Notebook_HOTT.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": [],
      "gpuType": "T4"
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "code",
      "source": [
        "! pip install POT"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "kZuJlJBxWH0V",
        "outputId": "7d367c91-5be2-4a91-93ee-56df2a204672"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Requirement already satisfied: POT in /usr/local/lib/python3.10/dist-packages (0.9.3)\n",
            "Requirement already satisfied: numpy>=1.16 in /usr/local/lib/python3.10/dist-packages (from POT) (1.23.5)\n",
            "Requirement already satisfied: scipy>=1.6 in /usr/local/lib/python3.10/dist-packages (from POT) (1.11.4)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "! pip install lda"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Bz-yCW_NWKXr",
        "outputId": "a2a0e93b-5bba-4ddf-e3a6-e74f3d5f2367"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Requirement already satisfied: lda in /usr/local/lib/python3.10/dist-packages (3.0.0)\n",
            "Requirement already satisfied: numpy<2.0,>=1.13.0 in /usr/local/lib/python3.10/dist-packages (from lda) (1.23.5)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import numpy as np\n",
        "import lda\n",
        "import nltk\n",
        "from sklearn.metrics.pairwise import euclidean_distances\n",
        "from scipy.io import loadmat\n",
        "from nltk.corpus import stopwords\n",
        "from nltk.stem.snowball import SnowballStemmer\n",
        "from scipy.special import kl_div\n",
        "from sklearn.manifold import TSNE\n",
        "import matplotlib.pyplot as plt\n",
        "import json\n",
        "import ot\n",
        "from sklearn.preprocessing import normalize\n",
        "from sklearn.model_selection import train_test_split\n",
        "import time"
      ],
      "metadata": {
        "id": "Qrvt48fppsdl"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "# HOTT.py"
      ],
      "metadata": {
        "id": "iJYIKdSLVe1p"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "import ot\n",
        "import numpy as np\n",
        "\n",
        "def sparse_ot(weights1, weights2, M):\n",
        "    \"\"\" Compute transport for (posssibly) un-normalized sparse distributions\"\"\"\n",
        "\n",
        "    weights1 = weights1/weights1.sum()\n",
        "    weights2 = weights2/weights2.sum()\n",
        "\n",
        "    active1 = np.where(weights1)[0]\n",
        "    active2 = np.where(weights2)[0]\n",
        "\n",
        "    weights_1_active = weights1[active1]\n",
        "    weights_2_active = weights2[active2]\n",
        "    M_reduced = np.ascontiguousarray(M[active1][:,active2])\n",
        "\n",
        "    return ot.emd2(weights_1_active,weights_2_active,M_reduced)\n",
        "\n",
        "def topic_ot(weights1, weights2, M):\n",
        "\n",
        "    return ot.emd2(weights1,weights2,M)\n",
        "def hott(p, q, C, threshold=None):\n",
        "    \"\"\" Hierarchical optimal topic transport.\"\"\"\n",
        "\n",
        "    # Avoid changing p and q outside of this function\n",
        "    p, q = np.copy(p), np.copy(q)\n",
        "\n",
        "    k = len(p)\n",
        "    if threshold is None:\n",
        "        threshold = 1. / (k + 1)\n",
        "\n",
        "    p[p<threshold] = 0\n",
        "    q[q<threshold] = 0\n",
        "\n",
        "    return sparse_ot(p, q, C)\n",
        "\n",
        "\n",
        "def hoftt(p, q, C):\n",
        "    \"\"\" Hierarchical optimal full topic transport.\"\"\"\n",
        "    return ot.emd2(p, q, C)\n"
      ],
      "metadata": {
        "id": "2CuL5UQ-Valq"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "# data.py"
      ],
      "metadata": {
        "id": "4VkqM_x2VO_Z"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "\n",
        "\n",
        "\n",
        "def load_wmd_data(path):\n",
        "    \"\"\"Load data used in the WMD paper.\n",
        "    \"\"\"\n",
        "    mat_data = loadmat(path, squeeze_me=True, chars_as_strings=True)\n",
        "\n",
        "    try:\n",
        "        y = mat_data['Y'].astype(np.int)\n",
        "    except KeyError:\n",
        "        y = np.concatenate((mat_data['ytr'].astype(np.int),\n",
        "                            mat_data['yte'].astype(np.int)))\n",
        "    try:\n",
        "        embeddings_of_doc_words = mat_data['X']\n",
        "    except KeyError:\n",
        "        embeddings_of_doc_words = np.concatenate((mat_data['xtr'],\n",
        "                                                  mat_data['xte']))\n",
        "    try:\n",
        "        doc_word_counts = mat_data['BOW_X']\n",
        "    except KeyError:\n",
        "        doc_word_counts = np.concatenate((mat_data['BOW_xtr'], mat_data['BOW_xte']))\n",
        "    try:\n",
        "        doc_words = mat_data['words']\n",
        "    except KeyError:\n",
        "        doc_words = np.concatenate((mat_data['words_tr'], mat_data['words_te']))\n",
        "\n",
        "    vocab = []\n",
        "    embed_vocab = {}\n",
        "    for d_w, d_e in zip(doc_words, embeddings_of_doc_words):\n",
        "        if type(d_w) == str:\n",
        "            d_w = [d_w]\n",
        "        words = [w for w in d_w if type(w) == str]\n",
        "        if len(words) == 1:\n",
        "            d_e = d_e.reshape((-1, 1))\n",
        "        for i, w in enumerate(words):\n",
        "            if w not in vocab:\n",
        "                vocab.append(w)\n",
        "                embed_vocab[w] = d_e[:, i]\n",
        "            else:\n",
        "                if not np.allclose(embed_vocab[w], d_e[:, i]):\n",
        "                    print('Problem with embeddings')\n",
        "                    break\n",
        "\n",
        "    bow_data = np.zeros((len(doc_word_counts), len(vocab)),\n",
        "                        dtype=np.int)\n",
        "    for doc_idx, (d_w, d_c) in enumerate(zip(doc_words, doc_word_counts)):\n",
        "        if type(d_w) == str:\n",
        "            d_w = [d_w]\n",
        "        words = [w for w in d_w if type(w) == str]\n",
        "        if len(words) == 1:\n",
        "            d_c = np.array([d_c])\n",
        "        words_idx = np.array([vocab.index(w) for w in words])\n",
        "        bow_data[doc_idx, words_idx] = d_c.astype(np.int)\n",
        "\n",
        "    return vocab, embed_vocab, bow_data, y\n",
        "\n",
        "\n",
        "def reduce_vocab(bow_data, vocab, embed_vocab, embed_aggregate='mean'):\n",
        "    \"\"\"Reduce vocabulary size by stemming and removing stop words.\n",
        "    \"\"\"\n",
        "    nltk.download('stopwords')\n",
        "    vocab = np.array(vocab)\n",
        "    short = np.array([len(w) > 2 for w in vocab])\n",
        "    stop_words = set(stopwords.words('english'))\n",
        "    stop = np.array([w not in stop_words for w in vocab])\n",
        "    reduced_vocab = vocab[np.logical_and(short, stop)]\n",
        "    reduced_bow_data = bow_data[:, np.logical_and(short, stop)]\n",
        "    stemmer = SnowballStemmer(\"english\")\n",
        "    stemmed_dict = {}\n",
        "    stemmed_idx_mapping = {}\n",
        "    stemmed_vocab = []\n",
        "    for i, w in enumerate(reduced_vocab):\n",
        "        stem_w = stemmer.stem(w)\n",
        "        if stem_w in stemmed_vocab:\n",
        "            stemmed_dict[stem_w].append(w)\n",
        "            stemmed_idx_mapping[stemmed_vocab.index(stem_w)].append(i)\n",
        "        else:\n",
        "            stemmed_dict[stem_w] = [w]\n",
        "            stemmed_vocab.append(stem_w)\n",
        "            stemmed_idx_mapping[stemmed_vocab.index(stem_w)] = [i]\n",
        "\n",
        "    stemmed_bow_data = np.zeros((bow_data.shape[0], len(stemmed_vocab)),\n",
        "                                dtype=np.int)\n",
        "    for i in range(len(stemmed_vocab)):\n",
        "        stemmed_bow_data[:, i] = reduced_bow_data[:, stemmed_idx_mapping[i]].sum(axis=1).flatten()\n",
        "\n",
        "    word_counts = stemmed_bow_data.sum(axis=0)\n",
        "    stemmed_reduced_vocab = np.array(stemmed_vocab)[word_counts > 2].tolist()\n",
        "    stemmed_reduced_bow_data = stemmed_bow_data[:, word_counts > 2]\n",
        "\n",
        "    stemmed_reduced_embed_vocab = {}\n",
        "    for w in stemmed_reduced_vocab:\n",
        "        old_w_embed = [embed_vocab[w_old] for w_old in stemmed_dict[w]]\n",
        "        if embed_aggregate == 'mean':\n",
        "            new_w_embed = np.mean(old_w_embed, axis=0)\n",
        "        elif embed_aggregate == 'first':\n",
        "            new_w_embed = old_w_embed[0]\n",
        "        else:\n",
        "            print('Unknown embedding aggregation')\n",
        "            break\n",
        "        stemmed_reduced_embed_vocab[w] = new_w_embed\n",
        "\n",
        "    return (stemmed_reduced_vocab,\n",
        "            stemmed_reduced_embed_vocab,\n",
        "            stemmed_reduced_bow_data)\n",
        "\n",
        "\n",
        "def get_embedded_data(bow_data, embed_vocab, vocab):\n",
        "    \"\"\"Map bag-of-words data to embedded representation.\"\"\"\n",
        "    M, V = bow_data.shape\n",
        "    embed_data = [[] for _ in range(M)]\n",
        "    for i in range(V):\n",
        "        for d in range(M):\n",
        "            if bow_data[d, i] > 0:\n",
        "                for _ in range(bow_data[d, i]):\n",
        "                    embed_data[d].append(embed_vocab[vocab[i]])\n",
        "    return [np.array(embed_doc) for embed_doc in embed_data]\n",
        "\n",
        "\n",
        "def change_embeddings(vocab, bow_data, embed_path):\n",
        "    \"\"\"Change embedding data if vocabulary has been reduced.\"\"\"\n",
        "    all_embed_vocab = {}\n",
        "    with open(embed_path, 'r') as file:\n",
        "        for line in file.readlines():\n",
        "            word = line.split(' ')[0]\n",
        "            embedding = [float(x) for x in line.split(' ')[1:]]\n",
        "            all_embed_vocab[word] = embedding\n",
        "\n",
        "    data_embed_vocab = {}\n",
        "    new_vocab_idx = []\n",
        "    new_vocab = []\n",
        "    for i, w in enumerate(vocab):\n",
        "        if w in all_embed_vocab:\n",
        "            data_embed_vocab[w] = all_embed_vocab[w]\n",
        "            new_vocab_idx.append(i)\n",
        "            new_vocab.append(w)\n",
        "    bow_data = bow_data[:, new_vocab_idx]\n",
        "    return new_vocab, data_embed_vocab, bow_data\n",
        "\n",
        "\n",
        "def fit_topics(data, embeddings, vocab, K):\n",
        "    \"\"\"Fit a topic model to bag-of-words data.\"\"\"\n",
        "    model = lda.LDA(n_topics=K, n_iter=1500, random_state=1)\n",
        "    model.fit(data)\n",
        "    topics = model.topic_word_\n",
        "    lda_centers = np.matmul(topics, embeddings)\n",
        "    print('LDA Gibbs topics')\n",
        "    n_top_words = 20\n",
        "    for i, topic_dist in enumerate(topics):\n",
        "        topic_words = np.array(vocab)[np.argsort(topic_dist)][:-(n_top_words + 1):-1]\n",
        "        topic_name = topic_words[0]\n",
        "        print(f'{topic_name}: {\", \".join(topic_words)}')\n",
        "    print('\\n')\n",
        "    topic_proportions = model.doc_topic_\n",
        "\n",
        "    return topics, lda_centers, topic_proportions\n",
        "\n",
        "def loader(data_path,\n",
        "           embeddings_path,\n",
        "           p=1,\n",
        "           K_lda=30,\n",
        "           glove_embeddings=True,\n",
        "           stemming=True,\n",
        "           n_words_keep = None):\n",
        "    \"\"\" Load dataset and embeddings from data path.\"\"\"\n",
        "    # Load dataset from data_path\n",
        "    vocab, embed_vocab, bow_data, y = load_wmd_data(data_path)\n",
        "    y = y - 1\n",
        "\n",
        "    # Use GLOVE word embeddings\n",
        "    if glove_embeddings:\n",
        "        vocab, embed_vocab, bow_data = change_embeddings(\n",
        "            vocab, bow_data, embeddings_path)\n",
        "    # Reduce vocabulary by removing short words, stop words, and stemming\n",
        "    if stemming:\n",
        "        vocab, embed_vocab, bow_data = reduce_vocab(\n",
        "            bow_data, vocab, embed_vocab, embed_aggregate='mean')\n",
        "\n",
        "    # Matrix of word embeddings\n",
        "    embeddings = np.array([embed_vocab[w] for w in vocab])\n",
