3d3t1itp9uzZbj01Hz11XfhTp05ddU9Tmc7sIkEo7CJBKOwiQSjsIkEo7CJBKOwiQSjsIkFMZn32RQD+BGAuAAOw2cx+R/IFAD8H0J/d9Tkz216pRq9l3vrpk7Fy5Uq37l13PrU+e09Pj1v/5JNP3Ppdd93l1j033nijW79w4YJbb2xsdOup6+1HM5kP1QwD+JWZfUyyGcBukjuy2m/N7L8q156IlMtk1mc/DuB4dvscyS8A+EuYiEjduarnlyTbAdwDYGe2aSPJT0luITnhZxdJbiDZSbKzv79/oruISBVMOuwkmwBsA7DJzM4C+D2AHwG4G2Nn/pcm2s/MNptZh5l1tLW1laFlESnFpMJOshFjQf+zmf0NAMys18xGzGwUwB8ALK9cmyJSVDLsHLv06WsAvjCz34zbPn/c3X4KYE/52xORcpnMu/ErAPwMwGcku7JtzwF4nOTdGBuOOwTgFxXpcApIXSq6qFtvvTW31tXVlVsD0sNXO3bscOtFht7OnTvn1gcGBkr+3QDQ29tbaP+pZjLvxv8TwER/rRpTF7mG6BN0IkEo7CJBKOwiQSjsIkEo7CJBKOwiQehS0lPA888/n1ubN2+eu29qnP2BBx4oqafJWLdunVufO3euW09NYV29evVV9zSV6cwuEoTCLhKEwi4ShMIuEoTCLhKEwi4ShMIuEgTNrHoPRvYDODxuUyuAE1Vr4OrUa2/12heg3kpVzt5uM7MJr/9W1bD/4MHJTjPrqFkDjnrtrV77AtRbqarVm57GiwShsIsEUeuwb67x43vqtbd67QtQb6WqSm81fc0uItVT6zO7iFSJwi4SRE3CTvIhkvtI7if5bC16yEPyEMnPSHaR7KxxL1tI9pHcM27bHJI7SH6dfZ9wjb0a9fYCyaPZsesiubZGvS0i+Q+Sn5PcS/KX2faaHjunr6oct6q/ZifZAOArAA8C6AbwEYDHzezzqjaSg+QhAB1mVvMPYJD8CYDzAP5kZv+WbftPAKfM7MXsH8oWM/uPOuntBQDna72Md7Za0fzxy4wDeBTAk6jhsXP6egxVOG61OLMvB7DfzA6a2SCAvwB4pAZ91D0z+xDAqSs2PwJga3Z7K8b+WKoup7e6YGbHzezj7PY5AN8vM17TY+f0VRW1CPsCAEfG/dyN+lrv3QD8neRukhtq3cwE5prZ8ex2DwD/2k3Vl1zGu5quWGa8bo5dKcufF6U36H5opZn9GMDDAJ7Jnq7WJRt7DVZPY6eTWsa7WiZYZvxfannsSl3+vKhahP0ogEXjfl6YbasLZnY0+94H4B3U31LUvd+voJt976txP/9ST8t4T7TMOOrg2NVy+fNahP0jAEtJLiY5HcB6AO/VoI8fIDkre+MEJGcBWIP6W4r6PQBPZLefAPBuDXu5TL0s4523zDhqfOxqvvy5mVX9C8BajL0jfwDA87XoIaevJQA+yb721ro3AG9h7GndEMbe23gawM0A3gfwNYD/AzCnjnp7A8BnAD7FWLDm16i3lRh7iv4pgK7sa22tj53TV1WOmz4uKxKE3qATCUJhFwlCYRcJQmEXCUJhFwlCYRcJQmEXCeL/ASFkGZTWLx26AAAAAElFTkSuQmCC\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[0. 1. 0. 0. 0. 0. 0. 0. 0. 0.]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "La clase del elemento 6 es la clase 1 (pantalones)."
