Redes CNN_Ejercicio ClaseACTUALIZADO.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU",
    "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/Redes%20CNN_Ejercicio%20ClaseACTUALIZADO.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 1,
      "metadata": {
        "id": "XpNGzfU8ACaQ",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "8b4ceebe-f635-442a-ee12-5da066ddcb4c"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Colab only includes TensorFlow 2.x; %tensorflow_version has no effect.\n",
            "2.9.2\n"
          ]
        }
      ],
      "source": [
        "%tensorflow_version 2.x\n",
        "import tensorflow as tf\n",
        "from tensorflow import keras\n",
        "\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "\n",
        "print(tf.__version__)#\n",
        "\n",
        "from tensorflow.keras.models import Sequential\n",
        "from tensorflow.keras.layers import Dense, Flatten, Conv2D, MaxPooling2D\n",
        "from tensorflow.keras.losses import sparse_categorical_crossentropy\n",
        "from tensorflow.keras.optimizers import Adam\n",
        "import matplotlib.pyplot as plt"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Configuración del modelo\n",
        "batch_size = 50\n",
        "img_width, img_height, img_num_channels =  60000, 32, 32  # Forma de la imagen\n",
        "loss_function = sparse_categorical_crossentropy\n",
        "no_classes = 100\n",
        "no_epochs = 100\n",
        "optimizer = Adam()\n",
        "validation_split = 0.2\n",
        "verbosity = 1\n",
        "input_shape = (60000, 32, 32) #img_width, img_height, img_num_channels"
      ],
      "metadata": {
        "id": "cvrXA-JaAF8h"
      },
      "execution_count": 2,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "(train_images, train_labels), (test_images, test_labels) = tf.keras.datasets.cifar100.load_data(label_mode=\"fine\")\n",
        "# CIFAR: https://www.cs.toronto.edu/~kriz/cifar.html\n",
        "assert train_images.shape == (50000, 32, 32, 3)  # Forma tensor juego imagenes de entrenamiento\n",
        "assert test_images.shape == (10000, 32, 32, 3)  # Forma tensor juego imagenes de test\n",
        "assert train_labels.shape == (50000,1)  # Forma tensor etiquetas de entrenamiento\n",
        "assert test_labels.shape == (10000,1)  # Forma tensor etiquetas de test"
      ],
      "metadata": {
        "id": "RYgWQWuZAPaH",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "99671bef-0e97-451f-9da2-f75af6b69e86"
      },
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Downloading data from https://www.cs.toronto.edu/~kriz/cifar-100-python.tar.gz\n",
            "169001437/169001437 [==============================] - 14s 0us/step\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Pintamos la tercera imagen de entrenamiento (la número 2 en el array)\n",
        "import matplotlib.pyplot as plt\n",
        "digit = train_images[2]\n",
        "plt.imshow(digit, cmap=plt.cm.binary)\n",
        "plt.show()"
      ],
      "metadata": {
        "id": "YOIWl3BhAvoH",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 266
        },
        "outputId": "688f71e5-2a75-4b3d-c1c2-48b897cf3485"
      },
      "execution_count": 4,
      "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": "code",
      "source": [
        "print (train_images.shape)\n",
        "print (train_labels.shape)\n",
        "\n",
        "print(train_images[2].shape)\n",
        "train_images[2]"
      ],
      "metadata": {
        "id": "CYxj_8S2AwiK",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "e0976e68-22cb-4a22-d0c8-ddf7d02edf01"
      },
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "(50000, 32, 32, 3)\n",
            "(50000, 1)\n",
            "(32, 32, 3)\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([[[250, 250, 248],\n",
              "        [248, 249, 243],\n",
              "        [247, 248, 239],\n",
              "        ...,\n",
              "        [250, 250, 246],\n",
              "        [250, 250, 246],\n",
              "        [249, 250, 246]],\n",
              "\n",
              "       [[250, 251, 245],\n",