        "\n",
        "    topics, lda_centers, topic_proportions = fit_topics(\n",
        "        bow_data, embeddings, vocab, K_lda)\n",
        "\n",
        "    cost_embeddings = euclidean_distances(embeddings, embeddings) ** p\n",
        "    cost_topics = np.zeros((topics.shape[0], topics.shape[0]))\n",
        "\n",
        "    ## Reduce topics to top-20 words\n",
        "    if n_words_keep is not None:\n",
        "        for k in range(K_lda):\n",
        "            to_0_idx = np.argsort(-topics[k])[n_words_keep:]\n",
        "            topics[k][to_0_idx] = 0\n",
        "\n",
        "\n",
        "    for i in range(cost_topics.shape[0]):\n",
        "        for j in range(i + 1, cost_topics.shape[1]):\n",
        "            cost_topics[i, j] = sparse_ot(topics[i], topics[j], cost_embeddings)\n",
        "    cost_topics = cost_topics + cost_topics.T\n",
        "\n",
        "    out = {'X': bow_data, 'y': y,\n",
        "           'embeddings': embeddings,\n",
        "           'topics': topics, 'proportions': topic_proportions,\n",
        "           'cost_E': cost_embeddings, 'cost_T': cost_topics}\n",
        "\n",
        "    return out"
      ],
      "metadata": {
        "id": "swVh-h5uVJNA"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "# knn_classifier.py"
      ],
      "metadata": {
        "id": "sgK0RfbfV5kH"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "import numpy as np\n",
        "\n",
        "\n",
        "def predict(neighbor_classes, C):\n",
        "    # Make sure all classes are considered\n",
        "    labels = np.concatenate((neighbor_classes, list(range(C))))\n",
        "    # Find class frequency among neighbors\n",
        "    weights = np.unique(labels, return_counts=True)[1]\n",
        "    # Find most popular class\n",
        "    prediction = np.argmax(weights)\n",
        "\n",
        "    # If most popular class is ambiguous try with fewer neighbors; else return\n",
        "    if sum(weights[prediction] == weights) > 1:\n",
        "        return predict(neighbor_classes[:-2], C)\n",
        "\n",
        "    else:\n",
        "        return prediction\n",
        "\n",
        "\n",
        "def knn(X_train, X_test, y_train, y_test, method, C, n_neighbors=3):\n",
        "    # Number of classes\n",
        "    n_classes = len(np.unique(y_train))\n",
        "\n",
        "    prediction = []\n",
        "    for doc in X_test:\n",
        "        doc_to_train = [method(doc, x, C) for x in X_train]\n",
        "        # Find indices of n_neighbors closest documents\n",
        "        rank = np.argsort(doc_to_train)[:n_neighbors]\n",
        "\n",
        "        # Make prediction based on most popular class among neighbors\n",
        "        prediction.append(predict(y_train[rank], n_classes))\n",
        "\n",
        "    # Print and return test error\n",
        "    test_error = 1 - (prediction == y_test).mean()\n",
        "    return test_error, prediction\n"
      ],
      "metadata": {
        "id": "LZjQsksvV1Kr"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "# main.py"
      ],
      "metadata": {
        "id": "7735Az8cWB6R"
      }
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 456
        },
        "id": "mQNRSSeJUEn3",
        "outputId": "26881232-8e2a-460d-c67d-a846324e4759"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "<ipython-input-5-d16ca4986818>:7: DeprecationWarning: `np.int` is a deprecated alias for the builtin `int`. To silence this warning, use `int` by itself. Doing this will not modify any behavior and is safe. When replacing `np.int`, you may wish to use e.g. `np.int64` or `np.int32` to specify the precision. If you wish to review your current use, check the release note link for additional information.\n",
            "Deprecated in NumPy 1.20; for more details and guidance: https://numpy.org/devdocs/release/1.20.0-notes.html#deprecations\n",
            "  y = mat_data['Y'].astype(np.int)\n"
          ]
        },
        {
          "output_type": "error",
          "ename": "KeyboardInterrupt",
          "evalue": "",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-8-8599bc19d51f>\u001b[0m in \u001b[0;36m<cell line: 25>\u001b[0;34m()\u001b[0m\n\u001b[1;32m     23\u001b[0m \u001b[0;31m# p=1 for W1 and p=2 for W2\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     24\u001b[0m \u001b[0mp\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;36m1\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 25\u001b[0;31m \u001b[0mdata\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mloader\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdata_name\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0membeddings_path\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mp\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mp\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     26\u001b[0m \u001b[0mprint\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdata\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     27\u001b[0m \u001b[0mcost_E\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mdata\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m'cost_E'\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m<ipython-input-5-d16ca4986818>\u001b[0m in \u001b[0;36mloader\u001b[0;34m(data_path, embeddings_path, p, K_lda, glove_embeddings, stemming, n_words_keep)\u001b[0m\n\u001b[1;32m    163\u001b[0m     \u001b[0;34m\"\"\" Load dataset and embeddings from data path.\"\"\"\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    164\u001b[0m     \u001b[0;31m# Load dataset from data_path\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 165\u001b[0;31m     \u001b[0mvocab\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0membed_vocab\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mbow_data\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mload_wmd_data\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdata_path\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    166\u001b[0m     \u001b[0my\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0my\u001b[0m \u001b[0;34m-\u001b[0m \u001b[0;36m1\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    167\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m<ipython-input-5-d16ca4986818>\u001b[0m in \u001b[0;36mload_wmd_data\u001b[0;34m(path)\u001b[0m\n\u001b[1;32m     32\u001b[0m             \u001b[0md_e\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0md_e\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mreshape\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m-\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;36m1\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     33\u001b[0m         \u001b[0;32mfor\u001b[0m \u001b[0mi\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mw\u001b[0m \u001b[0;32min\u001b[0m \u001b[0menumerate\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mwords\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[0;32m---> 34\u001b[0;31m             \u001b[0;32mif\u001b[0m \u001b[0mw\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mvocab\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     35\u001b[0m                 \u001b[0mvocab\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mw\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     36\u001b[0m                 \u001b[0membed_vocab\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mw\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0md_e\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mi\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
          ]
        }
      ],
      "source": [
        "from sklearn.preprocessing import normalize\n",
        "from sklearn.model_selection import train_test_split\n",
        "import time\n",
        "#from knn_classifier import knn\n",
        "\n",
        "#import distances\n",
        "#import hott\n",
        "\n",
        "# Download datasets used by Kusner et al from\n",
        "# https://www.dropbox.com/sh/nf532hddgdt68ix/AABGLUiPRyXv6UL2YAcHmAFqa?dl=0\n",
        "# and put them into\n",
        "data_path = ''\n",
        "\n",
        "\n",
        "# Download GloVe 6B tokens, 300d word embeddings from\n",
        "# https://nlp.stanford.edu/projects/glove/\n",
        "# and put them into\n",
        "embeddings_path = '/content/drive/MyDrive/projet_ot/glove.6B.300d.txt'\n",
        "\n",
        "# Pick a dataset (uncomment the line you want)\n",
        "data_name = '/content/drive/MyDrive/projet_ot/bbcsport-emd_tr_te_split.mat'\n",
        "\n",
        "# p=1 for W1 and p=2 for W2\n",
        "p = 1\n",
        "data = loader(data_name, embeddings_path, p=p)\n",
        "print(data)\n",
        "cost_E = data['cost_E']\n",
        "cost_T = data['cost_T']\n",
        "data['topics']"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.preprocessing import MinMaxScaler\n",
        "import numpy as np\n",
        "\n",
        "# Assuming data['topics'] is a 2D NumPy array\n",
        "data['topics'] = data['topics'] + np.random.randn(data['topics'].shape[0], data['topics'].shape[1])+ np.random.randn(data['topics'].shape[0], data['topics'].shape[1])\n",
        "\n",
        "# Create a MinMaxScaler\n",
        "#scaler = MinMaxScaler()\n",
        "\n",
        "# Fit and transform the 'topics' data\n",
        "#data['topics'] = scaler.fit_transform(data['topics'])\n"
      ],
      "metadata": {
        "id": "PR4ErsPavdWD"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.preprocessing import MinMaxScaler\n",
        "import numpy as np\n",
        "\n",
        "# Assuming data['topics'] is a 2D NumPy array\n",
        "data['X'] = data['X'] + np.random.randn(data['X'].shape[0], data['X'].shape[1])\n"
      ],
      "metadata": {
        "id": "gGqVRTCvzt_9"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Save datasets"
      ],
      "metadata": {
        "id": "2vD1syslRIwj"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Dataset 70 topics and sparse déjà enregitré avec ça\n",
        "cost_L1_topics = np.zeros((data[\"topics\"].shape[0], data[\"topics\"].shape[0]))\n",
        "for i in range(data['topics'].shape[0]):\n",
        "        for j in range(i + 1, data['topics'].shape[0]):\n",
        "            cost_L1_topics[i, j] = np.linalg.norm(data[\"topics\"][i] - data[\"topics\"][j])\n",
        "cost_L1_topics = cost_L1_topics + cost_L1_topics.T\n",
        "data['cost_L1'] = cost_L1_topics"
      ],
      "metadata": {
        "id": "7e7qtaMdfxtU"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# Dataset 70 topics and sparse déjà enregitré avec ça\n",
        "cost_ot_topics = np.zeros((data[\"topics\"].shape[0], data[\"topics\"].shape[0]))\n",
        "for i in range(data['topics'].shape[0]):\n",
        "        for j in range(i + 1, data['topics'].shape[0]):\n",
        "            cost_ot_topics[i, j] = ot.emd2(data[\"topics\"][i],data[\"topics\"][j], data['cost_E'])\n",
        "cost_ot_topics = cost_ot_topics + cost_ot_topics.T\n",
        "data['cost_OT'] = cost_ot_topics"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 589
        },
        "id": "TqCFJFiMipYP",
        "outputId": "e35f2c94-ea22-4a46-a34b-6f85fd25cf5a"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "error",
          "ename": "AssertionError",