      ],
      "metadata": {
        "id": "TxjXrHj3nd2e"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "predictions = model7.predict(test_data)\n",
        "print('Predicción del elemento 6: clase', np.argmax(predictions[5]))\n",
        "print(predictions[5])\n",
        "print('Suma de las predicciones del elemento 6:', np.sum(predictions[5]))"
      ],
      "metadata": {
        "id": "xRRBr0kmnd2e",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "37478890-a2c8-41c9-850e-dac942004abe"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 1s 3ms/step\n",
            "Predicción del elemento 6: clase 1\n",
            "[6.3703999e-02 7.4151599e-01 1.9667026e-02 9.5675565e-02 4.3927837e-02\n",
            " 9.6384028e-04 3.2216404e-02 7.6084677e-04 1.1589121e-03 4.0958388e-04]\n",
            "Suma de las predicciones del elemento 6: 1.0\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "def plot_image(i, predictions_array, true_label, img):\n",
        "  predictions_array, true_label, img = predictions_array, true_label[i], img[i]\n",
        "  plt.grid(False)\n",
        "  plt.xticks([])\n",
        "  plt.yticks([])\n",
        "\n",
        "  plt.imshow(img, cmap = plt.cm.binary)    \n",
        "\n",
        "  predicted_label = np.argmax(predictions_array)\n",
        "  if predicted_label.all() == true_label.all():\n",
        "    color = 'blue'\n",
        "  else:\n",
        "    color = 'red'\n",
        "\n",
        "  plt.xlabel('{} {:2.0f}% ({})'.format(predicted_label,\n",
        "                                100*np.max(predictions_array),\n",
        "                                true_label),\n",
        "                                color = color)\n",
        "\n",
        "def plot_value_array(i, predictions_array, true_label):\n",
        "  predictions_array, true_label = predictions_array, true_label[i]\n",
        "  plt.grid(False)\n",
        "  plt.xticks(range(10))\n",
        "  plt.yticks([])\n",
        "  thisplot = plt.bar(range(10), predictions_array, color = '#00FF00')\n",
        "  plt.ylim([0, 1])\n",
        "  predicted_label = np.argmax(predictions_array)\n",
        "\n",
        "  thisplot[predicted_label].set_color('red')\n",
        "  thisplot[np.argmax(true_label)].set_color('black')"
      ],
      "metadata": {
        "id": "GZpRjWQhnd2e"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "plt.figure(figsize = (6,3))\n",
        "plt.subplot(1,2,1)\n",
        "plot_image(5, predictions[5], test_labels, test_data.reshape(10000,28,28))\n",
        "plt.subplot(1,2,2)\n",
        "plot_value_array(5, predictions[5],  test_labels)\n",
        "plt.show()"
      ],
      "metadata": {
        "id": "CHDP1XAvnd2e",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 211
        },
        "outputId": "6d9c85fe-805b-4d03-d9fa-9697bf866c41"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x216 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Caso 7.5"
      ],
      "metadata": {
        "id": "aySiC2fyrSZw"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Crear la red de neuronas"
      ],
      "metadata": {
        "id": "XrPy96d5rSZx"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Se crea la red neuronal con una capa de \n",
        "entrada con 512 neuronas y función de activación ```sigmoid```, y una capa de salida con función de activación ```softmax```."
      ],
      "metadata": {
        "id": "N3PbHHwQrSZx"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model7b = models.Sequential()\n",
        "model7b.add(layers.Dense(512, activation ='sigmoid', name = 'Capa7', input_shape = (28*28,)))\n",
        "model7b.add(layers.Dense(10, name = 'CapaSalida7', activation = 'softmax'))"
      ],
      "metadata": {
        "id": "Op6OP_2QrSZx"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Explicar model.summary()\n"
      ],
      "metadata": {
        "id": "qHjUm1LyrSZx"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "El resumen del modelo indica el número de parámetros de salida de cada capa. Este número es igual al tamaño de los datos de entrada multiplicado por el número de neuronas, y sumando el número de umbrales de activación de la capa.\n",
        "\n",
        "\n",
        "En el caso de la primera capa: 401920 = 784 (tamaño de los datos) x 512 (número de neuronas) + 512 (número de umbrales de activación).\n",
        "\n",
        "En el caso de la segunda capa: 262656 = 512 (tamaño de los datos) x 512 (número de neuronas) + 512 (número de umbrales de activación)."
      ],
      "metadata": {
        "id": "i2fzTOHmrSZx"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model7b.summary()"
      ],
      "metadata": {
        "id": "LkL7pseyrSZy",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "44a7a35b-e1b7-438e-e216-39c08e3e8674"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential_15\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " Capa7 (Dense)               (None, 512)               401920    \n",
            "                                                                 \n",
            " CapaSalida7 (Dense)         (None, 10)                5130      \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 407,050\n",
            "Trainable params: 407,050\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Se definen las funciones del optimizador, la pérdida y la métrica."