              "        [248, 249, 238],\n",
              "        [247, 247, 234],\n",
              "        ...,\n",
              "        [251, 251, 242],\n",
              "        [251, 252, 243],\n",
              "        [250, 251, 243]],\n",
              "\n",
              "       [[251, 251, 244],\n",
              "        [250, 248, 237],\n",
              "        [250, 245, 233],\n",
              "        ...,\n",
              "        [250, 249, 238],\n",
              "        [250, 249, 240],\n",
              "        [250, 249, 242]],\n",
              "\n",
              "       ...,\n",
              "\n",
              "       [[221, 213, 191],\n",
              "        [221, 206, 176],\n",
              "        [225, 207, 181],\n",
              "        ...,\n",
              "        [199, 176, 134],\n",
              "        [207, 193, 165],\n",
              "        [233, 229, 226]],\n",
              "\n",
              "       [[225, 223, 204],\n",
              "        [227, 219, 196],\n",
              "        [229, 216, 200],\n",
              "        ...,\n",
              "        [204, 185, 151],\n",
              "        [212, 201, 180],\n",
              "        [234, 232, 228]],\n",
              "\n",
              "       [[233, 233, 226],\n",
              "        [234, 232, 224],\n",
              "        [235, 230, 225],\n",
              "        ...,\n",
              "        [219, 209, 194],\n",
              "        [223, 216, 207],\n",
              "        [232, 230, 228]]], dtype=uint8)"
            ]
          },
          "metadata": {},
          "execution_count": 5
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Normalizar imágenes de entrenamiento y de test\n"
      ],
      "metadata": {
        "id": "K0teESsKA8N7"
      },
      "execution_count": 6,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "print (train_images.shape)\n",
        "print (train_labels.shape)\n",
        "\n",
        "train_images = train_images.astype('float32') / 255\n",
        "test_images = test_images.astype('float32') / 255\n",
        "\n",
        "print (train_images.shape)\n",
        "print (train_labels.shape)"
      ],
      "metadata": {
        "id": "jHxaQNAdBDTQ",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "39ed8fca-ed55-4aff-aa87-6c810d7a1ed8"
      },
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "(50000, 32, 32, 3)\n",
            "(50000, 1)\n",
            "(50000, 32, 32, 3)\n",
            "(50000, 1)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Crear el Modelo de red neuronal CNN con:\n",
        "# stride = 1\n",
        "# ventanas convolucionales de 3x3 HECHO\n",
        "# ventanas de pooling de 2x2 HECHO\n",
        "# funciones de activación 'relu' HECHO\n",
        "# 3 capas convolucionales con 32, 64 y 128 filtros respectivamente HECHO\n",
        "# Aplanamos la última salida y añadimos dos capas densas con función activación \n",
        "# 'relu' de 256 y 128 neuronas cada una. Se aplana con Flatten HECHO\n",
        "\n",
        "\n",
        "model = Sequential()\n",
        "model.add(Conv2D(32, (3, 3), strides=(1,1), activation='relu', input_shape=(32, 32, 3)))\n",
        "model.add(MaxPooling2D((2, 2)))\n",
        "\n",
        "model.add(Conv2D(64, (3, 3), strides=(1,1), activation='relu', input_shape=(32, 32, 3)))\n",
        "model.add(MaxPooling2D((2, 2)))\n",
        "\n",
        "model.add(Conv2D(128, (3, 3), strides=(1,1), activation='relu', input_shape=(32, 32, 3)))\n",
        "model.add(MaxPooling2D((2, 2)))\n",
        "\n",
        "#from tensorflow.keras.layers import Flatten\n",
        "\n",
        "model.add(Flatten())\n",
        "model.add(Dense(256,activation='relu'))\n",
        "model.add(Dense(128,activation='relu'))\n",
        "model.add(Dense(100,activation='softmax'))\n",
        "\n",
        "# Tenemos en cuenta el número de clases a clasificar para poner la capa final\n",
        "model.summary() #272372"
      ],
      "metadata": {
        "id": "AQhtNiJlBGfx",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "fb70fc73-1b6b-4fc1-bd8d-04502d1642fc"