          "evalue": "\nArrays are not almost equal to 6 decimals\na and b vector must have the same sum\nMismatched elements: 1 / 1 (100%)\nMax absolute difference: 7.20696921\nMax relative difference: 0.00390853\n x: array(1851.114894)\n y: array([1843.907925])",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mAssertionError\u001b[0m                            Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-39-8d713205e2ed>\u001b[0m in \u001b[0;36m<cell line: 3>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      3\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mi\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mrange\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdata\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m'topics'\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mshape\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\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      4\u001b[0m         \u001b[0;32mfor\u001b[0m \u001b[0mj\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mrange\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mi\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0;36m1\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdata\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m'topics'\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mshape\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\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[0;32m----> 5\u001b[0;31m             \u001b[0mcost_ot_topics\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mi\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mj\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mot\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0memd2\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdata\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m\"topics\"\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mi\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mdata\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m\"topics\"\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mj\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdata\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m'cost_E'\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[1;32m      6\u001b[0m \u001b[0mcost_ot_topics\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcost_ot_topics\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0mcost_ot_topics\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mT\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      7\u001b[0m \u001b[0mdata\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m'cost_OT'\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcost_ot_topics\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.10/dist-packages/ot/lp/__init__.py\u001b[0m in \u001b[0;36memd2\u001b[0;34m(a, b, M, processes, numItermax, log, return_matrix, center_dual, numThreads, check_marginals)\u001b[0m\n\u001b[1;32m    503\u001b[0m     \u001b[0;31m# ensure that same mass\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    504\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mcheck_marginals\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 505\u001b[0;31m         np.testing.assert_almost_equal(a.sum(0),\n\u001b[0m\u001b[1;32m    506\u001b[0m                                     \u001b[0mb\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msum\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mkeepdims\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mTrue\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0merr_msg\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m'a and b vector must have the same sum'\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    507\u001b[0m                                     decimal=6)\n",
            "    \u001b[0;31m[... skipping hidden 2 frame]\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.10/dist-packages/numpy/testing/_private/utils.py\u001b[0m in \u001b[0;36massert_array_compare\u001b[0;34m(comparison, x, y, err_msg, verbose, header, precision, equal_nan, equal_inf)\u001b[0m\n\u001b[1;32m    842\u001b[0m                                 \u001b[0mverbose\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mverbose\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mheader\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mheader\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    843\u001b[0m                                 names=('x', 'y'), precision=precision)\n\u001b[0;32m--> 844\u001b[0;31m             \u001b[0;32mraise\u001b[0m \u001b[0mAssertionError\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mmsg\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    845\u001b[0m     \u001b[0;32mexcept\u001b[0m \u001b[0mValueError\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    846\u001b[0m         \u001b[0;32mimport\u001b[0m \u001b[0mtraceback\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mAssertionError\u001b[0m: \nArrays are not almost equal to 6 decimals\na and b vector must have the same sum\nMismatched elements: 1 / 1 (100%)\nMax absolute difference: 7.20696921\nMax relative difference: 0.00390853\n x: array(1851.114894)\n y: array([1843.907925])"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "data_converted = {key: value.tolist() if isinstance(value, np.ndarray) else value for key, value in data.items()}\n",
        "\n",
        "# Enregistrer le dictionnaire\n",
        "import json\n",
        "\n",
        "# Enregistrer le dictionnaire dans un fichier JSON\n",
        "with open('/content/drive/MyDrive/projet_ot/dataset_words_30.json', 'w') as fichier:\n",
        "    json.dump(data_converted, fichier)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 259
        },
        "id": "0XzTxvOqhanU",
        "outputId": "1aa1239f-67d9-4d2d-c80c-671fcdb4b3cf"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "error",
          "ename": "NameError",
          "evalue": "name 'data' is not defined",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-50-013e0eb223e9>\u001b[0m in \u001b[0;36m<cell line: 1>\u001b[0;34m()\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mdata_converted\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m{\u001b[0m\u001b[0mkey\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0mvalue\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtolist\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mif\u001b[0m \u001b[0misinstance\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mvalue\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mndarray\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32melse\u001b[0m \u001b[0mvalue\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mkey\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mvalue\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mdata\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mitems\u001b[0m\u001b[0;34m(\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[1;32m      2\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      3\u001b[0m \u001b[0;31m# Enregistrer le dictionnaire\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      4\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mjson\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      5\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mNameError\u001b[0m: name 'data' is not defined"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "with open('/content/drive/MyDrive/projet_ot/dataset_words_30.json', 'r') as fichier:\n",
        "    loaded_data = json.load(fichier)\n",
        "\n",
        "# Convertir les listes en numpy arrays si nécessaire\n",
        "data = {key: np.array(value) if isinstance(value, list) else value for key, value in loaded_data.items()}\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 436
        },
        "id": "LkSeLTEAiCb1",
        "outputId": "cd80c4ee-85f4-4a75-bbe4-26f9d36b2f63"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "error",
          "ename": "JSONDecodeError",
          "evalue": "Expecting ',' delimiter: line 1 column 65409586 (char 65409585)",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mJSONDecodeError\u001b[0m                           Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-51-0cf6c3291510>\u001b[0m in \u001b[0;36m<cell line: 1>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[0;32mwith\u001b[0m \u001b[0mopen\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m'/content/drive/MyDrive/projet_ot/dataset_words_30.json'\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m'r'\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0mfichier\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 2\u001b[0;31m     \u001b[0mloaded_data\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mjson\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mload\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfichier\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      3\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      4\u001b[0m \u001b[0;31m# Convertir les listes en numpy arrays si nécessaire\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      5\u001b[0m \u001b[0mdata\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m{\u001b[0m\u001b[0mkey\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0marray\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mvalue\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mif\u001b[0m \u001b[0misinstance\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mvalue\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlist\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32melse\u001b[0m \u001b[0mvalue\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mkey\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mvalue\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mloaded_data\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mitems\u001b[0m\u001b[0;34m(\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[0;32m/usr/lib/python3.10/json/__init__.py\u001b[0m in \u001b[0;36mload\u001b[0;34m(fp, cls, object_hook, parse_float, parse_int, parse_constant, object_pairs_hook, **kw)\u001b[0m\n\u001b[1;32m    291\u001b[0m     \u001b[0mkwarg\u001b[0m\u001b[0;34m;\u001b[0m \u001b[0motherwise\u001b[0m\u001b[0;31m \u001b[0m\u001b[0;31m`\u001b[0m\u001b[0;31m`\u001b[0m\u001b[0mJSONDecoder\u001b[0m\u001b[0;31m`\u001b[0m\u001b[0;31m`\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0mused\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    292\u001b[0m     \"\"\"\n\u001b[0;32m--> 293\u001b[0;31m     return loads(fp.read(),\n\u001b[0m\u001b[1;32m    294\u001b[0m         \u001b[0mcls\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mcls\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mobject_hook\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mobject_hook\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    295\u001b[0m         \u001b[0mparse_float\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mparse_float\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mparse_int\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mparse_int\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/lib/python3.10/json/__init__.py\u001b[0m in \u001b[0;36mloads\u001b[0;34m(s, cls, object_hook, parse_float, parse_int, parse_constant, object_pairs_hook, **kw)\u001b[0m\n\u001b[1;32m    344\u001b[0m             \u001b[0mparse_int\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mNone\u001b[0m \u001b[0;32mand\u001b[0m \u001b[0mparse_float\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mNone\u001b[0m \u001b[0;32mand\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    345\u001b[0m             parse_constant is None and object_pairs_hook is None and not kw):\n\u001b[0;32m--> 346\u001b[0;31m         \u001b[0;32mreturn\u001b[0m \u001b[0m_default_decoder\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdecode\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0ms\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    347\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mcls\u001b[0m \u001b[0;32mis\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[1;32m    348\u001b[0m         \u001b[0mcls\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mJSONDecoder\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/lib/python3.10/json/decoder.py\u001b[0m in \u001b[0;36mdecode\u001b[0;34m(self, s, _w)\u001b[0m\n\u001b[1;32m    335\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    336\u001b[0m         \"\"\"\n\u001b[0;32m--> 337\u001b[0;31m         \u001b[0mobj\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mend\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mraw_decode\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0ms\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0midx\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0m_w\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0ms\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;36m0\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mend\u001b[0m\u001b[0;34m(\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[1;32m    338\u001b[0m         \u001b[0mend\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0m_w\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0ms\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mend\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mend\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    339\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mend\u001b[0m \u001b[0;34m!=\u001b[0m \u001b[0mlen\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0ms\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[0;32m/usr/lib/python3.10/json/decoder.py\u001b[0m in \u001b[0;36mraw_decode\u001b[0;34m(self, s, idx)\u001b[0m\n\u001b[1;32m    351\u001b[0m         \"\"\"\n\u001b[1;32m    352\u001b[0m         \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 353\u001b[0;31m             \u001b[0mobj\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mend\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mscan_once\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0ms\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0midx\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    354\u001b[0m         \u001b[0;32mexcept\u001b[0m \u001b[0mStopIteration\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0merr\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    355\u001b[0m             \u001b[0;32mraise\u001b[0m \u001b[0mJSONDecodeError\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"Expecting value\"\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0ms\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0merr\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mvalue\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[0;31mJSONDecodeError\u001b[0m: Expecting ',' delimiter: line 1 column 65409586 (char 65409585)"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "bow_data, y = data['X'], data['y']\n",