      ],
      "metadata": {
        "id": "aNFO8YwZrSZy"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model7b.compile(optimizer = 'sgd',\n",
        "              loss = 'categorical_crossentropy',\n",
        "              metrics = ['accuracy'])"
      ],
      "metadata": {
        "id": "HlHwlExZrSZy"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Entrenar la red"
      ],
      "metadata": {
        "id": "8QNafZwhrSZy"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Para entrenar la red se realizan ```5``` ciclos con lotes de ```128``` imágenes. Se compara con los datos de validación para controlar el sobreentrenamiento. "
      ],
      "metadata": {
        "id": "qxca_mFtrSZy"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "history = model7b.fit(train_val_data, train_val_labels,\n",
        "                    epochs = 25,\n",
        "                    batch_size = 128,\n",
        "                    validation_data = (val_data, val_labels))"
      ],
      "metadata": {
        "id": "gCgTPpHPrSZy",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "0a83208e-3bc7-43b8-f032-9379ae68eb35"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/25\n",
            "391/391 [==============================] - 4s 10ms/step - loss: 1.7895 - accuracy: 0.5524 - val_loss: 1.4010 - val_accuracy: 0.6814\n",
            "Epoch 2/25\n",
            "391/391 [==============================] - 4s 9ms/step - loss: 1.2182 - accuracy: 0.6993 - val_loss: 1.0709 - val_accuracy: 0.7228\n",
            "Epoch 3/25\n",
            "391/391 [==============================] - 3s 9ms/step - loss: 0.9917 - accuracy: 0.7235 - val_loss: 0.9192 - val_accuracy: 0.7334\n",
            "Epoch 4/25\n",
            "391/391 [==============================] - 3s 9ms/step - loss: 0.8758 - accuracy: 0.7369 - val_loss: 0.8326 - val_accuracy: 0.7460\n",
            "Epoch 5/25\n",
            "391/391 [==============================] - 3s 9ms/step - loss: 0.8055 - accuracy: 0.7467 - val_loss: 0.7760 - val_accuracy: 0.7542\n",
            "Epoch 6/25\n",
            "391/391 [==============================] - 3s 9ms/step - loss: 0.7575 - accuracy: 0.7530 - val_loss: 0.7410 - val_accuracy: 0.7573\n",
            "Epoch 7/25\n",
            "391/391 [==============================] - 3s 9ms/step - loss: 0.7224 - accuracy: 0.7598 - val_loss: 0.7076 - val_accuracy: 0.7658\n",
            "Epoch 8/25\n",
            "391/391 [==============================] - 4s 9ms/step - loss: 0.6954 - accuracy: 0.7669 - val_loss: 0.6860 - val_accuracy: 0.7670\n",
            "Epoch 9/25\n",
            "391/391 [==============================] - 4s 9ms/step - loss: 0.6734 - accuracy: 0.7726 - val_loss: 0.6650 - val_accuracy: 0.7759\n",
            "Epoch 10/25\n",
            "391/391 [==============================] - 4s 10ms/step - loss: 0.6547 - accuracy: 0.7762 - val_loss: 0.6471 - val_accuracy: 0.7830\n",
            "Epoch 11/25\n",
            "391/391 [==============================] - 4s 9ms/step - loss: 0.6392 - accuracy: 0.7805 - val_loss: 0.6318 - val_accuracy: 0.7878\n",
            "Epoch 12/25\n",
            "391/391 [==============================] - 4s 9ms/step - loss: 0.6254 - accuracy: 0.7845 - val_loss: 0.6198 - val_accuracy: 0.7905\n",
            "Epoch 13/25\n",
            "391/391 [==============================] - 4s 9ms/step - loss: 0.6129 - accuracy: 0.7889 - val_loss: 0.6086 - val_accuracy: 0.7941\n",
            "Epoch 14/25\n",
            "391/391 [==============================] - 4s 9ms/step - loss: 0.6019 - accuracy: 0.7913 - val_loss: 0.5979 - val_accuracy: 0.7996\n",
            "Epoch 15/25\n",
            "391/391 [==============================] - 3s 9ms/step - loss: 0.5923 - accuracy: 0.7960 - val_loss: 0.5890 - val_accuracy: 0.8007\n",
            "Epoch 16/25\n",
            "391/391 [==============================] - 4s 9ms/step - loss: 0.5835 - accuracy: 0.7980 - val_loss: 0.5803 - val_accuracy: 0.8014\n",
            "Epoch 17/25\n",
            "391/391 [==============================] - 3s 9ms/step - loss: 0.5753 - accuracy: 0.8013 - val_loss: 0.5749 - val_accuracy: 0.8053\n",
            "Epoch 18/25\n",
            "391/391 [==============================] - 4s 10ms/step - loss: 0.5677 - accuracy: 0.8032 - val_loss: 0.5667 - val_accuracy: 0.8055\n",
            "Epoch 19/25\n",
            "391/391 [==============================] - 4s 9ms/step - loss: 0.5608 - accuracy: 0.8063 - val_loss: 0.5585 - val_accuracy: 0.8102\n",
            "Epoch 20/25\n",
            "391/391 [==============================] - 3s 9ms/step - loss: 0.5546 - accuracy: 0.8081 - val_loss: 0.5538 - val_accuracy: 0.8104\n",
            "Epoch 21/25\n",
            "391/391 [==============================] - 4s 9ms/step - loss: 0.5488 - accuracy: 0.8105 - val_loss: 0.5495 - val_accuracy: 0.8104\n",
            "Epoch 22/25\n",
            "391/391 [==============================] - 3s 9ms/step - loss: 0.5434 - accuracy: 0.8127 - val_loss: 0.5440 - val_accuracy: 0.8148\n",
            "Epoch 23/25\n",
            "391/391 [==============================] - 4s 9ms/step - loss: 0.5382 - accuracy: 0.8137 - val_loss: 0.5382 - val_accuracy: 0.8151\n",
            "Epoch 24/25\n",
            "391/391 [==============================] - 4s 9ms/step - loss: 0.5334 - accuracy: 0.8168 - val_loss: 0.5319 - val_accuracy: 0.8167\n",
            "Epoch 25/25\n",
            "391/391 [==============================] - 4s 9ms/step - loss: 0.5290 - accuracy: 0.8171 - val_loss: 0.5280 - val_accuracy: 0.8192\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "La pérdida indica el porcentaje de errores entre las etiquetas y las predicciones del modelo.\n",
        "\n",
        "La precisión indica el porcentaje de aciertos entre las predicciones y las etiquetas."