      },
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " conv2d (Conv2D)             (None, 30, 30, 32)        896       \n",
            "                                                                 \n",
            " max_pooling2d (MaxPooling2D  (None, 15, 15, 32)       0         \n",
            " )                                                               \n",
            "                                                                 \n",
            " conv2d_1 (Conv2D)           (None, 13, 13, 64)        18496     \n",
            "                                                                 \n",
            " max_pooling2d_1 (MaxPooling  (None, 6, 6, 64)         0         \n",
            " 2D)                                                             \n",
            "                                                                 \n",
            " conv2d_2 (Conv2D)           (None, 4, 4, 128)         73856     \n",
            "                                                                 \n",
            " max_pooling2d_2 (MaxPooling  (None, 2, 2, 128)        0         \n",
            " 2D)                                                             \n",
            "                                                                 \n",
            " flatten (Flatten)           (None, 512)               0         \n",
            "                                                                 \n",
            " dense (Dense)               (None, 256)               131328    \n",
            "                                                                 \n",
            " dense_1 (Dense)             (None, 128)               32896     \n",
            "                                                                 \n",
            " dense_2 (Dense)             (None, 100)               12900     \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 270,372\n",
            "Trainable params: 270,372\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Compilamos el modelo con métrica preción\n",
        "\n",
        "model.compile(loss='sparse_categorical_crossentropy',\n",
        "              optimizer='adam',\n",
        "              metrics=['accuracy'])\n",
        "\n",
        "history = model.fit(train_images, train_labels,\n",
        "          batch_size= batch_size,\n",
        "          epochs= no_epochs,\n",
        "          verbose= verbosity,\n",
        "          validation_split = validation_split)\n",
        "\n",
        "test_loss, test_acc = model.evaluate(test_images, test_labels)\n",
        "\n",
        "print('Test accuracy:', test_acc)"
      ],
      "metadata": {
        "id": "t631tbjkCFRr",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "127394a3-b389-444e-c7ed-64de07195ead"
      },
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/100\n",
            "800/800 [==============================] - 13s 6ms/step - loss: 4.1412 - accuracy: 0.0567 - val_loss: 3.7859 - val_accuracy: 0.1144\n",
            "Epoch 2/100\n",
            "800/800 [==============================] - 4s 5ms/step - loss: 3.5283 - accuracy: 0.1539 - val_loss: 3.3674 - val_accuracy: 0.1856\n",
            "Epoch 3/100\n",
            "800/800 [==============================] - 4s 5ms/step - loss: 3.1618 - accuracy: 0.2224 - val_loss: 3.1415 - val_accuracy: 0.2313\n",
            "Epoch 4/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 2.9179 - accuracy: 0.2690 - val_loss: 2.9519 - val_accuracy: 0.2634\n",
            "Epoch 5/100\n",
            "800/800 [==============================] - 4s 5ms/step - loss: 2.7224 - accuracy: 0.3073 - val_loss: 2.8223 - val_accuracy: 0.3010\n",
            "Epoch 6/100\n",
            "800/800 [==============================] - 4s 4ms/step - loss: 2.5623 - accuracy: 0.3413 - val_loss: 2.7302 - val_accuracy: 0.3135\n",
            "Epoch 7/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 2.4254 - accuracy: 0.3679 - val_loss: 2.7129 - val_accuracy: 0.3272\n",
            "Epoch 8/100\n",
            "800/800 [==============================] - 4s 5ms/step - loss: 2.3029 - accuracy: 0.3948 - val_loss: 2.6802 - val_accuracy: 0.3309\n",