        "topic_proportions = data['proportions']\n",
        "\n",
        "seed = 0\n",
        "bow_train, bow_test, topic_train, topic_test, y_train, y_test = train_test_split(bow_data, topic_proportions, y, random_state=seed)\n",
        "\n",
        "methods = {'HOTT': hoftt, 'L1_W1': hoftt}\n",
        "\n",
        "for method in methods.keys():\n",
        "\n",
        "    t_s = time.time()\n",
        "    # Compute HOTT distance\n",
        "    if method == 'HOTT':\n",
        "      X_train, X_test = topic_train, topic_test\n",
        "      C = data['cost_T']\n",
        "\n",
        "    # Compute test error\n",
        "      test_error, prediction_HOTT = knn(X_train, X_test, y_train, y_test, methods[method], C, n_neighbors=7)\n",
        "      print(method + ' test error is %f; took %.2f seconds' % (test_error, time.time()-t_s))\n",
        "      t_s = time.time()\n",
        "\n",
        "    #Compute L1 then Wasserstein distance\n",
        "    if method == 'L1_W1':\n",
        "      X_train, X_test = topic_train, topic_test\n",
        "      C = data['cost_L1']\n",
        "\n",
        "    # Compute test error\n",
        "      test_error, prediction_L1_W1 = knn(X_train, X_test, y_train, y_test, methods[method], C, n_neighbors=7)\n",
        "      print(method + ' test error is %f; took %.2f seconds' % (test_error, time.time()-t_s))\n",
        "      t_s = time.time()\n",
        "\n",
        "# Done!"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "tguiF1SKgfP4",
        "outputId": "e02dca0c-18fa-497b-9f4e-c6f05d68abf9"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "HOTT test error is 0.048649; took 80.19 seconds\n",
            "L1_W1 test error is 0.054054; took 51.29 seconds\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Apply t-SNE for dimensionality reduction to 2D\n",
        "predicted_HOTT = np.array(prediction_HOTT)\n",
        "predicted_L1_W1 = np.array(prediction_L1_W1)\n",
        "tsne = TSNE(n_components=2, random_state=42)\n",
        "X_test_tsne = tsne.fit_transform(X_test)\n",
        "\n",
        "# Create a scatter plot\n",
        "plt.figure()\n",
        "\n",
        "# Get unique labels from y_test\n",
        "unique_labels = np.unique(y_test)\n",
        "\n",
        "# Plot the points with different colors for each class\n",
        "for label in unique_labels:\n",
        "    indices = np.where(y_test == label)\n",
        "    plt.scatter(X_test_tsne[indices, 0], X_test_tsne[indices, 1], label=f'Class {label}')\n",
        "\n",
        "#plt.title('t-SNE Visualization of X_test')\n",
        "#plt.xlabel('t-SNE Dimension 1')\n",
        "#plt.ylabel('t-SNE Dimension 2')\n",
        "\n",
        "# Add legend\n",
        "plt.axis(\"off\")\n",
        "plt.legend()\n",
        "plt.savefig(\"true_values.png\")\n",
        "plt.show()\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 406
        },
        "id": "Ez0p5Y-iWnP1",
        "outputId": "b056fe9a-bcbb-44da-c095-918b08fc0269"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [],
      "metadata": {
        "id": "Ir2d2RhQmH9Z"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# Create a scatter plot\n",
        "plt.figure()\n",
        "\n",
        "# Get unique labels from y_test\n",
        "unique_labels = np.unique(predicted_HOTT)\n",
        "\n",
        "# Plot the points with different colors for each class\n",
        "for label in unique_labels:\n",
        "    indices = np.where(predicted_HOTT == label)\n",
        "    plt.scatter(X_test_tsne[indices, 0], X_test_tsne[indices, 1], label=f'Class {label}')\n",
        "\n",
        "#plt.title('t-SNE Visualization of X_test')\n",
        "#plt.xlabel('t-SNE Dimension 1')\n",
        "#plt.ylabel('t-SNE Dimension 2')\n",
        "\n",
        "# Add legend\n",
        "plt.axis(\"off\")\n",
        "plt.legend()\n",
        "plt.savefig(\"HOTT_values.png\")\n",
        "plt.show()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 406
        },
        "id": "PTImcOhwU2YW",
        "outputId": "b80c8c38-e141-440f-894d-33973e5d4411"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Create a scatter plot\n",
        "plt.figure()\n",
        "\n",
        "# Get unique labels from y_test\n",
        "unique_labels = np.unique(predicted_L1_W1)\n",
        "\n",
        "# Plot the points with different colors for each class\n",
        "for label in unique_labels:\n",
        "    indices = np.where(predicted_L1_W1 == label)\n",
        "    plt.scatter(X_test_tsne[indices, 0], X_test_tsne[indices, 1], label=f'Class {label}')\n",
        "\n",
        "#plt.title('t-SNE Visualization of X_test')\n",
        "#plt.xlabel('t-SNE Dimension 1')\n",
        "#plt.ylabel('t-SNE Dimension 2')\n",
        "\n",
        "# Add legend\n",
        "plt.axis(\"off\")\n",
        "plt.legend()\n",
        "plt.savefig(\"L1_values.png\")\n",
        "plt.show()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 406
        },
        "id": "CjMhjoqwYAXY",
        "outputId": "8b24da30-67e1-409b-eae4-af7a5f7dbf17"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# KL + OT\n",
        "def kl(p, q):\n",
        "\n",
        "    return np.sum(np.where((p != 0), p * np.log(p / q), 0))\n",
        "\n",
        "cost_KL_topics = np.zeros((data[\"topics\"].shape[0], data[\"topics\"].shape[0]))\n",
        "for i in range(data['topics'].shape[0]):\n",
        "        for j in range(i + 1, data['topics'].shape[0]):\n",
        "            cost_KL_topics[i, j] = kl(data[\"topics\"][i],data[\"topics\"][j])\n",
        "cost_KL_topics = cost_KL_topics + cost_KL_topics.T\n",
        "data['cost_KL'] = cost_KL_topics\n",
        "C = data['cost_KL']\n",
        "\n",
        "    # Compute test error\n",
        "test_error, prediction_KL_W1 = knn(X_train, X_test, y_train, y_test, methods[method], C, n_neighbors=7)\n",
        "print(method + ' test error is %f; took %.2f seconds' % (test_error, time.time()-t_s))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "iIpI8WW7oz7r",
        "outputId": "95b5f5a2-ab82-4d66-d33d-3dff0184168f"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "<ipython-input-42-a94ee407e7da>:4: RuntimeWarning: divide by zero encountered in divide\n",
            "  return np.sum(np.where((p != 0), p * np.log(p / q), 0))\n",
            "<ipython-input-42-a94ee407e7da>:4: RuntimeWarning: divide by zero encountered in log\n",
            "  return np.sum(np.where((p != 0), p * np.log(p / q), 0))\n",
            "<ipython-input-42-a94ee407e7da>:4: RuntimeWarning: invalid value encountered in multiply\n",
            "  return np.sum(np.where((p != 0), p * np.log(p / q), 0))\n",
            "/usr/local/lib/python3.10/dist-packages/ot/lp/__init__.py:571: UserWarning: Problem infeasible. Check that a and b are in the simplex\n",
            "  check_result(result_code)\n",
            "/usr/local/lib/python3.10/dist-packages/ot/lp/__init__.py:571: UserWarning: Problem infeasible. Check that a and b are in the simplex\n",
            "  check_result(result_code)\n",
            "/usr/local/lib/python3.10/dist-packages/ot/lp/__init__.py:571: UserWarning: Problem infeasible. Check that a and b are in the simplex\n",
            "  check_result(result_code)\n",
            "/usr/local/lib/python3.10/dist-packages/ot/lp/__init__.py:571: UserWarning: Problem infeasible. Check that a and b are in the simplex\n",
            "  check_result(result_code)\n",
            "/usr/local/lib/python3.10/dist-packages/ot/lp/__init__.py:571: UserWarning: Problem infeasible. Check that a and b are in the simplex\n",
            "  check_result(result_code)\n",
            "/usr/local/lib/python3.10/dist-packages/ot/lp/__init__.py:571: UserWarning: Problem infeasible. Check that a and b are in the simplex\n",
            "  check_result(result_code)\n",
            "/usr/local/lib/python3.10/dist-packages/ot/lp/__init__.py:571: UserWarning: Problem infeasible. Check that a and b are in the simplex\n",
            "  check_result(result_code)\n",
            "/usr/local/lib/python3.10/dist-packages/ot/lp/__init__.py:571: UserWarning: Problem infeasible. Check that a and b are in the simplex\n",
            "  check_result(result_code)\n",
            "/usr/local/lib/python3.10/dist-packages/ot/lp/__init__.py:571: UserWarning: Problem infeasible. Check that a and b are in the simplex\n",
            "  check_result(result_code)\n",
            "/usr/local/lib/python3.10/dist-packages/ot/lp/__init__.py:571: UserWarning: Problem infeasible. Check that a and b are in the simplex\n",
            "  check_result(result_code)\n",
            "/usr/local/lib/python3.10/dist-packages/ot/lp/__init__.py:571: UserWarning: Problem infeasible. Check that a and b are in the simplex\n",
            "  check_result(result_code)\n",
            "/usr/local/lib/python3.10/dist-packages/ot/lp/__init__.py:571: UserWarning: Problem infeasible. Check that a and b are in the simplex\n",
            "  check_result(result_code)\n",
            "/usr/local/lib/python3.10/dist-packages/ot/lp/__init__.py:571: UserWarning: Problem infeasible. Check that a and b are in the simplex\n",
            "  check_result(result_code)\n",
            "/usr/local/lib/python3.10/dist-packages/ot/lp/__init__.py:571: UserWarning: Problem infeasible. Check that a and b are in the simplex\n",
            "  check_result(result_code)\n",
            "/usr/local/lib/python3.10/dist-packages/ot/lp/__init__.py:571: UserWarning: Problem infeasible. Check that a and b are in the simplex\n",
            "  check_result(result_code)\n",
            "/usr/local/lib/python3.10/dist-packages/ot/lp/__init__.py:571: UserWarning: Problem infeasible. Check that a and b are in the simplex\n",
            "  check_result(result_code)\n",
            "/usr/local/lib/python3.10/dist-packages/ot/lp/__init__.py:571: UserWarning: Problem infeasible. Check that a and b are in the simplex\n",
            "  check_result(result_code)\n",
            "/usr/local/lib/python3.10/dist-packages/ot/lp/__init__.py:571: UserWarning: Problem infeasible. Check that a and b are in the simplex\n",
            "  check_result(result_code)\n",
            "/usr/local/lib/python3.10/dist-packages/ot/lp/__init__.py:571: UserWarning: Problem infeasible. Check that a and b are in the simplex\n",
            "  check_result(result_code)\n",
            "/usr/local/lib/python3.10/dist-packages/ot/lp/__init__.py:571: UserWarning: Problem infeasible. Check that a and b are in the simplex\n",
            "  check_result(result_code)\n",
            "/usr/local/lib/python3.10/dist-packages/ot/lp/__init__.py:571: UserWarning: Problem infeasible. Check that a and b are in the simplex\n",
            "  check_result(result_code)\n",
            "/usr/local/lib/python3.10/dist-packages/ot/lp/__init__.py:571: UserWarning: Problem infeasible. Check that a and b are in the simplex\n",
            "  check_result(result_code)\n",
            "/usr/local/lib/python3.10/dist-packages/ot/lp/__init__.py:571: UserWarning: Problem infeasible. Check that a and b are in the simplex\n",
            "  check_result(result_code)\n",
            "/usr/local/lib/python3.10/dist-packages/ot/lp/__init__.py:571: UserWarning: Problem infeasible. Check that a and b are in the simplex\n",
            "  check_result(result_code)\n",
            "/usr/local/lib/python3.10/dist-packages/ot/lp/__init__.py:571: UserWarning: Problem infeasible. Check that a and b are in the simplex\n",
            "  check_result(result_code)\n",
            "/usr/local/lib/python3.10/dist-packages/ot/lp/__init__.py:571: UserWarning: Problem infeasible. Check that a and b are in the simplex\n",