      ],
      "metadata": {
        "id": "epQ6PWfirSZy"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "test_loss, test_acc = model7b.evaluate(test_data, test_labels)"
      ],
      "metadata": {
        "id": "dpY3zvZwrSZz",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "7ced56ff-da62-4dfc-9739-dc6990a64f2e"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 1s 3ms/step - loss: 0.5496 - accuracy: 0.8060\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "acc = history.history['accuracy']\n",
        "val_acc = history.history['val_accuracy']\n",
        "loss = history.history['loss']\n",
        "val_loss = history.history['val_loss']\n",
        "\n",
        "epochs = range(1, len(acc) + 1)\n",
        "\n",
        "# \"bo\" es para \"punto azun\"\n",
        "plt.plot(epochs, loss, 'bo', label='Pérdida de Entrenamiento')\n",
        "# \"b\" es para \"linea continua azul\"\n",
        "plt.plot(epochs, val_loss, 'b', label='Pérdida de Validación')\n",
        "plt.title('Pérdida de Entrenamiento y Validación')\n",
        "plt.xlabel('Epochs')\n",
        "plt.ylabel('Loss')\n",
        "plt.legend()\n",
        "\n",
        "plt.show()"
      ],
      "metadata": {
        "id": "wiO20x5XzQcX",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "outputId": "6d691c61-7949-47e2-9283-e1635eea76cf"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Predecir valores"
      ],
      "metadata": {
        "id": "1ZARjkoqrSZz"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Imprimir la matriz de confusión, que muestra los valores predichos frente a los reales."
      ],
      "metadata": {
        "id": "BvcHg4ErrSZz"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Note, this code is taken straight from the SKLEARN website, an nice way of \n",
        "# viewing confusion matrix.\n",
        "def plot_confusion_matrix(cm, classes,\n",
        "                          normalize = False,\n",
        "                          title = 'Confusion matrix',\n",
        "                          cmap = plt.cm.Blues):\n",
        "    '''\n",
        "    This function prints and plots the confusion matrix.\n",
        "    Normalization can be applied by setting `normalize = True`.\n",
        "    '''\n",
        "    plt.imshow(cm, interpolation = 'nearest', cmap = cmap)\n",
        "    plt.title(title)\n",
        "    plt.colorbar()\n",
        "    tick_marks = np.arange(len(classes))\n",
        "    plt.xticks(tick_marks, classes, rotation = 45)\n",
        "    plt.yticks(tick_marks, classes)\n",
        "\n",
        "    if normalize:\n",
        "        cm = cm.astype('float') / cm.sum(axis = 1)[:, np.newaxis]\n",
        "\n",
        "    thresh = cm.max() / 2.\n",
        "    for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):\n",
        "        plt.text(j, i, cm[i, j],\n",
        "                 horizontalalignment = 'center',\n",
        "                 color = 'white' if cm[i, j] > thresh else 'black')\n",
        "\n",
        "    plt.tight_layout()\n",
        "    plt.ylabel('Observación')\n",
        "    plt.xlabel('Predicción')"
      ],
      "metadata": {
        "id": "vu9XYDdurSZz"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "Y_pred = model7b.predict(test_data)\n",
        "print('Valores de predicción del modelo')\n",
        "# Convertir en predicciones de clases desde los vectores one-hot encoding \n",
        "Y_pred_classes = np.argmax(Y_pred, axis = 1) \n",
        "# Convertir en clases de observación desde los vectores one-hot encoding\n",
        "Y_true = np.argmax(test_labels, axis = 1) \n",
        "# Computar la matriz de confusión\n",
        "confusion_mtx = confusion_matrix(Y_true, Y_pred_classes) \n",
        "# Dibujar la matriz de confusión\n",
        "print('Matriz de confusión')\n",
        "plot_confusion_matrix(confusion_mtx, classes = range(10))"
      ],
      "metadata": {
        "id": "6SXfWWibrSZz",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 366
        },
        "outputId": "324e60f6-479d-456d-e4dd-fa02ae19f5e2"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 1s 3ms/step\n",
            "Valores de predicción del modelo\n",
            "Matriz de confusión\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Se grafica el elemento 6 (```test_data[5]```) para comprobar las soluciones. "
      ],
      "metadata": {
        "id": "OlCm0q0KrSZz"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "digit = test_data.reshape(10000,28, 28)[5]\n",
        "plt.imshow(digit, cmap = plt.cm.binary)\n",
        "plt.show()\n",
        "print(test_labels[5])"
      ],
      "metadata": {
        "id": "ds7qBYgBrSZz",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 283
        },
        "outputId": "54fa7a13-1492-4d6e-8789-cf76a18d92a3"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[0. 1. 0. 0. 0. 0. 0. 0. 0. 0.]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "La clase del elemento 6 es la clase 1 (pantalones)."