            "Epoch 9/100\n",
            "800/800 [==============================] - 4s 5ms/step - loss: 2.1893 - accuracy: 0.4196 - val_loss: 2.6789 - val_accuracy: 0.3386\n",
            "Epoch 10/100\n",
            "800/800 [==============================] - 4s 5ms/step - loss: 2.0843 - accuracy: 0.4429 - val_loss: 2.6543 - val_accuracy: 0.3499\n",
            "Epoch 11/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 1.9819 - accuracy: 0.4638 - val_loss: 2.6412 - val_accuracy: 0.3527\n",
            "Epoch 12/100\n",
            "800/800 [==============================] - 4s 5ms/step - loss: 1.8854 - accuracy: 0.4877 - val_loss: 2.6768 - val_accuracy: 0.3523\n",
            "Epoch 13/100\n",
            "800/800 [==============================] - 4s 5ms/step - loss: 1.7870 - accuracy: 0.5095 - val_loss: 2.6885 - val_accuracy: 0.3504\n",
            "Epoch 14/100\n",
            "800/800 [==============================] - 4s 5ms/step - loss: 1.6995 - accuracy: 0.5262 - val_loss: 2.7614 - val_accuracy: 0.3508\n",
            "Epoch 15/100\n",
            "800/800 [==============================] - 4s 4ms/step - loss: 1.6154 - accuracy: 0.5470 - val_loss: 2.8399 - val_accuracy: 0.3477\n",
            "Epoch 16/100\n",
            "800/800 [==============================] - 4s 5ms/step - loss: 1.5340 - accuracy: 0.5690 - val_loss: 2.8909 - val_accuracy: 0.3491\n",
            "Epoch 17/100\n",
            "800/800 [==============================] - 4s 4ms/step - loss: 1.4531 - accuracy: 0.5865 - val_loss: 2.9524 - val_accuracy: 0.3409\n",
            "Epoch 18/100\n",
            "800/800 [==============================] - 4s 5ms/step - loss: 1.3728 - accuracy: 0.6072 - val_loss: 3.0568 - val_accuracy: 0.3491\n",
            "Epoch 19/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 1.2948 - accuracy: 0.6246 - val_loss: 3.1118 - val_accuracy: 0.3472\n",
            "Epoch 20/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 1.2262 - accuracy: 0.6428 - val_loss: 3.2063 - val_accuracy: 0.3419\n",
            "Epoch 21/100\n",
            "800/800 [==============================] - 4s 5ms/step - loss: 1.1645 - accuracy: 0.6585 - val_loss: 3.3343 - val_accuracy: 0.3369\n",
            "Epoch 22/100\n",
            "800/800 [==============================] - 4s 4ms/step - loss: 1.0838 - accuracy: 0.6797 - val_loss: 3.5363 - val_accuracy: 0.3377\n",
            "Epoch 23/100\n",
            "800/800 [==============================] - 4s 5ms/step - loss: 1.0294 - accuracy: 0.6926 - val_loss: 3.5377 - val_accuracy: 0.3356\n",
            "Epoch 24/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.9639 - accuracy: 0.7117 - val_loss: 3.7152 - val_accuracy: 0.3294\n",
            "Epoch 25/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.9219 - accuracy: 0.7212 - val_loss: 3.8931 - val_accuracy: 0.3257\n",
            "Epoch 26/100\n",
            "800/800 [==============================] - 4s 5ms/step - loss: 0.8566 - accuracy: 0.7417 - val_loss: 4.0551 - val_accuracy: 0.3254\n",
            "Epoch 27/100\n",
            "800/800 [==============================] - 4s 4ms/step - loss: 0.8145 - accuracy: 0.7480 - val_loss: 4.2620 - val_accuracy: 0.3270\n",
            "Epoch 28/100\n",
            "800/800 [==============================] - 4s 5ms/step - loss: 0.7691 - accuracy: 0.7621 - val_loss: 4.4412 - val_accuracy: 0.3271\n",
            "Epoch 29/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.7337 - accuracy: 0.7701 - val_loss: 4.6413 - val_accuracy: 0.3251\n",
            "Epoch 30/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.6939 - accuracy: 0.7836 - val_loss: 4.7588 - val_accuracy: 0.3176\n",
            "Epoch 31/100\n",
            "800/800 [==============================] - 4s 4ms/step - loss: 0.6606 - accuracy: 0.7915 - val_loss: 4.8148 - val_accuracy: 0.3185\n",
            "Epoch 32/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.6196 - accuracy: 0.8033 - val_loss: 5.1082 - val_accuracy: 0.3119\n",
            "Epoch 33/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.5984 - accuracy: 0.8101 - val_loss: 5.1957 - val_accuracy: 0.3156\n",