            "  check_result(result_code)\n",
            "/usr/local/lib/python3.10/dist-packages/ot/lp/__init__.py:571: UserWarning: Problem infeasible. Check that a and b are in the simplex\n",
            "  check_result(result_code)\n",
            "/usr/local/lib/python3.10/dist-packages/ot/lp/__init__.py:571: UserWarning: Problem infeasible. Check that a and b are in the simplex\n",
            "  check_result(result_code)\n",
            "/usr/local/lib/python3.10/dist-packages/ot/lp/__init__.py:571: UserWarning: Problem infeasible. Check that a and b are in the simplex\n",
            "  check_result(result_code)\n",
            "/usr/local/lib/python3.10/dist-packages/ot/lp/__init__.py:571: UserWarning: Problem infeasible. Check that a and b are in the simplex\n",
            "  check_result(result_code)\n",
            "/usr/local/lib/python3.10/dist-packages/ot/lp/__init__.py:571: UserWarning: Problem infeasible. Check that a and b are in the simplex\n",
            "  check_result(result_code)\n",
            "/usr/local/lib/python3.10/dist-packages/ot/lp/__init__.py:571: UserWarning: Problem infeasible. Check that a and b are in the simplex\n",
            "  check_result(result_code)\n",
            "/usr/local/lib/python3.10/dist-packages/ot/lp/__init__.py:571: UserWarning: Problem infeasible. Check that a and b are in the simplex\n",
            "  check_result(result_code)\n",
            "/usr/local/lib/python3.10/dist-packages/ot/lp/__init__.py:571: UserWarning: Problem infeasible. Check that a and b are in the simplex\n",
            "  check_result(result_code)\n",
            "/usr/local/lib/python3.10/dist-packages/ot/lp/__init__.py:571: UserWarning: Problem infeasible. Check that a and b are in the simplex\n",
            "  check_result(result_code)\n",
            "/usr/local/lib/python3.10/dist-packages/ot/lp/__init__.py:571: UserWarning: Problem infeasible. Check that a and b are in the simplex\n",
            "  check_result(result_code)\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "L1_W1 test error is 0.891892; took 131.59 seconds\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "print(data['cost_T'].shape)\n",
        "print(data['cost_L1'].shape)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "mbRPN1z4ZId8",
        "outputId": "bb069334-e81c-4e10-f792-22f59fdbf247"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "(30, 30)\n",
            "(30, 30)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from matplotlib.lines import Line2D\n",
        "predicted_KL_W1 = np.array(prediction_KL_W1)\n",
        "\n",
        "plt.figure()\n",
        "\n",
        "# Get unique labels from y_test\n",
        "unique_labels = np.unique(predicted_KL_W1)\n",
        "\n",
        "# Plot the points with different colors for each class\n",
        "for label in unique_labels:\n",
        "    indices = np.where(predicted_KL_W1 == label)\n",
        "    plt.scatter(X_test_tsne[indices, 0], X_test_tsne[indices, 1], label=f'Class {label}')\n",
        "\n",
        "#plt.title('t-SNE Visualization of X_test')\n",
        "#plt.xlabel('t-SNE Dimension 1')\n",
        "#plt.ylabel('t-SNE Dimension 2')\n",
        "\n",
        "# Add legend\n",
        "plt.axis(\"off\")\n",
        "plt.legend()\n",
        "plt.savefig(\"KL_values.png\")\n",
        "plt.show()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 406
        },
        "id": "BQDkRbH0y8En",
        "outputId": "58ec65e0-eae4-4c8a-f7e7-8d69c101a827"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.metrics import accuracy_score,precision_score, recall_score\n",
        "\n",
        "# Assuming predicted_HOTT is a 1D array of predicted labels\n",
        "accuracy = accuracy_score(y_test, predicted_HOTT)\n",
        "precision_micro = precision_score(y_test, predicted_HOTT, average='micro')\n",
        "recall_micro = recall_score(y_test, predicted_HOTT, average='micro')\n",
        "print(f'Accuracy HOTT: {accuracy * 100:.2f}%')\n",
        "print(f'Recall HOTT: {recall_micro * 100:.2f}%')\n",
        "print(f'Precision HOTT: {precision_micro * 100:.2f}%')\n",
        "accuracy = accuracy_score(y_test, predicted_L1_W1)\n",
        "precision_micro = precision_score(y_test, predicted_L1_W1, average='micro')\n",
        "recall_micro = recall_score(y_test, predicted_L1_W1, average='micro')\n",
        "print(f'Accuracy L1-OT: {accuracy * 100:.2f}%')\n",
        "print(f'Recall L1-OT: {recall_micro * 100:.2f}%')\n",
        "print(f'Precision L1-OT: {precision_micro * 100:.2f}%')\n",
        "# Assuming predicted_HOTT is a 1D array of predicted labels\n",
        "accuracy = accuracy_score(y_test, predicted_KL_W1)\n",
        "precision_micro = precision_score(y_test, predicted_KL_W1, average='micro')\n",
        "recall_micro = recall_score(y_test, predicted_KL_W1, average='micro')\n",
        "print(f'Accuracy KL: {accuracy * 100:.2f}%')\n",
        "print(f'Recall KL-OT: {recall_micro * 100:.2f}%')\n",
        "print(f'Precision KL-OT: {precision_micro * 100:.2f}%')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "RfsXSPrlmRkK",
        "outputId": "595abf52-f76d-4180-a0f7-1a233ed594e1"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Accuracy HOTT: 97.84%\n",
            "Recall HOTT: 97.84%\n",
            "Precision HOTT: 97.84%\n",
            "Accuracy L1-OT: 96.76%\n",
            "Recall L1-OT: 96.76%\n",
            "Precision L1-OT: 96.76%\n",
            "Accuracy KL: 94.59%\n",
            "Recall KL-OT: 94.59%\n",
            "Precision KL-OT: 94.59%\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# DATA 70"
      ],
      "metadata": {
        "id": "8wKjnnrobSBG"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "print(data_70)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "hT7uz58mPWyS",
        "outputId": "763c89fe-5eb8-424c-83b2-9f449b4bfcf4"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "{'X': array([[1, 2, 4, ..., 0, 0, 0],\n",
            "       [0, 0, 0, ..., 0, 0, 0],\n",
            "       [0, 0, 1, ..., 0, 0, 0],\n",
            "       ...,\n",
            "       [0, 0, 0, ..., 0, 0, 0],\n",
            "       [0, 0, 0, ..., 0, 0, 0],\n",
            "       [0, 0, 0, ..., 0, 0, 0]]), 'y': array([0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
            "       0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
            "       0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
            "       0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
            "       0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
            "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
            "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
            "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
            "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
            "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
            "       1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,\n",
            "       2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,\n",
            "       2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,\n",
            "       2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,\n",
            "       2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,\n",
            "       2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,\n",
            "       2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,\n",
            "       2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,\n",
            "       2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,\n",
            "       2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,\n",
            "       2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,\n",
            "       2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,\n",
            "       2, 2, 2, 2, 2, 2, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,\n",
            "       3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,\n",
            "       3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,\n",
            "       3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,\n",
            "       3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,\n",
            "       3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,\n",
            "       3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 4,\n",
            "       4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4,\n",
            "       4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4,\n",
            "       4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4,\n",
            "       4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4,\n",
            "       4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4]), 'embeddings': array([[-0.27643   ,  0.320935  ,  0.23405   , ...,  0.1866295 ,\n",
            "        -0.02417   , -0.1088535 ],\n",
            "       [-0.175304  ,  0.24774767,  0.14967   , ..., -0.72072333,\n",
            "        -0.15531667, -0.21975667],\n",
            "       [ 0.10689   ,  0.566245  ,  0.1396955 , ..., -0.03442   ,\n",
            "         0.062178  , -0.587205  ],\n",
            "       ...,\n",
            "       [ 0.22219   ,  0.92183   ,  0.10766   , ..., -0.081299  ,\n",
            "         0.1556    ,  0.33486   ],\n",
            "       [ 0.45473   ,  0.18718   ,  0.37229   , ..., -0.52887   ,\n",
            "         1.1778    , -0.32292   ],\n",
            "       [ 0.30182   ,  0.35544   , -0.1622    , ...,  0.26037   ,\n",
            "        -0.011199  ,  0.052316  ]]), 'topics': array([[0., 0., 0., ..., 0., 0., 0.],\n",
            "       [0., 0., 0., ..., 0., 0., 0.],\n",
            "       [0., 0., 0., ..., 0., 0., 0.],\n",
            "       ...,\n",
            "       [0., 0., 0., ..., 0., 0., 0.],\n",
            "       [0., 0., 0., ..., 0., 0., 0.],\n",
            "       [0., 0., 0., ..., 0., 0., 0.]]), 'proportions': array([[0.00102041, 0.00102041, 0.00102041, ..., 0.01122449, 0.00102041,\n",
            "        0.0622449 ],\n",
            "       [0.00142857, 0.00142857, 0.00142857, ..., 0.00142857, 0.00142857,\n",
            "        0.18714286],\n",
            "       [0.02066667, 0.00066667, 0.00066667, ..., 0.00066667, 0.02733333,\n",
            "        0.00066667],\n",
            "       ...,\n",
            "       [0.00126582, 0.00126582, 0.00126582, ..., 0.08987342, 0.01392405,\n",
            "        0.00126582],\n",
            "       [0.00093458, 0.00093458, 0.00093458, ..., 0.00093458, 0.01962617,\n",
            "        0.02897196],\n",
            "       [0.00096154, 0.00096154, 0.00096154, ..., 0.00096154, 0.00096154,\n",
            "        0.00096154]]), 'cost_E': array([[ 0.        ,  6.60246383,  8.92130749, ...,  8.92280546,\n",
            "        11.20264757,  8.68721576],\n",
            "       [ 6.60246383,  0.        ,  7.60976123, ...,  8.27806809,\n",
            "        10.96897321,  7.47870144],\n",
            "       [ 8.92130749,  7.60976123,  0.        , ...,  9.74017468,\n",
            "        11.64515142,  9.62337975],\n",
            "       ...,\n",
            "       [ 8.92280546,  8.27806809,  9.74017468, ...,  0.        ,\n",
            "        11.26659304,  8.65675009],\n",
            "       [11.20264757, 10.96897321, 11.64515142, ..., 11.26659304,\n",
            "         0.        , 11.20142335],\n",
            "       [ 8.68721576,  7.47870144,  9.62337975, ...,  8.65675009,\n",
            "        11.20142335,  0.        ]]), 'cost_T': array([[0.        , 6.68011692, 6.47477294, ..., 6.19678536, 6.29592686,\n",
            "        6.58870439],\n",
            "       [6.68011692, 0.        , 6.14679514, ..., 5.42666547, 5.77127954,\n",
            "        6.30865949],\n",
            "       [6.47477294, 6.14679514, 0.        , ..., 5.57354814, 6.28063566,\n",
            "        6.75680953],\n",
            "       ...,\n",
            "       [6.19678536, 5.42666547, 5.57354814, ..., 0.        , 5.58624175,\n",
            "        5.70014553],\n",
            "       [6.29592686, 5.77127954, 6.28063566, ..., 5.58624175, 0.        ,\n",
            "        5.88535366],\n",
            "       [6.58870439, 6.30865949, 6.75680953, ..., 5.70014553, 5.88535366,\n",
            "        0.        ]]), 'cost_L1': array([[0.        , 0.17099752, 0.18080012, ..., 0.19529256, 0.14820951,\n",
            "        0.19049337],\n",
            "       [0.        , 0.        , 0.19300013, ..., 0.18242319, 0.14671363,\n",
            "        0.19312297],\n",