      ],
      "metadata": {
        "id": "MBy-CwhVrSZ0"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "predictions = model7b.predict(test_data)\n",
        "print('Predicción del elemento 6: clase', np.argmax(predictions[5]))\n",
        "print(predictions[5])\n",
        "print('Suma de las predicciones del elemento 6:', np.sum(predictions[5]))"
      ],
      "metadata": {
        "id": "zINTkEmCrSZ0",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "9b3e8085-dda7-49e4-a077-c2e0cfcb8f01"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 1s 3ms/step\n",
            "Predicción del elemento 6: clase 1\n",
            "[6.2911315e-03 9.7520256e-01 1.3679471e-03 4.5774151e-03 1.1252260e-02\n",
            " 3.1324223e-06 1.2764327e-03 8.0710261e-06 1.9323830e-05 1.6630870e-06]\n",
            "Suma de las predicciones del elemento 6: 1.0\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "def plot_image(i, predictions_array, true_label, img):\n",
        "  predictions_array, true_label, img = predictions_array, true_label[i], img[i]\n",
        "  plt.grid(False)\n",
        "  plt.xticks([])\n",
        "  plt.yticks([])\n",
        "\n",
        "  plt.imshow(img, cmap = plt.cm.binary)    \n",
        "\n",
        "  predicted_label = np.argmax(predictions_array)\n",
        "  if predicted_label.all() == true_label.all():\n",
        "    color = 'blue'\n",
        "  else:\n",
        "    color = 'red'\n",
        "\n",
        "  plt.xlabel('{} {:2.0f}% ({})'.format(predicted_label,\n",
        "                                100*np.max(predictions_array),\n",
        "                                true_label),\n",
        "                                color = color)\n",
        "\n",
        "def plot_value_array(i, predictions_array, true_label):\n",
        "  predictions_array, true_label = predictions_array, true_label[i]\n",
        "  plt.grid(False)\n",
        "  plt.xticks(range(10))\n",
        "  plt.yticks([])\n",
        "  thisplot = plt.bar(range(10), predictions_array, color = '#00FF00')\n",
        "  plt.ylim([0, 1])\n",
        "  predicted_label = np.argmax(predictions_array)\n",
        "\n",
        "  thisplot[predicted_label].set_color('red')\n",
        "  thisplot[np.argmax(true_label)].set_color('black')"
      ],
      "metadata": {
        "id": "KqT4u7_jrSZ0"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "plt.figure(figsize = (6,3))\n",
        "plt.subplot(1,2,1)\n",
        "plot_image(5, predictions[5], test_labels, test_data.reshape(10000,28,28))\n",
        "plt.subplot(1,2,2)\n",
        "plot_value_array(5, predictions[5],  test_labels)\n",
        "plt.show()"
      ],
      "metadata": {
        "id": "ReDfVSMrrSZ0",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 211
        },
        "outputId": "44e49c3b-54c7-4d83-bfaf-6de003c89036"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x216 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Caso 8"
      ],
      "metadata": {
        "id": "Bhw1lF8And2e"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Crear la red de neuronas"
      ],
      "metadata": {
        "id": "5e-tHDPHnd2e"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Se crea la red neuronal con una capa de entrada con 512 neuronas y función de activación ```sigmoid```, y una capa de salida con función de activación ```softmax```."