            "Epoch 34/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.5731 - accuracy: 0.8172 - val_loss: 5.3144 - val_accuracy: 0.3178\n",
            "Epoch 35/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.5444 - accuracy: 0.8264 - val_loss: 5.6158 - val_accuracy: 0.3217\n",
            "Epoch 36/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.5243 - accuracy: 0.8302 - val_loss: 5.6570 - val_accuracy: 0.3168\n",
            "Epoch 37/100\n",
            "800/800 [==============================] - 4s 4ms/step - loss: 0.5172 - accuracy: 0.8343 - val_loss: 5.9077 - val_accuracy: 0.3106\n",
            "Epoch 38/100\n",
            "800/800 [==============================] - 4s 5ms/step - loss: 0.4811 - accuracy: 0.8429 - val_loss: 6.1464 - val_accuracy: 0.3127\n",
            "Epoch 39/100\n",
            "800/800 [==============================] - 4s 5ms/step - loss: 0.4783 - accuracy: 0.8437 - val_loss: 6.2786 - val_accuracy: 0.3098\n",
            "Epoch 40/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.4595 - accuracy: 0.8515 - val_loss: 6.3844 - val_accuracy: 0.3097\n",
            "Epoch 41/100\n",
            "800/800 [==============================] - 4s 5ms/step - loss: 0.4430 - accuracy: 0.8550 - val_loss: 6.5950 - val_accuracy: 0.3115\n",
            "Epoch 42/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.4545 - accuracy: 0.8535 - val_loss: 6.6427 - val_accuracy: 0.3086\n",
            "Epoch 43/100\n",
            "800/800 [==============================] - 4s 5ms/step - loss: 0.4089 - accuracy: 0.8654 - val_loss: 6.9105 - val_accuracy: 0.3103\n",
            "Epoch 44/100\n",
            "800/800 [==============================] - 4s 5ms/step - loss: 0.4131 - accuracy: 0.8646 - val_loss: 6.9741 - val_accuracy: 0.3124\n",
            "Epoch 45/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.4141 - accuracy: 0.8649 - val_loss: 7.0102 - val_accuracy: 0.3114\n",
            "Epoch 46/100\n",
            "800/800 [==============================] - 4s 4ms/step - loss: 0.3952 - accuracy: 0.8692 - val_loss: 7.1968 - val_accuracy: 0.3117\n",
            "Epoch 47/100\n",
            "800/800 [==============================] - 4s 5ms/step - loss: 0.3747 - accuracy: 0.8765 - val_loss: 7.5186 - val_accuracy: 0.3134\n",
            "Epoch 48/100\n",
            "800/800 [==============================] - 4s 4ms/step - loss: 0.3847 - accuracy: 0.8732 - val_loss: 7.3439 - val_accuracy: 0.3044\n",
            "Epoch 49/100\n",
            "800/800 [==============================] - 4s 4ms/step - loss: 0.3783 - accuracy: 0.8771 - val_loss: 7.5900 - val_accuracy: 0.3112\n",
            "Epoch 50/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.3726 - accuracy: 0.8793 - val_loss: 7.7421 - val_accuracy: 0.3110\n",
            "Epoch 51/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.3621 - accuracy: 0.8809 - val_loss: 7.8228 - val_accuracy: 0.3065\n",
            "Epoch 52/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.3713 - accuracy: 0.8791 - val_loss: 8.0863 - val_accuracy: 0.3099\n",
            "Epoch 53/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.3440 - accuracy: 0.8882 - val_loss: 8.2099 - val_accuracy: 0.3085\n",
            "Epoch 54/100\n",
            "800/800 [==============================] - 4s 5ms/step - loss: 0.3530 - accuracy: 0.8846 - val_loss: 8.3613 - val_accuracy: 0.3107\n",
            "Epoch 55/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.3359 - accuracy: 0.8916 - val_loss: 8.2845 - val_accuracy: 0.2987\n",
            "Epoch 56/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.3463 - accuracy: 0.8902 - val_loss: 8.3165 - val_accuracy: 0.3059\n",
            "Epoch 57/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.3340 - accuracy: 0.8916 - val_loss: 8.4509 - val_accuracy: 0.3090\n",
            "Epoch 58/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.3300 - accuracy: 0.8955 - val_loss: 8.5965 - val_accuracy: 0.3054\n",
            "Epoch 59/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.3579 - accuracy: 0.8880 - val_loss: 8.7224 - val_accuracy: 0.2977\n",