            "       [0.        , 0.        , 0.        , ..., 0.20634629, 0.18167115,\n",
            "        0.22252364],\n",
            "       ...,\n",
            "       [0.        , 0.        , 0.        , ..., 0.        , 0.18842389,\n",
            "        0.22008431],\n",
            "       [0.        , 0.        , 0.        , ..., 0.        , 0.        ,\n",
            "        0.17854911],\n",
            "       [0.        , 0.        , 0.        , ..., 0.        , 0.        ,\n",
            "        0.        ]])}\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Get data back\n",
        "file_path = '/content/drive/MyDrive/projet_ot/dataset_words.json'\n",
        "\n",
        "# Load the data from the JSON file\n",
        "with open(file_path, 'r') as json_file:\n",
        "    loaded_data = json.load(json_file)\n",
        "\n",
        "# Convert lists back to NumPy arrays in the loaded data\n",
        "data_70 = {key: np.array(value) if isinstance(value, list) else value for key, value in loaded_data.items()}\n"
      ],
      "metadata": {
        "id": "yFNcV7_ANlO2"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "print(data_70[\"topics\"].shape)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "LUWVblk6VIEC",
        "outputId": "8dc8ca49-8650-41af-be76-caff0adc4d4b"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "(70, 3657)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "print(data_70[\"cost_L1\"])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "uWfhuK7YVK24",
        "outputId": "5f9b2cca-9670-42f8-e84a-7a60ecb681a0"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[[0.         1.46938776 1.22541497 ... 1.28397522 0.59619431 0.87932404]\n",
            " [1.46938776 0.         1.8807619  ... 1.89200667 1.72313883 1.80249515]\n",
            " [1.22541497 1.8807619  0.         ... 0.84250584 0.93686385 1.06772851]\n",
            " ...\n",
            " [1.28397522 1.89200667 0.84250584 ... 0.         1.17957801 1.31828794]\n",
            " [0.59619431 1.72313883 0.93686385 ... 1.17957801 0.         0.84951628]\n",
            " [0.87932404 1.80249515 1.06772851 ... 1.31828794 0.84951628 0.        ]]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Dataset 70 topics and sparse déjà enregitré avec ça\n",
        "cost_L1_topics = np.zeros((data_70[\"topics\"].shape[0], data_70[\"topics\"].shape[0]))\n",
        "for i in range(data_70['topics'].shape[0]):\n",
        "        for j in range(i + 1, data_70['topics'].shape[0]):\n",
        "            cost_L1_topics[i, j] = np.linalg.norm(data_70[\"proportions\"][i] - data_70[\"proportions\"][j], 1)\n",
        "cost_L1_topics = cost_L1_topics + cost_L1_topics.T\n",
        "data_70['cost_L1'] = cost_L1_topics"
      ],
      "metadata": {
        "id": "jeqF7sBDjsSP"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "print(bow_train)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "uanZwJtCV0Y8",
        "outputId": "6a720c16-1725-4322-82e9-7ccfc3493206"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[[0 0 0 ... 0 0 0]\n",
            " [0 0 0 ... 0 0 0]\n",
            " [0 0 0 ... 0 0 0]\n",
            " ...\n",
            " [0 0 0 ... 0 0 0]\n",
            " [0 0 0 ... 0 0 0]\n",
            " [0 0 0 ... 0 0 0]]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "bow_data, y = data_70['X'], data_70['y']\n",
        "topic_proportions = data_70['proportions']\n",
        "\n",
        "seed = 0\n",
        "bow_train, bow_test, topic_train, topic_test, y_train, y_test = train_test_split(bow_data, topic_proportions, y, random_state=seed)\n",
        "\n",
        "methods = {'HOTT': hoftt}\n",
        "\n",
        "for method in methods.keys():\n",
        "\n",
        "    t_s = time.time()\n",
        "    # Compute HOTT distance\n",
        "    X_train, X_test = topic_train, topic_test\n",
        "    C = data_70['cost_T']\n",
        "\n",
        "    # Compute test error\n",
        "    test_error, prediction_HOTT = knn(X_train, X_test, y_train, y_test, methods[method], C, n_neighbors=7)\n",
        "    print(' HOTT test error is %f; took %.2f seconds' % (test_error, time.time()-t_s))\n",
        "    t_s = time.time()\n",
        "\n",
        "    #Compute L1 then Wasserstein distance\n",
        "    X_train, X_test = topic_train, topic_test\n",
        "    C = data_70['cost_L1']\n",
        "\n",
        "    # Compute test error\n",
        "    test_error, prediction_L1_W1 = knn(X_train, X_test, y_train, y_test, methods[method], C, n_neighbors=7)\n",
        "    print(' L1 test error is %f; took %.2f seconds' % (test_error, time.time()-t_s))\n",
        "    t_s = time.time()\n",
        "\n",
        "# Done!"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "qV2NOpxsOlHf",
        "outputId": "c85c9128-a453-4f2b-d441-9a894f8f75d4"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            " HOTT test error is 0.021622; took 130.22 seconds\n",
            " L1 test error is 0.064865; took 122.51 seconds\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Apply t-SNE for dimensionality reduction to 2D\n",
        "predicted_HOTT = np.array(prediction_HOTT)\n",
        "predicted_L1_W1 = np.array(prediction_L1_W1)\n",
        "tsne = TSNE(n_components=2, random_state=42)\n",
        "X_test_tsne = tsne.fit_transform(X_test)\n",
        "\n",
        "# Create a scatter plot\n",
        "plt.figure()\n",
        "\n",
        "# Get unique labels from y_test\n",
        "unique_labels = np.unique(y_test)\n",
        "\n",
        "# Plot the points with different colors for each class\n",
        "for label in unique_labels:\n",
        "    indices = np.where(y_test == label)\n",
        "    plt.scatter(X_test_tsne[indices, 0], X_test_tsne[indices, 1], label=f'Class {label}')\n",
        "\n",
        "#plt.title('t-SNE Visualization of X_test')\n",
        "#plt.xlabel('t-SNE Dimension 1')\n",
        "#plt.ylabel('t-SNE Dimension 2')\n",
        "\n",
        "# Add legend\n",
        "plt.axis(\"off\")\n",
        "plt.legend()\n",
        "plt.savefig(\"true_values_70.png\")\n",
        "plt.show()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 406
        },
        "id": "3r4TA7ITl-u8",
        "outputId": "bbd2498b-3438-437a-8193-f58a0d368e3a"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Create a scatter plot\n",
        "plt.figure()\n",
        "\n",
        "# Get unique labels from y_test\n",
        "unique_labels = np.unique(predicted_HOTT)\n",
        "\n",
        "# Plot the points with different colors for each class\n",
        "for label in unique_labels:\n",
        "    indices = np.where(predicted_HOTT == label)\n",
        "    plt.scatter(X_test_tsne[indices, 0], X_test_tsne[indices, 1], label=f'Class {label}')\n",
        "\n",
        "#plt.title('t-SNE Visualization of X_test')\n",
        "#plt.xlabel('t-SNE Dimension 1')\n",
        "#plt.ylabel('t-SNE Dimension 2')\n",
        "\n",
        "# Add legend\n",
        "plt.axis(\"off\")\n",
        "plt.legend()\n",
        "plt.savefig(\"HOTT_values_70.png\")\n",
        "plt.show()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 406
        },
        "id": "hUocroE_mbCW",
        "outputId": "1ebe5662-7d7c-474c-c587-6c2a000100c5"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Create a scatter plot\n",
        "plt.figure()\n",
        "\n",
        "# Get unique labels from y_test\n",
        "unique_labels = np.unique(predicted_L1_W1)\n",
        "\n",
        "# Plot the points with different colors for each class\n",
        "for label in unique_labels:\n",
        "    indices = np.where(predicted_L1_W1 == label)\n",
        "    plt.scatter(X_test_tsne[indices, 0], X_test_tsne[indices, 1], label=f'Class {label}')\n",
        "\n",
        "#plt.title('t-SNE Visualization of X_test')\n",
        "#plt.xlabel('t-SNE Dimension 1')\n",
        "#plt.ylabel('t-SNE Dimension 2')\n",
        "\n",
        "# Add legend\n",
        "plt.axis(\"off\")\n",
        "plt.legend()\n",
        "plt.savefig(\"L1_values_70.png\")\n",
        "plt.show()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 406
        },
        "id": "GH101sDgmN7Y",
        "outputId": "ab24feaf-7402-41ed-a2fd-a3345311a843"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "print(methods.keys())"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "vh3oDLTPnPkw",
        "outputId": "2eab8a27-8c21-437c-fce3-a9bd71ef3354"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "dict_keys(['HOTT', 'L1_W1'])\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# KL + OT\n",
        "def kl(p, q):\n",
        "\n",
        "    return np.sum(np.where((p != 0), p * np.log(p / q), 0))\n",
        "\n",
        "cost_KL_topics = np.zeros((data_70[\"topics\"].shape[0], data_70[\"topics\"].shape[0]))\n",
        "for i in range(data_70['topics'].shape[0]):\n",
        "        for j in range(i + 1, data_70['topics'].shape[0]):\n",
        "            cost_KL_topics[i, j] = kl(data_70[\"topics\"][i],data_70[\"topics\"][j])\n",
        "cost_KL_topics = cost_KL_topics + cost_KL_topics.T\n",
        "data_70['cost_KL'] = cost_KL_topics\n",
        "C = data_70['cost_KL']\n",
        "\n",
        "    # Compute test error\n",
        "for method in methods.keys():\n",
        "  test_error, prediction_KL_W1 = knn(X_train, X_test, y_train, y_test, methods['HOTT'], C, n_neighbors=3)\n",
        "  print(method + ' test error is %f; took %.2f seconds' % (test_error, time.time()-t_s))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "yaOguvfJlkxi",
        "outputId": "235c1ae0-fab2-4784-ad63-fdae6b4c43d7"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "<ipython-input-57-df7e942f0044>:4: RuntimeWarning: divide by zero encountered in divide\n",
            "  return np.sum(np.where((p != 0), p * np.log(p / q), 0))\n",
            "<ipython-input-57-df7e942f0044>:4: RuntimeWarning: invalid value encountered in divide\n",
            "  return np.sum(np.where((p != 0), p * np.log(p / q), 0))\n",
            "<ipython-input-57-df7e942f0044>:4: RuntimeWarning: divide by zero encountered in log\n",
            "  return np.sum(np.where((p != 0), p * np.log(p / q), 0))\n",
            "<ipython-input-57-df7e942f0044>:4: RuntimeWarning: invalid value encountered in multiply\n",
            "  return np.sum(np.where((p != 0), p * np.log(p / q), 0))\n",
            "/usr/local/lib/python3.10/dist-packages/ot/lp/__init__.py:571: UserWarning: Problem infeasible. Check that a and b are in the simplex\n",
            "  check_result(result_code)\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "HOTT test error is 0.632432; took 1445.92 seconds\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from matplotlib.lines import Line2D\n",
        "predicted_KL_W1 = np.array(prediction_KL_W1)\n",
        "\n",
        "plt.figure()\n",
        "\n",
        "# Get unique labels from y_test\n",
        "unique_labels = np.unique(predicted_KL_W1)\n",
        "\n",
        "# Plot the points with different colors for each class\n",
        "for label in unique_labels:\n",
        "    indices = np.where(predicted_KL_W1 == label)\n",
        "    plt.scatter(X_test_tsne[indices, 0], X_test_tsne[indices, 1], label=f'Class {label}')\n",
        "\n",
        "#plt.title('t-SNE Visualization of X_test')\n",
        "#plt.xlabel('t-SNE Dimension 1')\n",
        "#plt.ylabel('t-SNE Dimension 2')\n",
        "\n",
        "# Add legend\n",
        "plt.axis(\"off\")\n",
        "plt.legend()\n",
        "plt.savefig(\"KL_values_70.png\")\n",
        "plt.show()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 406
        },
        "id": "ciBEyXaLl3Qi",
        "outputId": "7225eb9c-8b32-4537-c2cb-81941f105e28"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.metrics import accuracy_score,precision_score, recall_score\n",
        "\n",
        "# Assuming predicted_HOTT is a 1D array of predicted labels\n",
        "accuracy = accuracy_score(y_test, predicted_HOTT)\n",
        "precision_micro = precision_score(y_test, predicted_HOTT, average='micro')\n",
        "recall_micro = recall_score(y_test, predicted_HOTT, average='micro')\n",
        "print(f'Accuracy HOTT: {accuracy * 100:.2f}%')\n",
        "print(f'Recall HOTT: {recall_micro * 100:.2f}%')\n",
        "print(f'Precision HOTT: {precision_micro * 100:.2f}%')\n",
        "accuracy = accuracy_score(y_test, predicted_L1_W1)\n",
        "precision_micro = precision_score(y_test, predicted_L1_W1, average='micro')\n",
        "recall_micro = recall_score(y_test, predicted_L1_W1, average='micro')\n",
        "print(f'Accuracy L1-OT: {accuracy * 100:.2f}%')\n",