      ],
      "metadata": {
        "id": "Tz5qH7g-nd2e"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model8 = models.Sequential()\n",
        "model8.add(layers.Dense(512, activation ='sigmoid', name = 'Capa8', input_shape = (28*28,)))\n",
        "model8.add(layers.Dense(10, name = 'CapaSalida8', activation = 'softmax'))"
      ],
      "metadata": {
        "id": "sHE5QeQEnd2f"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Explicar model.summary()\n"
      ],
      "metadata": {
        "id": "zPOSVYtNnd2f"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "El resumen del modelo indica el número de parámetros de salida de cada capa. Este número es igual al tamaño de los datos de entrada multiplicado por el número de neuronas, y sumando el número de umbrales de activación de la capa.\n",
        "\n",
        "\n",
        "En el caso de la primera capa: 401920 = 784 (tamaño de los datos) x 512 (número de neuronas) + 512 (número de umbrales de activación).\n",
        "\n",
        "En el caso de la segunda capa: 262656 = 512 (tamaño de los datos) x 512 (número de neuronas) + 512 (número de umbrales de activación)."
      ],
      "metadata": {
        "id": "wYb5bK-Lnd2f"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model8.summary()"
      ],
      "metadata": {
        "id": "SOyBZTzvnd2f",
        "outputId": "76d41e98-3d4c-4892-b97d-f085440e5fa4",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential_16\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " Capa8 (Dense)               (None, 512)               401920    \n",
            "                                                                 \n",
            " CapaSalida8 (Dense)         (None, 10)                5130      \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 407,050\n",
            "Trainable params: 407,050\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Se definen las funciones del optimizador, la pérdida y la métrica."
      ],
      "metadata": {
        "id": "8aOOKQMsnd2f"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model8.compile(optimizer = 'rmsprop',\n",
        "              loss = 'categorical_crossentropy',\n",
        "              metrics = ['accuracy'])"
      ],
      "metadata": {
        "id": "RmNUdhy4nd2f"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Entrenar la red"
      ],
      "metadata": {
        "id": "OhPvgbLCnd2f"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Para entrenar la red se realizan ```5``` ciclos con lotes de ```128``` imágenes. Se compara con los datos de validación para controlar el sobreentrenamiento. "
      ],
      "metadata": {
        "id": "AgIMPiOdnd2f"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "history = model8.fit(train_val_data, train_val_labels,\n",
        "                    epochs = 5,\n",
        "                    batch_size = 524,\n",
        "                    validation_data = (val_data, val_labels))"
      ],
      "metadata": {
        "id": "JWlfDykcnd2f",
        "outputId": "8cd2cffb-27e0-47f2-ed8e-ede5f7da6e3b",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/5\n",
            "96/96 [==============================] - 4s 39ms/step - loss: 0.3244 - accuracy: 0.8827 - val_loss: 0.3717 - val_accuracy: 0.8616\n",
            "Epoch 2/5\n",
            "96/96 [==============================] - 3s 29ms/step - loss: 0.3196 - accuracy: 0.8827 - val_loss: 0.3639 - val_accuracy: 0.8667\n",
            "Epoch 3/5\n",
            "96/96 [==============================] - 3s 32ms/step - loss: 0.3141 - accuracy: 0.8848 - val_loss: 0.3401 - val_accuracy: 0.8751\n",
            "Epoch 4/5\n",
            "96/96 [==============================] - 3s 28ms/step - loss: 0.3070 - accuracy: 0.8878 - val_loss: 0.3303 - val_accuracy: 0.8783\n",
            "Epoch 5/5\n",
            "96/96 [==============================] - 3s 28ms/step - loss: 0.3012 - accuracy: 0.8891 - val_loss: 0.3281 - val_accuracy: 0.8819\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Al aumentar el batch_size a 524 la precisión del modelo aumenta.\n",
        "\n",
        "La pérdida indica el porcentaje de errores entre las etiquetas y las predicciones del modelo.\n",
        "\n",
        "La precisión indica el porcentaje de aciertos entre las predicciones y las etiquetas."
      ],
      "metadata": {
        "id": "-AkE1HWxnd2f"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "test_loss, test_acc = model8.evaluate(test_data, test_labels)"
      ],
      "metadata": {
        "id": "-92kjkjwnd2f",
        "outputId": "b36ff26d-d9b3-4435-80a0-184f823bdebe",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 1s 3ms/step - loss: 0.3565 - accuracy: 0.8704\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "acc = history.history['accuracy']\n",
        "val_acc = history.history['val_accuracy']\n",
        "loss = history.history['loss']\n",
        "val_loss = history.history['val_loss']\n",
        "\n",
        "epochs = range(1, len(acc) + 1)\n",
        "\n",
        "# \"bo\" es para \"punto azun\"\n",
        "plt.plot(epochs, loss, 'bo', label='Pérdida de Entrenamiento')\n",
        "# \"b\" es para \"linea continua azul\"\n",
        "plt.plot(epochs, val_loss, 'b', label='Pérdida de Validación')\n",
        "plt.title('Pérdida de Entrenamiento y Validación')\n",
        "plt.xlabel('Epochs')\n",
        "plt.ylabel('Loss')\n",
        "plt.legend()\n",
        "\n",
        "plt.show()"
      ],
      "metadata": {
        "id": "9gK6wy94zStn",
        "outputId": "b2bf09f2-8f20-4225-d3ec-4097720a3f44",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Predecir valores"
      ],
      "metadata": {
        "id": "iwUEfXINnd2g"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Imprimir la matriz de confusión, que muestra los valores predichos frente a los reales."