            "Epoch 60/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.3290 - accuracy: 0.8938 - val_loss: 8.9531 - val_accuracy: 0.3039\n",
            "Epoch 61/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.3328 - accuracy: 0.8953 - val_loss: 8.8287 - val_accuracy: 0.3025\n",
            "Epoch 62/100\n",
            "800/800 [==============================] - 4s 4ms/step - loss: 0.3183 - accuracy: 0.8994 - val_loss: 9.0994 - val_accuracy: 0.3116\n",
            "Epoch 63/100\n",
            "800/800 [==============================] - 4s 4ms/step - loss: 0.3161 - accuracy: 0.9007 - val_loss: 9.0463 - val_accuracy: 0.3058\n",
            "Epoch 64/100\n",
            "800/800 [==============================] - 4s 5ms/step - loss: 0.3235 - accuracy: 0.8977 - val_loss: 9.0135 - val_accuracy: 0.3095\n",
            "Epoch 65/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.3093 - accuracy: 0.9013 - val_loss: 9.4231 - val_accuracy: 0.2990\n",
            "Epoch 66/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.3321 - accuracy: 0.8942 - val_loss: 9.2576 - val_accuracy: 0.3091\n",
            "Epoch 67/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.3153 - accuracy: 0.9020 - val_loss: 9.2779 - val_accuracy: 0.3056\n",
            "Epoch 68/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.2821 - accuracy: 0.9103 - val_loss: 9.6791 - val_accuracy: 0.2961\n",
            "Epoch 69/100\n",
            "800/800 [==============================] - 4s 5ms/step - loss: 0.3075 - accuracy: 0.9032 - val_loss: 9.7057 - val_accuracy: 0.3058\n",
            "Epoch 70/100\n",
            "800/800 [==============================] - 4s 5ms/step - loss: 0.3026 - accuracy: 0.9061 - val_loss: 9.7693 - val_accuracy: 0.3037\n",
            "Epoch 71/100\n",
            "800/800 [==============================] - 4s 5ms/step - loss: 0.3016 - accuracy: 0.9064 - val_loss: 9.8079 - val_accuracy: 0.3022\n",
            "Epoch 72/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.3151 - accuracy: 0.9032 - val_loss: 9.9460 - val_accuracy: 0.3018\n",
            "Epoch 73/100\n",
            "800/800 [==============================] - 4s 5ms/step - loss: 0.3070 - accuracy: 0.9051 - val_loss: 9.8002 - val_accuracy: 0.2984\n",
            "Epoch 74/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.2840 - accuracy: 0.9107 - val_loss: 10.0853 - val_accuracy: 0.3039\n",
            "Epoch 75/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.3060 - accuracy: 0.9068 - val_loss: 9.8928 - val_accuracy: 0.3002\n",
            "Epoch 76/100\n",
            "800/800 [==============================] - 4s 5ms/step - loss: 0.3203 - accuracy: 0.9040 - val_loss: 10.2965 - val_accuracy: 0.3018\n",
            "Epoch 77/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.2903 - accuracy: 0.9108 - val_loss: 10.0621 - val_accuracy: 0.2990\n",
            "Epoch 78/100\n",
            "800/800 [==============================] - 4s 4ms/step - loss: 0.2834 - accuracy: 0.9144 - val_loss: 10.2846 - val_accuracy: 0.2941\n",
            "Epoch 79/100\n",
            "800/800 [==============================] - 4s 5ms/step - loss: 0.3146 - accuracy: 0.9061 - val_loss: 10.2823 - val_accuracy: 0.3035\n",
            "Epoch 80/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.2773 - accuracy: 0.9139 - val_loss: 10.6207 - val_accuracy: 0.2998\n",
            "Epoch 81/100\n",
            "800/800 [==============================] - 4s 5ms/step - loss: 0.2971 - accuracy: 0.9101 - val_loss: 10.4437 - val_accuracy: 0.2991\n",
            "Epoch 82/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.2994 - accuracy: 0.9100 - val_loss: 10.4779 - val_accuracy: 0.3049\n",
            "Epoch 83/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.2953 - accuracy: 0.9124 - val_loss: 10.6941 - val_accuracy: 0.2972\n",
            "Epoch 84/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.2935 - accuracy: 0.9105 - val_loss: 10.8822 - val_accuracy: 0.3032\n",
            "Epoch 85/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.2924 - accuracy: 0.9136 - val_loss: 10.6323 - val_accuracy: 0.3006\n",