      ],
      "metadata": {
        "id": "DZSsk0eOnd2g"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Note, this code is taken straight from the SKLEARN website, an nice way of \n",
        "# viewing confusion matrix.\n",
        "def plot_confusion_matrix(cm, classes,\n",
        "                          normalize = False,\n",
        "                          title = 'Confusion matrix',\n",
        "                          cmap = plt.cm.Blues):\n",
        "    '''\n",
        "    This function prints and plots the confusion matrix.\n",
        "    Normalization can be applied by setting `normalize = True`.\n",
        "    '''\n",
        "    plt.imshow(cm, interpolation = 'nearest', cmap = cmap)\n",
        "    plt.title(title)\n",
        "    plt.colorbar()\n",
        "    tick_marks = np.arange(len(classes))\n",
        "    plt.xticks(tick_marks, classes, rotation = 45)\n",
        "    plt.yticks(tick_marks, classes)\n",
        "\n",
        "    if normalize:\n",
        "        cm = cm.astype('float') / cm.sum(axis = 1)[:, np.newaxis]\n",
        "\n",
        "    thresh = cm.max() / 2.\n",
        "    for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):\n",
        "        plt.text(j, i, cm[i, j],\n",
        "                 horizontalalignment = 'center',\n",
        "                 color = 'white' if cm[i, j] > thresh else 'black')\n",
        "\n",
        "    plt.tight_layout()\n",
        "    plt.ylabel('Observación')\n",
        "    plt.xlabel('Predicción')"
      ],
      "metadata": {
        "id": "gSHAF0dsnd2g"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "Y_pred = model8.predict(test_data)\n",
        "print('Valores de predicción del modelo')\n",
        "# Convertir en predicciones de clases desde los vectores one-hot encoding \n",
        "Y_pred_classes = np.argmax(Y_pred, axis = 1) \n",
        "# Convertir en clases de observación desde los vectores one-hot encoding\n",
        "Y_true = np.argmax(test_labels, axis = 1) \n",
        "# Computar la matriz de confusión\n",
        "confusion_mtx = confusion_matrix(Y_true, Y_pred_classes) \n",
        "# Dibujar la matriz de confusión\n",
        "print('Matriz de confusión')\n",
        "plot_confusion_matrix(confusion_mtx, classes = range(10))"
      ],
      "metadata": {
        "id": "bO6Ll8gNnd2g",
        "outputId": "1f1d0860-e89e-4abf-88d6-37e4e2daed83",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 366
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 1s 3ms/step\n",
            "Valores de predicción del modelo\n",
            "Matriz de confusión\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Se grafica el elemento 6 (```test_data[5]```) para comprobar las soluciones. "
      ],
      "metadata": {
        "id": "2BxXnxM3nd2g"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "digit = test_data.reshape(10000,28, 28)[5]\n",
        "plt.imshow(digit, cmap = plt.cm.binary)\n",
        "plt.show()\n",
        "print(test_labels[5])"
      ],
      "metadata": {
        "id": "VqJasvFind2g",
        "outputId": "459504d7-51c3-4b98-f7e1-5bdea40875ba",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 283
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[0. 1. 0. 0. 0. 0. 0. 0. 0. 0.]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "La clase del elemento 6 es la clase 1 (pantalones)."
      ],
      "metadata": {
        "id": "xf5zMsPsnd2g"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "predictions = model8.predict(test_data)\n",
        "print('Predicción del elemento 6: clase', np.argmax(predictions[5]))\n",
        "print(predictions[5])\n",
        "print('Suma de las predicciones del elemento 6:', np.sum(predictions[5]))"
      ],
      "metadata": {
        "id": "SR6O7nUZnd2g",
        "outputId": "4833bee1-b87a-4825-dae3-5f02b3c7c95d",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 1s 3ms/step\n",
            "Predicción del elemento 6: clase 1\n",
            "[9.5273199e-04 9.9876368e-01 8.8163273e-05 6.8759327e-05 1.0342503e-04\n",
            " 3.9242907e-09 2.2170978e-05 2.6210438e-07 7.8846108e-07 3.1833867e-09]\n",
            "Suma de las predicciones del elemento 6: 1.0\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "def plot_image(i, predictions_array, true_label, img):\n",
        "  predictions_array, true_label, img = predictions_array, true_label[i], img[i]\n",
        "  plt.grid(False)\n",
        "  plt.xticks([])\n",
        "  plt.yticks([])\n",
        "\n",
        "  plt.imshow(img, cmap = plt.cm.binary)    \n",
        "\n",
        "  predicted_label = np.argmax(predictions_array)\n",
        "  if predicted_label.all() == true_label.all():\n",
        "    color = 'blue'\n",
        "  else:\n",
        "    color = 'red'\n",
        "\n",
        "  plt.xlabel('{} {:2.0f}% ({})'.format(predicted_label,\n",
        "                                100*np.max(predictions_array),\n",
        "                                true_label),\n",
        "                                color = color)\n",
        "\n",
        "def plot_value_array(i, predictions_array, true_label):\n",
        "  predictions_array, true_label = predictions_array, true_label[i]\n",
        "  plt.grid(False)\n",
        "  plt.xticks(range(10))\n",
        "  plt.yticks([])\n",
        "  thisplot = plt.bar(range(10), predictions_array, color = '#00FF00')\n",
        "  plt.ylim([0, 1])\n",
        "  predicted_label = np.argmax(predictions_array)\n",
        "\n",
        "  thisplot[predicted_label].set_color('red')\n",
        "  thisplot[np.argmax(true_label)].set_color('black')"
      ],
      "metadata": {
        "id": "V-WgItlYnd2g"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "plt.figure(figsize = (6,3))\n",
        "plt.subplot(1,2,1)\n",
        "plot_image(5, predictions[5], test_labels, test_data.reshape(10000,28,28))\n",
        "plt.subplot(1,2,2)\n",
        "plot_value_array(5, predictions[5],  test_labels)\n",
        "plt.show()"
      ],
      "metadata": {
        "id": "_hU9WOQfnd2g",
        "outputId": "967903cb-781b-4fc7-af1a-1881c8aecea9",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 211
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x216 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Conclusiones\n"
      ],
      "metadata": {
        "id": "x15DhcxgTvb8"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "En la siguiente tabla observamos los distintos casos con su valor de pérdida y su precisión. \n",
        "\n",
        "> ![MicrosoftTeams-image (1).png](data:image/png;base64,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)\n",
        "\n",
        "\n",
        "\n",
        "*   En los casos 1 y 2 ambos tienen capas de 10 neuronas y función de activación \n",
        "'relu' , lo que cambia es el optimizador. En el caso 1, el optimizador es el 'sgd' y observamos que tiene una precisión de 0.77 y el del caso 2, el 'rmsprop' con una precisión de 0.83. Por tanto podemos concluir que es más preciso el 'sgd'.\n",
        "*   En los casos 3 y 4 ambos tienen capas de 10 neuronas y función de activación 'sigmoid' , lo que cambia es el optimizador. En el caso 3, el optimizador es el 'sgd' y observamos que tiene una precisión de 0.64 y el del caso 4, el 'rmsprop' con una precisión de 0.83. Por tanto podemos concluir que es más preciso el optimizador 'sgd'.\n",
        "*  Comparando la función de activación'relu' y 'sigmoid' de los casos 1, 2, 3 y 4. La función de activación 'relu'(acc 0.77) tiene mayor precisión que 'sigmoid' (acc 0.64)\n",
        "\n",
        "\n",
        "*   En los casos 5 y 6 ambos tienen capas de 512 neuronas y función de activación 'relu' , lo que cambia es el optimizador. En el caso 5, el optimizador es el 'sgd' y observamos que tiene una precisión de 0.82 y el del caso 6, el 'rmsprop' con una precisión 0.89. Por tanto podemos concluir que es más preciso el optimizador 'rmsprop'.\n",
        "*   En los casos 7 y 8 ambos tienen capas de 512 neuronas y función de activación 'sigmoid' , lo que cambia es el optimizador. En el caso 7, el optimizador es el 'sgd' y observamos que tiene una precisión de 0.74 y el del caso 8, el 'rmsprop' con una precisión 0.81. Por tanto podemos concluir que es más preciso el optimizador 'rmsprop'.\n",
        "\n",
        "*   Comparando la función de activación'relu' y 'sigmoid' de los casos 5, 6, 7 y 8. La función de activación 'relu' tiene mayor precisión que 'sigmoid'.\n",
        "\n",
        "*  En los casos 2.5, 7.5 añadimos 25 épocas lo que nos lleva al sobreentrenamiento de nuestro modelo.\n",
        "\n",
        "*   Por último, observamos que los modelos con 10 neuronas son menos precisas que los modelos con 512 neuronas.\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "\n"
      ],
      "metadata": {
        "id": "GH0mlY6AT_lW"
      }
    }
  ]
}