            "Epoch 86/100\n",
            "800/800 [==============================] - 4s 5ms/step - loss: 0.2850 - accuracy: 0.9144 - val_loss: 10.8200 - val_accuracy: 0.2967\n",
            "Epoch 87/100\n",
            "800/800 [==============================] - 4s 5ms/step - loss: 0.2714 - accuracy: 0.9180 - val_loss: 11.2558 - val_accuracy: 0.3015\n",
            "Epoch 88/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.2743 - accuracy: 0.9165 - val_loss: 10.8870 - val_accuracy: 0.3054\n",
            "Epoch 89/100\n",
            "800/800 [==============================] - 4s 5ms/step - loss: 0.2955 - accuracy: 0.9129 - val_loss: 11.2053 - val_accuracy: 0.3038\n",
            "Epoch 90/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.2613 - accuracy: 0.9219 - val_loss: 11.1760 - val_accuracy: 0.3028\n",
            "Epoch 91/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.2853 - accuracy: 0.9165 - val_loss: 11.1694 - val_accuracy: 0.3034\n",
            "Epoch 92/100\n",
            "800/800 [==============================] - 4s 5ms/step - loss: 0.2990 - accuracy: 0.9132 - val_loss: 11.4517 - val_accuracy: 0.2985\n",
            "Epoch 93/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.2922 - accuracy: 0.9148 - val_loss: 11.1390 - val_accuracy: 0.2982\n",
            "Epoch 94/100\n",
            "800/800 [==============================] - 4s 5ms/step - loss: 0.2714 - accuracy: 0.9189 - val_loss: 11.3496 - val_accuracy: 0.3040\n",
            "Epoch 95/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.2647 - accuracy: 0.9233 - val_loss: 11.5358 - val_accuracy: 0.2943\n",
            "Epoch 96/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.3219 - accuracy: 0.9073 - val_loss: 11.3736 - val_accuracy: 0.3000\n",
            "Epoch 97/100\n",
            "800/800 [==============================] - 4s 5ms/step - loss: 0.2744 - accuracy: 0.9204 - val_loss: 11.4617 - val_accuracy: 0.3011\n",
            "Epoch 98/100\n",
            "800/800 [==============================] - 4s 5ms/step - loss: 0.2535 - accuracy: 0.9248 - val_loss: 11.4754 - val_accuracy: 0.2934\n",
            "Epoch 99/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.2914 - accuracy: 0.9151 - val_loss: 11.6896 - val_accuracy: 0.3009\n",
            "Epoch 100/100\n",
            "800/800 [==============================] - 3s 4ms/step - loss: 0.3021 - accuracy: 0.9149 - val_loss: 11.7094 - val_accuracy: 0.2994\n",
            "313/313 [==============================] - 1s 3ms/step - loss: 11.5094 - accuracy: 0.3014\n",
            "Test accuracy: 0.30140000581741333\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Ejecutamos el entrenamiento con el validation_split explicitado anteriormente\n",
        "\n"
      ],
      "metadata": {
        "id": "SQMB2DMRCkcw"
      },
      "execution_count": 10,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# Generamos las métricas de generalización\n",
        "score = model.evaluate(train_images, train_labels, verbose=0)\n",
        "print(f'Test loss: {score[0]} / Test accuracy: {score[1]}')\n",
        "\n",
        "# Visualizamos history\n",
        "# Pintamos history: Pérdida\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.title('Validation loss history')\n",
        "plt.ylabel('Loss value')\n",
        "plt.xlabel('No. epoch')\n",
        "plt.show()\n",
        "\n",
        "# Pintamos history: Precisión\n",
        "plt.plot(history.history['val_accuracy'])\n",
        "plt.title('Validation accuracy history')\n",
        "plt.ylabel('Accuracy value (%)')\n",
        "plt.xlabel('No. epoch')\n",
        "plt.show()"
      ],
      "metadata": {
        "id": "Ib6jlUAllODh",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 591
        },
        "outputId": "dd845802-0c06-4abf-d7e4-6b3e333fa9f9"
      },
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Test loss: 2.4945602416992188 / Test accuracy: 0.8137400150299072\n"
          ]
        },
        {
          "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": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    }
  ]
}