        "print(f'Recall L1-OT: {recall_micro * 100:.2f}%')\n",
        "print(f'Precision L1-OT: {precision_micro * 100:.2f}%')\n",
        "print(accuracy_score(y_test, predicted_KL_W1))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Ma3pGntgnq0N",
        "outputId": "7e2e84e5-1567-4946-9934-9d3472bb3128"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Accuracy HOTT: 97.84%\n",
            "Recall HOTT: 97.84%\n",
            "Precision HOTT: 97.84%\n",
            "Accuracy L1-OT: 96.76%\n",
            "Recall L1-OT: 96.76%\n",
            "Precision L1-OT: 96.76%\n",
            "0.3675675675675676\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "def mmd(p, q, kernel='rbf'):\n",
        "    # Ensure p and q are 2D arrays\n",
        "    p, q = np.atleast_2d(p), np.atleast_2d(q)\n",
        "\n",
        "    # Calculate pairwise kernel matrices\n",
        "    K_pp = pairwise_kernels(p, p, metric=kernel)\n",
        "    K_qq = pairwise_kernels(q, q, metric=kernel)\n",
        "    K_pq = pairwise_kernels(p, q, metric=kernel)\n",
        "\n",
        "    # Compute mean embeddings\n",
        "    mu_p = np.mean(K_pp, axis=0)\n",
        "    mu_q = np.mean(K_qq, axis=0)\n",
        "\n",
        "    # Calculate MMD\n",
        "    mmd = np.linalg.norm(mu_p - mu_q)\n",
        "\n",
        "    return mmd\n"
      ],
      "metadata": {
        "id": "kQiPyHeqShJF"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# MMD + OT\n",
        "def kl(p, q):\n",
        "\n",
        "    return np.sum(np.where((p != 0), p * np.log(p / q), 0))\n",
        "\n",
        "cost_MMD_topics = np.zeros((data_70[\"topics\"].shape[0], data_70[\"topics\"].shape[0]))\n",
        "for i in range(data_70['topics'].shape[0]):\n",
        "        for j in range(i + 1, data_70['topics'].shape[0]):\n",
        "            cost_MMD_topics[i, j] = mmd(data_70[\"topics\"][i],data_70[\"topics\"][j])\n",
        "cost_MMD_topics = cost_MMD_topics + cost_MMD_topics.T\n",
        "data_70['cost_MMD'] = cost_MMD_topics\n",
        "C = data_70['cost_MMD']\n",
        "\n",
        "    # Compute test error\n",
        "for method in methods.keys():\n",
        "  test_error, prediction_KL_W1 = knn(X_train, X_test, y_train, y_test, methods['HOTT'], C, n_neighbors=3)\n",
        "  print(method + ' test error is %f; took %.2f seconds' % (test_error, time.time()-t_s))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "5MY-JPtkTHPV",
        "outputId": "2cc3adc7-90eb-431b-e536-1c78c901b0b4"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "HOTT test error is 0.632432; took 388.27 seconds\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Create a scatter plot\n",
        "plt.figure()\n",
        "\n",
        "# Get unique labels from y_test\n",
        "unique_labels = np.unique(predicted_L1_W1)\n",
        "\n",
        "# Plot the points with different colors for each class\n",
        "for label in unique_labels:\n",
        "    indices = np.where(predicted_L1_W1 == label)\n",
        "    plt.scatter(X_test_tsne[indices, 0], X_test_tsne[indices, 1], label=f'Class {label}')\n",
        "\n",
        "#plt.title('t-SNE Visualization of X_test')\n",
        "#plt.xlabel('t-SNE Dimension 1')\n",
        "#plt.ylabel('t-SNE Dimension 2')\n",
        "\n",
        "# Add legend\n",
        "plt.axis(\"off\")\n",
        "plt.legend()\n",
        "plt.savefig(\"MMD_values_70.png\")\n",
        "plt.show()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 406
        },
        "id": "4tKVMkpxYLMI",
        "outputId": "510e126f-7779-47ed-9d78-f4ff1da12d5a"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAgMAAAGFCAYAAABg2vAPAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/bCgiHAAAACXBIWXMAAA9hAAAPYQGoP6dpAABbgUlEQVR4nO3de3gUZZo3/m+lg2BCQoKS7sAvQoKJEg7KKY5G351IJAEGxtcgjsMAOyzsqEGEAAY5KLgiMMyAh0HW9RUPMzsD7iisIhtCUC5xQRMMKBh1ADERzIEhhIwJENLdvz+aapJOH6q6q6qrqr+fuebSdFd3V3diP3c9z/3ct+B0Op0gIiKiiBUV7hMgIiKi8GIwQEREFOEYDBAREUU4BgNEREQRjsEAERFRhGMwQEREFOEYDBAREUU4BgNEREQRjsEAERFRhGMwQEREFOEYDBAREUU4BgNEREQRjsEAERFRhGMwQEREFOGiw30CRERkXHa7HZcvXw73aUSsbt26wWKxhPw8DAaIiEg2p9OJuro6NDU1hftUIl5CQgJsNhsEQQj6ORgMEBGRbGIgkJSUhJiYmJAGIgqO0+lEa2srGhoaAADJyclBPxeDASIiksVut7sDgeuuuy7cpxPRrr32WgBAQ0MDkpKSgl4yYAIhERHJIuYIxMTEhPlMCLj6ewgld4PBABERBYVLA/qgxO+BywREZDoOhxO1x5rQ0nwJsfHdkZyegKgoDlxEvjAYICJTOXGoAfu2HkNL0yX3bbEJ3XHXA+kYODwpjGdGpF9cJiAi0zhxqAElLx/tFAgAQEvTJZS8fBQnDjWE6czIaARBwPbt28N9GpphMEBEpuBwOLFv6zG/x3z81jE4HE6NzoiksDucOHDiLP778GkcOHEWdg1+P3V1dXj00UeRlpaG7t27IyUlBRMnTsSePXtUf20pnE4nnnzySSQnJ+Paa69Fbm4ujh3z/7cdKi4TEJEp1B5r6jIj4OnHc5dQe6wJ/W5K1OisyJ+So7VY+V4Vas9fdN+W3KsHnpqYifwhwe+Z9+e7775DdnY2EhISsG7dOgwdOhSXL1/Grl27UFhYiK+//lqV15Xjt7/9LV544QW88cYbSE1NxfLly5GXl4eqqir06NFDldfkzAARmUJLs/9AQO5xpK6So7V4+E+VnQIBAKg7fxEP/6kSJUdrVXndRx55BIIgoLy8HAUFBcjIyMDgwYNRVFSETz75xOfjiouLkZGRgZiYGKSlpWH58uWdtvJ9/vnnyMnJQVxcHOLj4zFy5EgcPHgQAFBdXY2JEyciMTERsbGxGDx4MHbu3On1dZxOJ5577jksW7YMP//5zzFs2DC8+eab+OGHH1RdtuDMABGZQmx8d0WPI/XYHU6sfK8K3hYEnAAEACvfq8I9mTZYFNwF0tjYiJKSEqxatQqxsbFd7k9ISPD52Li4OLz++uvo27cvjhw5gtmzZyMuLg6PP/44AGDq1KkYPnw4Nm3aBIvFgsOHD6Nbt24AgMLCQrS1teGjjz5CbGwsqqqq0LNnT6+vc/LkSdTV1SE3N9d9W69evXDbbbfhwIED+MUvfhHCJ+AbgwEiMoXk9ATEJnT3u1TQM9G1zZDCq/xkY5cZgY6cAGrPX0T5yUbcPlC5CofHjx+H0+nEzTffLPuxy5Ytc//7gAEDsHDhQmzZssUdDNTU1GDRokXu505PT3cfX1NTg4KCAgwdOhQAkJaW5vN16urqAABWq7XT7Var1X2fGrhMQESmEBUl4K4H0v0ekz35RtYb0IGGf/gOBII5TiqnM/jkxK1btyI7Oxs2mw09e/bEsmXLUFNT476/qKgIs2bNQm5uLtasWYMTJ06475s7dy6eeeYZZGdn46mnnsIXX3wR0vtQA4MBItI1h8OJ09+cw98q6nD6m3N+dwMMHJ6E/N8MQWyC96WAj//rOLcX6kBSnLQkOKnHSZWeng5BEGQnCR44cABTp07F+PHjsWPHDhw6dAhLly5FW1ub+5gVK1bgyy+/xIQJE/DBBx8gMzMT27ZtAwDMmjUL3377LaZNm4YjR45g1KhRePHFF72+ls1mAwDU19d3ur2+vt59nxoYDBCRbp041IA3l+zH9g2HsPvVKmzfcAhvLtnvd0AfODwJd97vfYYg1HoDcgIT8i0rtTeSe/WArzkaAa5dBVmpvRV93d69eyMvLw8bN25ES0tLl/t9tWPev38/+vfvj6VLl2LUqFFIT09HdXV1l+MyMjIwf/58lJaW4r777sNrr73mvi8lJQUPPfQQ3nnnHSxYsACvvPKK19dKTU2FzWbrtM2xubkZn376KW6//XaZ71g65gwQGUyklNoVCwh5Egf0/N8M8VpR0OFw4uP/ClxvIPWWPrI+N1Y2VI4lSsBTEzPx8J8qIQCdEgnF38hTEzMVTR4Ubdy4EdnZ2cjKysLTTz+NYcOGob29Hbt378amTZvw1VdfdXlMeno6ampqsGXLFowePRrvv/+++6ofAC5cuIBFixZh8uTJSE1NxalTp1BRUYGCggIAwLx58zBu3DhkZGTg3Llz+PDDDzFo0CCv5ycIAubNm4dnnnkG6enp7q2Fffv2xb333qv45yFiMEBkIJEyIEktIORtQFej3kCwgQn5lj8kGZt+NaJLnQGbynUG0tLSUFlZiVWrVmHBggWora1Fnz59MHLkSGzatMnrYyZNmoT58+djzpw5uHTpEiZMmIDly5djxYoVAACLxYKzZ89i+vTpqK+vx/XXX4/77rsPK1euBOBq+VxYWIhTp04hPj4e+fn52LBhg89zfPzxx9HS0oJ//dd/RVNTE+68806UlJSoVmMAAARnKBkVRKQZXwOSyEwD0ulvzmH7hkMBj7t3/vAuA/rfKuqw+9WqgI+9518ykTE68Bqsw+HEm0v2B9ylMG3VHaacofHm4sWLOHnyJFJTU0MeoOwOJ8pPNqLhHxeRFOdaGlBjRsDMlPh9cGaAyABCuVI2olAKCMmtNxBo2YWVDdVliRIU3T5IwWEwQGQAkTYghVJASE69ASnLLqxsSJGAuwmIDCDSBiRxQPfHVwEhKfUG7pySjpOfn5HU4ZCVDSkSMBggMoBIG5CkDui+lkR81Rvomdgd+b8ZgtRb+kjucBhKYEJkFFwmIDKASCy1Kw7ontP4PRO7484pgXdPDByehNRb+njNBzj9zTlZyy53PZDuN3nTX2BCZAQMBogMQLxSjrQByd+A7slXIqC3HAq5yy6hBiZEesdggMggInVA8jWgdyS3/kIwyy5yAhMio2EwQGQgHJC6CqYgULDLLlICEyIjYgIhkcGIA1LGaBv63ZQY0YGA1PoLnj0EQk1QJPMTBAHbt28P92lohsEAERlKx2ZBX3zwveREQE+BdhyYddmFgLq6Ojz66KNIS0tD9+7dkZKSgokTJ3ZqDhRO77zzDsaOHYvrrrsOgiDg8OHDqr8mlwmIyDC85QZI4SthkMsuOuCwA9X7gR/rgZ5WoP8dQJRFtZf77rvvkJ2djYSEBKxbtw5Dhw7F5cuXsWvXLhQWFspub6yGlpYW3HnnnZgyZQpmz56tyWsyGCAiv/TSJTFQbwZ//CUMMg8gjKreBUqKgeYfrt4W3xfIXwtkTlLlJR955BEIgoDy8nLExsa6bx88eDBmzpzp83HFxcXYtm0bTp06BZvNhqlTp+LJJ59Et27dAACff/455s2bh4MHD0IQBKSnp+Pll1/GqFGjUF1djTlz5uDjjz9GW1sbBgwYgHXr1mH8+PFeX2vatGkAXIGLVhgMEJFPeumSKCU3wBez1V8wjap3gbemo3MDYwDNta7bp7ypeEDQ2NiIkpISrFq1qlMgIEpISPD52Li4OLz++uvo27cvjhw5gtmzZyMuLg6PP/44AGDq1KkYPnw4Nm3aBIvFgsOHD7sDhcLCQrS1teGjjz5CbGwsqqqq0LNnT0XfW6gYDBCRV3pq2yulN4MvTATUIYfdNSPgGQgAV24TgJLFwM0TFF0yOH78OJxOJ26++WbZj122bJn73wcMGICFCxdiy5Yt7mCgpqYGixYtcj93evrVBNWamhoUFBRg6NChAFxtlPWGCYRE1EWwWfpqCabnAhMBdax6f+elgS6cQPNp13EKcjqD/3vdunUrsrOzYbPZ0LNnTyxbtgw1NTXu+4uKijBr1izk5uZizZo1OHHihPu+uXPn4plnnkF2djaeeuopfPHFFyG9DzUwGCCiLuR0SdSC1CJB2ZNvxD3/kol75w/HtFV3MBDQqx/rlT1OovT0dAiCIDtJ8MCBA5g6dSrGjx+PHTt24NChQ1i6dCna2trcx6xYsQJffvklJkyYgA8++ACZmZnYtm0bAGDWrFn49ttvMW3aNBw5cgSjRo3Ciy++qOh7CxWDAaII1nGb3ulvzrmv9KVeiZ/6urHLY9UgtVnQsLtTWH/BCHpalT1Oot69eyMvLw8bN25ES0tLl/ubmpq8Pm7//v3o378/li5dilGjRiE9PR3V1dVdjsvIyMD8+fNRWlqK++67D6+99pr7vpSUFDz00EN45513sGDBArzyyiuKvS8lMGeAKEL5Sw6UeiV+8H+ufiGqmVgYqb0ZTKv/Ha5dA8218J43ILju73+H4i+9ceNGZGdnIysrC08//TSGDRuG9vZ27N69G5s2bcJXX33V5THp6emoqanBli1bMHr0aLz//vvuq34AuHDhAhYtWoTJkycjNTUVp06dQkVFBQoKCgAA8+bNw7hx45CRkYFz587hww8/xKBBg3yeY2NjI2pqavDDD66llG+++QYAYLPZYLPZlPw43DgzQBSBxORAz6UAMTnwwo9tAa/EPYmPPXGoQclTdWORIBOJsri2DwIAPAO4Kz/nr1Gl3kBaWhoqKyuRk5ODBQsWYMiQIbjnnnuwZ88ebNq0yetjJk2ahPnz52POnDm49dZbsX//fixfvtx9v8ViwdmzZzF9+nRkZGRgypQpGDduHFauXAkAsNvtKCwsxKBBg5Cfn4+MjAy89NJLPs/x3XffxfDhwzFhwgQAwC9+8QsMHz4c//7v/67gJ9GZ4Awlo4KIDMfhcOLNJfsD1uXPnpyOXa/I39ffM7E7pq26Q7WrdL3UPYhkFy9exMmTJ5GamooePXoE/0Re6wz0cwUCKtUZMCMlfh9cJiCKMFKTA6/t2c1rl8RAxMRCtQr5sEiQiWROcm0f1LACIXnHYIAowkhNDmxpvoSM0bZO5Xoba1vw2c6uiVPBvgYRoixA6l3hPouIx5wBoggjNTlQPK5jl8SUm3or+hpEpA8MBogijNRtet5K+IbyWCLSLwYDRBFG3Kbnj69teqE8loj0i8EAUQQKZZset/gRmQ8TCIki1MDhSZ2SA+Vs0wvlsUSkPwwGiCJYKNv0uMWPyDy4TEBERBThGAwQERF5EAQB27dvD/dpaIbBAJEJ+epGSERAXV0dHn30UaSlpaF79+5ISUnBxIkTsWfPnnCfGi5fvozi4mIMHToUsbGx6Nu3L6ZPn+5uWqQW5gwQmYy/boRSMv1Z+5+0ZHfYUdlQiTOtZ9Anpg9GJI2ARcVyxN999x2ys7ORkJCAdevWYejQobh8+TJ27dqFwsJCfP3116q9thStra2orKzE8uXLccstt+DcuXN47LHHMGnSJBw8eFC112WjIiITEbsR+hJo61+ogQRFBqUaFZVVl2FN+RrUt9a7b7PGWLE4azFy++cqcapdjB8/Hl988QW++eYbxMbGdrqvqakJCQkJAFzLBNu2bcO9994LACguLsa2bdtw6tQp2Gw2TJ06FU8++SS6desGAPj8888xb948HDx4EIIgID09HS+//DJGjRqF6upqzJkzBx9//DHa2towYMAArFu3DuPHj5d0zhUVFcjKykJ1dTVuuOGGLvezURERuTkcTuzbeszvMR+/dQypt/TxeqXvK5AQWxOzhgApqay6DEV7i+BE5+vRhtYGFO0twvqfrlc8IGhsbERJSQlWrVrVJRAA4A4EvImLi8Prr7+Ovn374siRI5g9ezbi4uLw+OOPAwCmTp2K4cOHY9OmTbBYLDh8+LA7UCgsLERbWxs++ugjxMbGoqqqCj179pR83ufPn4cgCH7PL1QMBoi8cdgN10lNajdCbx0FQw0kiOSwO+xYU76mSyAAAE44IUDA2vK1yEnJUXTJ4Pjx43A6nbj55ptlP3bZsmXufx8wYAAWLlyILVu2uIOBmpoaLFq0yP3c6elXK3XW1NSgoKAAQ4cOBQCkpaVJft2LFy+iuLgYDz74IOLj42Wft1QMBog8ee2x3hfIX6vrHutyuhF6CiWQIJKrsqGy09KAJyecqGutQ2VDJUbbRiv2uqGsim/duhUvvPACTpw4gR9//BHt7e2dBueioiLMmjULf/zjH5Gbm4v7778fAwcOBADMnTsXDz/8MEpLS5Gbm4uCggIMGzYs4GtevnwZU6ZMgdPpxKZNm4I+dym4m4Coo6p3gbemdw4EAKC51nV71bvhOS8fOu4aaD3fJukx3joKhhJIEMl1pvWMosdJlZ6eDkEQZCcJHjhwAFOnTsX48eOxY8cOHDp0CEuXLkVb29X/5lasWIEvv/wSEyZMwAcffIDMzExs27YNADBr1ix8++23mDZtGo4cOYJRo0bhxRdf9PuaYiBQXV2N3bt3qzorADAYILrKYXfNCHiZunTfVrLYdZyUp1N5e9+JQw14c8l+bN9wCLtfrcL//vU4hAAz+L46Cspta0wUij4xfRQ9TqrevXsjLy8PGzduREtLS5f7m5qavD5u//796N+/P5YuXYpRo0YhPT0d1dXVXY7LyMjA/PnzUVpaivvuuw+vvfaa+76UlBQ89NBDeOedd7BgwQK88sorPs9TDASOHTuGsrIyXHfddfLfrExcJiASVe/vOiPQiRNoPu06LvUuv0+ldla+r2S/QLOgvjoKiq2J/S0VsDUxKWVE0ghYY6xoaG3wmjcgQIA1xooRSSMUf+2NGzciOzsbWVlZePrppzFs2DC0t7dj9+7d2LRpE7766qsuj0lPT0dNTQ22bNmC0aNH4/3333df9QPAhQsXsGjRIkyePBmpqak4deoUKioqUFBQAACYN28exo0bh4yMDJw7dw4ffvghBg0a5PX8Ll++jMmTJ6OyshI7duyA3W5HXV0dAFcwc8011yj+mQCcGSC66kffa5hyjhMHas+BVczKP3GoIdgzBCAt2c9zhiBQR0G2JiYtWaIsWJy1GIBr4O9I/Lk4q1iVegNpaWmorKxETk4OFixYgCFDhuCee+7Bnj17fK7LT5o0CfPnz8ecOXNw6623Yv/+/Vi+fPnV92Ox4OzZs5g+fToyMjIwZcoUjBs3DitXrgQA2O12FBYWYtCgQcjPz0dGRgZeeuklr691+vRpvPvuuzh16hRuvfVWJCcnu/+/f/9+xT8PEesMEIlO7gPe+Fng42bs8Dkz4HA48eaS/QGvsKetuiPogfX0N+ewfcOhgMdlT74RMb2ukVU4yNuMRs/E7rhzCusM0FVq1hmwxdhQnFWsWp0BM2KdASIl9b/DtWuguRbe8wYE1/397/D5FFpk5UtN4ovpdQ0yRttkPTdbE5OWcvvnIiclR9MKhOQdgwEiUZTFtX3wrekABHQOCK4Mhvlr/NYb0CIrX+1kv2BaE7OEMQXLEmVRdPsgBYfBAFFHmZOAKW/6qDOwJmCdAS2y8vWW7McSxkTGx2CAyFPmJODmCUFVINRioBaT/fz1INAq2Y8ljInMgbsJiLyJsriSBIdOdv1T4hqmVln5A4cnIf83QxCb0HmGIdCuASVJLWHM9slE+seZASKFiQO12ln54U72YwljIvNgMECkAq0G6mCS/ZTCEsZE5sFggDRhd9gjbvtQOAdqLUhNgmysbcHpb85xhwGRjjEYINV5KyxijbFicdZiFhYxMCnJkgDw2c5qfLazmjsMiHSMCYSkqrLqMhTtLerSrrShtQFFe4tQVl0WpjOjUElJluxIqXLMRFoQBAHbt28P92lohsEAqcbusGNN+RqvjUjE29aWr4VdYhdA0h9fuxr84Q4DCre6ujo8+uijSEtLQ/fu3ZGSkoKJEydiz5494T41AK52yDfffDNiY2ORmJiI3NxcfPrpp6q+JpcJSDWVDZVdZgQ6csKJutY6VDZUsgKZgXVMljz1dSMO/k/X1q4dcYcBdeS029F68DO0nzmD6D59EDNqJASLevlE3333HbKzs5GQkIB169Zh6NChuHz5Mnbt2oXCwkJ8/fXXqr22VBkZGfjDH/6AtLQ0XLhwARs2bMDYsWNx/Phx9OmjbFtnEWcGSDVnWs8oehzpl5gsmdg3VtLx3GFAANBcWorjY3JRM2MGfli4EDUzZuD4mFw0l5aq9pqPPPIIBEFAeXk5CgoKkJGRgcGDB6OoqAiffPKJz8cVFxcjIyMDMTExSEtLw/Lly3H58mX3/Z9//jlycnIQFxeH+Ph4jBw5EgcPHgQAVFdXY+LEiUhMTERsbCwGDx6MnTt3+nytX/7yl8jNzUVaWhoGDx6M9evXo7m5GV988YVyH4QHzgyQavrESItgpR5H+qdFOWYyh+bSUpx+bB7g0Ti3vb7edfvzzyF+7FhFX7OxsRElJSVYtWoVYmO7Bq4JCQk+HxsXF4fXX38dffv2xZEjRzB79mzExcXh8ccfBwBMnToVw4cPx6ZNm2CxWHD48GF069YNAFBYWIi2tjZ89NFHiI2NRVVVFXr27CnpnNva2vAf//Ef6NWrF2655Rb5b1oiBgOkmhFJI2CNsaKhtcFr3oAAAdYYK0YkjQjD2ZEa9NY3gfTJabej/tnVXQIB151OQBBQ/+xqxI0Zo+iSwfHjx+F0OnHzzTfLfuyyZcvc/z5gwAAsXLgQW7ZscQcDNTU1WLRokfu509OvJtfW1NSgoKAAQ4cOBQCkpaUFfL0dO3bgF7/4BVpbW5GcnIzdu3fj+uuvl33eUnGZgFRjibJgcdZiAK6BvyPx5+KsYtPXG4gkWpVjJmNrPfgZ2uvqfB/gdKK9rg6tBz9T9HWd3oIPibZu3Yrs7GzYbDb07NkTy5YtQ01Njfv+oqIizJo1C7m5uVizZg1OnDjhvm/u3Ll45plnkJ2djaeeekrSdH9OTg4OHz6M/fv3Iz8/H1OmTEFDg3o7cRgMkKpy++di/U/XIymm895ya4wV63+6nnUGTEgPfRNI39rPSMsTknqcVOnp6RAEQXaS4IEDBzB16lSMHz8eO3bswKFDh7B06VK0tbW5j1mxYgW+/PJLTJgwAR988AEyMzOxbds2AMCsWbPw7bffYtq0aThy5AhGjRqFF1980e9rxsbG4sYbb8RPfvITvPrqq4iOjsarr74q/01LxGUCUl1u/1zkpOREXAXCSBbuvgmkb9ESM+KlHidV7969kZeXh40bN2Lu3Lld8gaampq85g3s378f/fv3x9KlS923VVd33TWTkZGBjIwMzJ8/Hw8++CBee+01/N//+38BACkpKXjooYfw0EMP4YknnsArr7yCRx99VPK5OxwOXLqkXuItgwHShCXKwu2DEcbs5ZgpeDGjRiLaZkN7fb33vAFBQLTViphRIxV/7Y0bNyI7OxtZWVl4+umnMWzYMLS3t2P37t3YtGkTvvrqqy6PSU9PR01NDbZs2YLRo0fj/fffd1/1A8CFCxewaNEiTJ48GampqTh16hQqKipQUFAAAJg3bx7GjRuHjIwMnDt3Dh9++CEGDRrk9fxaWlqwatUqTJo0CcnJyfj73/+OjRs34vTp07j//vsV/zxEXCYgIiJNCRYLrEueuPKDx2zRlZ+tS55Qpd5AWloaKisrkZOTgwULFmDIkCG45557sGfPHmzatMnrYyZNmoT58+djzpw5uPXWW7F//34sX77cfb/FYsHZs2cxffp0ZGRkYMqUKRg3bhxWrlwJALDb7SgsLMSgQYOQn5+PjIwMvPTSS15fy2Kx4Ouvv3Zve5w4cSLOnj2Lffv2YfDgwYp/HiLBGUpGBRERRZyLFy/i5MmTSE1NRY8ePYJ+nubSUtQ/u7pTMmG0zQbrkicU31ZoZkr8PrhMQEREYRE/dizixozRtAIhecdggIiIwkawWBB7W1a4TyPiMWeAiIgowjEYICIiinBcJiAyMYfDyb3+RBQQgwEikzpxqAH7th7r1CcgNqE77nognVUAiagTLhMQmdCJQw0oeflol4ZBLU2XUPLyUZw4pF6NcyIyHgYDRCbjcDixb+sxv8d8/NYxOBwsMUJELgwGiEym9liT3xbCAPDjuUuoPdakzQkRke4xGCAymZZmac1MpB5HKnLYgZP7gCN/df3TYQ/3GdEVgiBg+/bt4T4NzTAYIDKZ2PjugQ+ScRyppOpd4LkhwBs/A97+F9c/nxviup1UVVdXh0cffRRpaWno3r07UlJSMHHiROzZsyfcp9bFQw89BEEQ8Nxzz6n6OtxNQGQyyekJiE3o7nepoGeia5shhUnVu8Bb0wF45G0017pun/ImkDkpLKemNa23v3733XfIzs5GQkIC1q1bh6FDh+Ly5cvYtWsXCgsL8fXXX6v22nJt27YNn3zyCfr27av6a3FmgMhkoqIE3PVAut9j7pySznoD4eKwAyXF6BIIAFdvK1kcEUsGJw414M0l+7F9wyHsfrUK2zccwptL9qu62+WRRx6BIAgoLy93dwYcPHgwioqK8Mknn/h8XHFxMTIyMhATE4O0tDQsX74cly9fdt//+eefIycnB3FxcYiPj8fIkSNx8OBBAEB1dTUmTpyIxMRExMbGYvDgwdi5c6ff8zx9+jQeffRR/Od//ie6deumzJv3gzMDRCY0cHgS8n8zpEudgZ6J3XHnFNYZCKvq/UDzD34OcALNp13Hpd6l2WlpTdz+6knc/pr/myGK/502NjaipKQEq1atQmxsbJf7ExISfD42Li4Or7/+Ovr27YsjR45g9uzZiIuLw+OPPw4AmDp1KoYPH45NmzbBYrHg8OHD7kG8sLAQbW1t+OijjxAbG4uqqir07NnT52s5HA5MmzYNixYtUrVtcUcMBohMauDwJKTe0ocVCPXmx3pljzMgqdtfU2/po+jf6/Hjx+F0OnHzzTfLfuyyZcvc/z5gwAAsXLgQW7ZscQcDNTU1WLRokfu509Ovzs7V1NSgoKAAQ4cOBQCkpaX5fa21a9ciOjoac+fOlX2ewWIwQGRiUVEC+t2UGO7ToI56WpU9LhgOu2vm4cd61+v0vwOI0q5tsJztr0r+/TqdwdfW2Lp1K1544QWcOHECP/74I9rb2xEfH+++v6ioCLNmzcIf//hH5Obm4v7778fAgQMBAHPnzsXDDz+M0tJS5ObmoqCgAMOGDfP6Op999hmef/55VFZWQhC0C9yZM0DkhdNuR8un5Ti/4320fFoOp93867ekkf53APF9Afj6oheA+H6u49Sgg10M4dr+mp6eDkEQZCcJHjhwAFOnTsX48eOxY8cOHDp0CEuXLkVbW5v7mBUrVuDLL7/EhAkT8MEHHyAzMxPbtm0DAMyaNQvffvstpk2bhiNHjmDUqFF48cUXvb7Wvn370NDQgBtuuAHR0dGIjo5GdXU1FixYgAEDBgT93gMRnKGESkQm1FxaivpnV6O9rs59W7TNBuuSJxA/dmwYz4xMw72bAOicSHglQFBrN4GvXQwyX/fixYs4efIkUlNT0aNHD9mncfqbc9i+4VDA4+6dP1zxma1x48bhyJEj+Oabb7rkDTQ1NbnzBgRBwLZt23Dvvffi97//PV566SWcOHHCfeysWbPw17/+FU1NTV5f58EHH0RLSwvefbdrkPXEE0/g/fffxxdffNHlvrNnz6K2trbTbXl5eZg2bRp+/etf46abburymFB/HwBnBog6aS4txenH5nUKBACgvb4epx+bh+bS0jCdGZlK5iTXwBuf3Pn2+L7qBQI62sUgbn/1R63trxs3boTdbkdWVhbefvttHDt2DF999RVeeOEF3H777V4fk56ejpqaGmzZsgUnTpzACy+84L7qB4ALFy5gzpw52Lt3L6qrq/G///u/qKiowKBBgwAA8+bNw65du3Dy5ElUVlbiww8/dN/n6brrrsOQIUM6/b9bt26w2WxeAwGlMGeA6Aqn3Y76Z1cD3ibLnE5AEFD/7GrEjRkDwaLd+iqZVOYk4OYJyq3dB8oD0NEuBnH7q7fdBCK1tr+mpaWhsrISq1atwoIFC1BbW4s+ffpg5MiR2LRpk9fHTJo0CfPnz8ecOXNw6dIlTJgwAcuXL8eKFSsAABaLBWfPnsX06dNRX1+P66+/Hvfddx9WrlwJALDb7SgsLMSpU6cQHx+P/Px8bNiwQfH3FgouExBd0fJpOWpmzAh43A1vvIHY27I0OCMyNSWT+KredV31dxzs4/sC+WuvzjIc+asrRyCQgleBoZP9HqLEtDTgvc02t7/Kp8TvgzMDRFe0nzmj6HFEPkkZvOU8l5RqhnrYxeCB21/1gzkDRFdE9+mj6HFEXomDt+eUvTh4y8nql5MH0P8OIC7Zy3EilXcx+CBuf80YbUO/mxIZCIQJgwGiK2JGjUS0zQb42tsrCIi22RAzaqS2J0bmoXQSn5w8gK/fB9p9bdW78jefv0bTegOkHwwGiK4QLBZYlzxx5QePgODKz9YlTzB5kIInZ/CWQmqVwm92umYdLjR6v//axIhqjkRdMRgg6iB+7Fj0e/45RFs7r5tGW63o9/xzrDNAoVG6FLHU9f0vtsL7bMQV3a517WyQifnn+qDE74EJhEQe4seORdyYMWg9+Bnaz5xBdJ8+iBk1kjMCFDqlkvjEnQj/qAVirgNaz/o4ULhy/9/9P5/MLYViA57W1lZce+21kh5D6mltbQWAkLobMhgg8kKwWLh9kJQnliJuroX3K3XBdb+/JD5vOxG8urLUNWwK8MlLgc9NRmMki8WChIQENDS4Wg3HxMRoWkefXJxOJ1pbW9HQ0ICEhARYQrhgYTBARKSVKItr++Bb0+EarL2UIvaXxOeznLAX8X1dz3VtorRgQOaWQpvNBgDugEA3nE7AfglwOICoKMDS3XdSsEkkJCS4fx/BYtEhA3La7Wgpr0Drp58CAGJuy0JsVhansYmMwmudgX6uwdtXEp/D7moo5HNG4MqSQP5q1xZCsYiR+3EBZiPmHQlqJ4Hdbsfly5dlP04Vxz8A9v0OaOkQoMQmAXctBG68O3znpaJu3bqFNCMgMkUw0Nbehq1/24rvm79HSnwKHsh4ANdEXxPu01JFc2kpap98Cg6P5hhRCQlIfnolE9xkcNrtzAug8JFbgfDkPleHwUBm7Oi69h+uxkhaUqgJU6QyfDCw/uB6vFH1BhxOh/u2KCEKMzJnoGhUURjPTHnNpaU4Pfcxv8f0e+F5BgQShKMzIYMPCkmo5YSDmY0wCimzJiHMfkQCQwcD6w+ux2tfvubz/l8P/rVpAgKn3Y5jd4+Bvd5/kk+0zYYb95TpYpDR6+Andibs0pDoyrpisFsI/b1ftkWmkIUyMyBSsh+Cnijx2UQ4wwYDbe1tGP3n0Z1mBDxFCVGo+GWFKZYMpDbRAfTRSEevg5/TbsfxMbldWhS7CQKirVbZAZW/9wtAleCDIozKa/+GpmATpkhl2KJDW/+21W8gAAAOpwNb/7ZVozNSl5zmOOFupCNeeXsOuO319Tj92Dw0l5aG6czgunL3FQgAgNOJ9ro6tB78TPJz+n2/cx9D7ZNP+W6LDKD+2dVw2tXvIU8GJ+5EAOBeB3eL8HLCOmzCZDSGDQa+b/5e0eP0Tk5znHA20nHa7ah/drVuBz+lOxNKeb+eyZ6ex8gNPiiCZU5yJcLFezQciu8b2QlyYv2GLkGSKDxNmIzEsHUGUuJTFD1O72JGjYTFapWUMxDORjpyrrzDsZShdGfCgO9XonDP5pCBZE5ylQ420tq/2rkKcuo3mDVvIkSGDQYeyHgAvzv4u4A5Aw9kPKDhWalHsFhgW7ok4G6CcDfSUfrKW2liZ0J/A7icgEqp98G2yFfwi1qaKItxEuG87mLo6xq8lZzJEGdNvL7WlR0TWp2LARk2GLgm+hrMyJzhdzfBjMwZpkgeFMWPHQu88LzXOgOWhATYdFBnQOkrb6UJFgviJ4xH46ubfR4TP2G85IAq5PdxJWGRbZHBL2oz8rX3v7nWdXuoSxuewePNE3zPmqh9LgZn2N0EokiqMyDScwVCd7Z+fb33dfQgs/UVP78AMwNSz0/K+43q1QuO8+evPMDZ6T6AuwkAsGCMGam9919O8Mg6BAEZPhgAIqsCoRG49/EDsgc/tWsTSN2iKWd7ppT3C0CXWy11gV/U5qTm3n+5wSPrEARk2GWCjq6JvgbTMqeF+zToivixY4Hnn+s6+Fmtfgc/LWoTqJHTIPX9si2yD9X7A3Tgc8pusUs6ILULooxuiQBcwWNJMbzXWnACEICSxa7lAjF4VOtcTMQUwQDpT/zYsbIGP19VAcXaBFBoKl2tnAYp75dtkX3gF3Xo9Jh4qdbe/2CCR9YhCIjBAKlG6uAXcK++IKD+2dWIGzMm5Ctp926CADkNwST0cbAPEr+oQ6PXxEtx73+gion+9v57C3KCCR7lnosegyuVGbboEJmHGlUBfREsFneJ4C49zq/8HO7tmRGHBWOCJ66de14pixnyVe+G57yA0CsmVr3ryiV542euUsNv/Mz189kT0l6/Y/Ao51x8vW44P0sNMBigsNO6NkH82LHo9/xziLZ2vtKMtlqZ2R8OLLMbnIBr53CtnTvCWOo62IqJ/oKcvauBaxMhO3iUci56Dq5UZordBGRsamT4S6HXrooRy8wtdtVgpAx5OdPuUnaXXJsIXGiEz2qD/gINX+cS4btamDNAYafmOr4/XOPXGSOW2Q0nvSReShno5VRMlJIgeKER+OkSoPJ139UGffF1LhG+q4XBAIWduI5/+rF5rnV7L3v1Q1nH5wyAgRipzG646SHxUo3kRanBy3UDgXlH5QePvoIXvQRXYcJggHQh2NoEgShdu4CBhQYiMJM7KEpk64dCrfK+coIcucGjv+BFD8FVGDFngHRFycHWV+2CYMsAa1EUKeLpdZucXrkHZED22jkQfOCl5vq6+7kDBDlynztQ1cL7Xwd2PaH86xoEgwEypYA9CGT2SFA6sCAv2J8gOMEmXoYSeKmdvBhqkONJavAy9lngr/+s3OsaCLcWkikpWbsgYFEkuPoOOO1h3MJldEbYJqdXmZNca+czdgAFr7r+Oe9I4EAglC10aq+vB7sl0RepyYGx1yn7ugbCnAEyJSVrF8gJLPSyO8HusKOyoRJnWs+gT0wfjEgaAYuepzalfll/9zGQ9k+anZZhyFk7D6a2vyct1teV3F0iJ3gZOjkid7UwGCBTUrIHgdZFkXyROsCXVZdhTfka1Lde/QK0xlixOGsxcvvnqnqOQZP6Zf1f04GJL5r6Ck11Smyh0yp5UandJXKDlwjc1cJlAjIlsXZBl5LDIkFAtM0mqXaBWs2N5CirLkPe23mYuWsmivcVY+aumch7Ow9l1WVdjivaW9QpEACAhtYGFO0t6nK8bkj9sr7QZPpKcKpTYorfaFUjWfI6IAYDZEpK9iBQMrAIhtQB3u6wY035Gji9XKmJt60tXwu7HtfdA35Ze2D+QPCUmuJXel1fTUYLXsKAwQCZllI9CMLZ3EjOAF/ZUNklYPA8vq61DpUNlYqfZ8g6fVkH0mEam+RT8io5mOTFcDFS8BIGzBkgU4sfOxZxY8aEXLtAraJIgcgZ4M+0SstZkHqc5sQv6/fmAhfOBT7epJXgVCcGXm9Nh8/a/nKuko20vs6S1z4xGCDTU6oHgVKBhRxyBvg+MdJyFqQeFxaZk4Du8cAffx74WJNWgtOEGHh5rTNg8sZQRgpeNMRggEgGrZsbyRngRySNgDXGiobWBq/LCgIEWGOsGJE0QunTVFbqXeEts6u1cJVf5lUydcBggEjH5AzwligLFmctRtHeIggQOh0vXJn+Lc4q1ne9AUD5aWw9C3f5ZV4l0xVMICTSMXGAB64O6CJvA3xu/1ys/+l6JMUkdTrWGmPF+p+u12+dAU+RkOwVahVAIgWxNwGRAXgrJGSLsaE4q9jrAG+4CoS+mLWDoZqNfoiCwGCAyCBMM8CT+o1+iGRizgCRQViiLBhtGx3u0yAlqN3oh0gmBgOkGV7ZEl2hRaMfIhkYDJBPSg7ehmyeQxQKf/kOWjX6IZKIOQPklZKDt1hb33NrnJgNr0SWO2cdSFeObgd2FgGtZ6/e5rllUNxNAMDr9kmz7JogQ2AwQF0oOXjbHXbkvZ3ns6SuuE++pKCEsw5kDqXLgf0v+LhT6DzIe60z0M/8VQBJdxgMUCdKD94VdRWYuWtmwOM2520OKjlOi1kHIsm+3A781wz/x8T367xl0KzbJ8lQWHSIOlG6852azXMM3bKXzMdhB95fEPg4z46LYhXAoZNd/2QgQGHAYIA6UXrwVrN5jqFb9pL5VO8HWv8u7VhuGSSdYTBAnSg9eIu19T1L6YoECLDF2IJqnmP4lr1kLnIGeG4ZJJ1hMECdKD14y62tL4cpWvaSeUgd4GOu55ZB0h0GA9SJGoO3Ws1z1Jx10ILdYUdFXQV2frsTFXUVzG0wOrF2QCDjf8+8ANId7iYgr+Q2xpEimFoAgR4j7iYA4LVlr153E3A7pEm5awf4+Fq9Yy4w9t80PSUiKRgMkE9KFvIJ5rmkDphqBC5q4nZIk/NWOyDmemDC74HB94bttIj8YTBAqgvmKljugGmUCoRaFGEiHWDtADIYBgMUUCgDbTBXwWYeMNUuwkQycdAmAsBGRRRAKGvbgYoCCRCwtnwtclJyOg3qcuoHGG3A5HZIHfFaCtijfwBRhOBuAvJJvKr3HJgbWhtQtLcIZdVlfh8fbFEgvQ6YTrsdLZ+W4/yO99HyaTmcdvnZ/9wOqRNiol/HQABwdRF8a7rrfqIIwpkB8iqYq3rP5YT6FmlFWDwHdT0OmM2lpah/djXa6+rct0XbbLAueQLxY8dKfh5xO2RDa4PXz1ZcAtHrdkhTcNhdMwJeM/6dAASgZDFw8wQuGVDEME0wIHVd22m3o/XgZ2g/cwbRffogZtRICBb+B+9J7lS9t+WExO6Jkl7Lc1DX24DZXFqK04/NAzzSa9rr6123P/+c5IBArONQtLcIAgSv2yGDLcJEElXv7zoj0Inzav+A1Ls0Oy2icDJFMCB1XVupq7tIIGeq3leS4LlL5/w+1tegrqcB02m3o/7Z1V0CAdedTkAQUP/sasSNGSM5qBSLMHn7m9XrdkhTkVo2OML7BzgcTtQea0JL8yXExndHcnoCoqK8F/gi4zN8MOBrIBLXtcVsdSWv7tSkly1yUqfge/fojWX/u8zrFXxHcgd1vQyYrQc/6xQ8duF0or2uDq0HP0PsbVmSnze3fy5yUnJ08buOOFLLBkdw/4AThxqwb+sxtDRdct8Wm9Addz2QjoHDk/w8kozK0MGA1HXtn/b9P4pf3alBT1XppE7VC4LgdzlBlNA9odNMgZRBXQ8DZvsZaTMkUo/ryBJl0d1uiIhYRhPLBjfXwnvegOC6P0L7B5w41ICSl492ub2l6RJKXj6K/N8MYUBgQoYOBqSua39RtgXXqnB1J4XUK32pMxxakTpVf/bCWUnP9/jox2GNtcoe1MM9YH7uqIGUOZLoPsbP/o+YZbQoi2v74FvTAQjoHBBcmQbPXxORyYMOhxP7th7ze8zHbx1D6i19uGRgMobeWih1XfsftTWSjgvm6s6fsuoy5L2dh5m7ZqJ4XzFm7pqJvLfzumzJCzTDAQBry9dq3shGSoMhqcsJ1lgrRttGY3zaeIy2jTbEdLjdYce/Xfgr/h4HOHwc44RrwIwZNVLLU5NM6nZIcRnNc0lEXEZrLi3V4nS1kzkJmPImEJ/c+fb4vq7bI7TOQO2xpk5LA978eO4Sao81aXNCpBlDzwxIHYjikm+QdJySV3dyrvT1XGQn0FS93jL/lVTZUIm6iw14/Z4oLHjHAQc6R88OuK4j//HI/bqcSpd6pa9GkqQhZE5ybR9kBUK3lmb/gYDc48g4DD0zILWF7bDcXyDaZgMEH9NagqDo1Z3cK329FtkRiVP13q7q1Wh5rBfi511+UxR+f18UGuM6398YB/z+vijUjRqgyuuHUuRIzpW+nCRJ04myuLYPDp3s+qcB/06VFBvfXdHjyDgMPTMgdV07uts1sC55wrVrQBA6XwFdCRCsS55Q7KpH7pW+HovsyKGXzH+ldfy8y2+KQkW6gEHfO5H4I3CuJ/BVigBnlICHVPi9hLJ+L/dKX80kSTKW5PQExCZ097tU0DPRtc2QzMXQwQAgfSCKHzsWeP65rl+wVmtQCVL+EgPlXumbYapdD5n/SvP8vTijBFT1vzr74Zp5Uv73Euo2WLnbIaUuj3k7LiJ2H0SQqCgBdz2Q7nU3gejOKelMHjQhwwcDgPSBKH7sWMSNGRPyl1egLYByr/T1VGQnFOHO/FdaOH4vSqzfy73Sjxk1EtE2G9rr672/riAg2mrtsowWMbsPIszA4UnI/82QLnUGeiZ2x51TWGfArNjCWCYpLXlzUnKQ93ZewCt9zxa83oIMW4zN0FPtZqDl76Xl03LUzJgR8Lgb3njD5zbYYJ7DPRsBeF1G6+cxG+Fr9sLX8WQ8rEAYWRgMyGB32JH3dp7PfICOg/yH33+Ior1FAOD1itJX3QC9VCCkzrT6vZzf8T5+WLgw4HF9f/c79PrZBK/3Oe12HB+TG/BK/8Y9ZZ1mF+TsPjg+Jtf3UoSP5yci/TLFMoFW5CQGBptUF+xUO4MIdWm1BBLK+r1IsFiCSpiVuoymVolmIgofBgMyyE0MlJrLEOpArqcyxhSaYNfvPQWbMCtYLAEHcO4+IDIfBgOQPhgHswUw0BVlqAO53soYU2iCvar3RqmEWU9KzF4Qkb5EfM6AnMFYzBmQmxjo77UDJSP6G8jl5DBwycBY9JypH2xOAhHpV0QHA8EMxuJjAHmJgZ6UGMgr6iowc9fMgK+1OW+zqbb8mZ177359PdobG2Hp3RvdriwN6GVwlbv7gIj0zdDliEMRbHMgKc17pJCTjOiL3ssYk3zNpaU4PiYXNTNm4IfHH0fDmjU4s3497OebdBMIAK4liH7PP4doq7XT7dFWKwMBIgOK2JyBUJoDKVFtT4mB3OhljENltup3oVYeVIO/z1itnAQi0p5ugwG1t8qFOhiHutVMiYHcDGWMg6XnNfVg6LFzoJTPWMruAyLSP10uE5RVlyHv7TzM3DUTxfuKMXPXTOS9nYey6jLFXkONq2q7w46Kugrs/HYnKuoquiwxdCS146K/gdzMHQP9kdORzyj01jnQjJ8xEfmmu2BATNDznMIXt8opFRAoMRh3JDeAUWogVyqHwSgCXkEDqH92tax2v3qgp737Zv2Micg3XQUDwSb1ybkiFyl5VR1sAKPUQJ7bPxe7CnZhc95mrL1rLTbnbUZJQYnpAgFAf1fQStHT3n2zfsZE5JuucgaCSeoLpWhPsCWDOwoUwAgQsLZ8LXJScrwGFkq1/jVbx0Bf9HQFraT2c41AVBTgcHg/QGLlQUXOxaSfMRH5pqtgQG5SnxLV90IdjEPZlSDqOJCzx4B/erqCVkpzaSl+mF/kfVq+A6mVB0Nlxs9YCrPtTiGSQ1fBgJykvlCvyDsK5apayb3+7DEQWLC1+/X6Re93fV4UFYV+63+v2S4JpfojGInZdqcQyaWrnAE5SX1KFO1RglK7ErRKnDQ6sXa/6wePvxMftfs7FfJZuBA1M2bg+JhcXWTEB1yfBwCHA5bE3tqcEIL7jI2MOyeIdBYMyEnq00v1PSV2JQSbOBmp5FS/0/sXvV7X5yOlwiB3ThC56GqZAJCe1KeX6ntiAFO0twgCBK/9CgLtSlAi70DvlJ6ml1L9To+FfDzpeX3eLBUG/f3tydk5weJKZGa6CwYAaUl9eqq+F+quBL3McqhFrfXYQNXv9PRF72tA0vv6vNErDAb625M8M1O1D4ipAXpagf53AEzqJZPRZTAABE7qU+KKXEmh7ErQyyyHGsJZb18vU/CBBiTrkidcn4UgeO0AaKT1eT3thpHytyd5ZqZ8DVDd5vohvi+QvxbInKTsCROFkeFbGHvLwLfF2GTVCQj3l5fYzjjQLIe/dsZ65O577+vqXOW+9y2flqNmxoyAx93wxhuqXf36GpA8W/2aIZtdT7thpP7tDSzdhRNj83zPzMCJ6Bg7bvxZAwR3htWV/KApbzIgINMwfDAABD+g6+nLS9xNAMDrLIcRSwuHezB2Dwg+v+gBREWh7/rfo1d+vnqvLzEY0uv2Ryl81fwI19+vnL89+/kmV8AGePk7caJf9jnEp1z0uF1wzRDMO8IlAzIFXe0mCJa4pDA+bTxG20arWkJYLWbsMRDuafpOW+R8cTjww/wiVXYVyC3rK67P9/rZBMTelmWYQECPu2Hk/O353DkR0+4jEAAAJ9B8Gqjer8DZEoWfbnMG1KRkwSIlKVWaWC/0kCkfP3YssGE9Thct8F3qF1BlV0G4gyGt6HE3jNy/vS47J36sQkzVMx2WBnz40ff7JjKSiAwG9PjlJTJTjwG9ZMpbEnv7DQTU2lWgh2BIC3rcDRPM316nnRMn44GvJbxQT2vgY4gMwBTLBHLp8cvLjPRSyS5cV+jigNTlvYsEAdE2m+HL+upxN0zIf3v973DlBPgoJubKGejnOo7IBCIyGNDjl5dZ6aGSXbiu0PUSDKlNiSqcagjpby/K4to+CKBrQHDl5/w1TB4k0zDFbgK5zLqVT8/CmSkfcFeBylsczbBtMBA974YJ6W+v6l2gpBho/uHqbfH9XIEAtxWSiURkMADo+8uLlOfe7w94Leyj9iyFkbcNShVqzQ/dcthduwZ+rGcFQjKtiA0GABN/eZFXUq/QI2HgVoseingRkXwRHQwA/PKKNIEG+kiY0ici8hTxwQCRSGrpYCLSht3hRPnJRjT84yKS4nogK7U3LFG+dnhQKBgMECH8fRSCxSUNMquSo7VY+V4Vas9frQCZ3KsHnpqYifwhyWE8M3NiMEAEZfsoaLX0xCUNaRgwGU/J0Vo8/KfKLnu9xDmBTb8awYBAYRFZgZDCT2+5GkoVJtKq+VU4W0MbCQMm47E7nFj5XpWXTd+AE66AYOV7Vbgn08YlAwVFZNEhCq+y6jLkvZ2HmbtmonhfMWbumom8t/M0bw7VkRKFibRqfuW021H/7GrvNROu3Fb/7Go47do1BtIjMWDyXPoRAyY1mlNR6MpPNnZaGvDkBFB7/iLKTzZqd1IRgMEAyea029HyaTnO73gfLZ+Wyxp09NYtUhRq6WAtO/fJ7YYYiRgwGVfDP3wHAsEcR9IwGCBZmktLcXxMLmpmzMAPCxeiZsYMHB+TK+kqS4+tbkWhlg6W0/wqVEosadgddlTUVWDntztRUVcRls9cTQyYjCsproeix5E0DAZIslCnXbUcMIMRSi17LZtfhbqkocdlGqVFSvtoM8pK7Y3kXj38tYhCci/XNkNSDhMITUrpBL2A066CgPpnVyNuzBifV89yB8xwJBl26WsvMftcy+ZXobSGFpdpPGdnxGUas5ThjpT20WZkiRLw1MRMPPynSghAp79U8edxQ2woP9nIugMK4tZCE1Ijo12JrXcVdRWYuWtmwOfYnLcZ5y+d1yQrXylaN78SZ2mcTmfnKyg/BZLEc/Q1O6NFgy6tArxwN6ei0HmrMxAlAI4Ov07WHVAOlwlMRq0EPSWmXaW2uj138Zwukwz9sURZsDhrMQB0eX/iz8VZxYoNfOU3ReH/PZCIs3Gdb2+/vpfPJY1wL9NouTwRKe2jzSx/SDI+Lr4bf5n9E/xL9gAAnQMBAKg7fxEP/6kSJUdrtT9Bk2EwYCJqJugpMe0qZcBcNHoRflvxW10mGQaS2z8X63+6HkkxSZ1ut8ZYFZ1+FwO+0tRmFD5iwYpfRuH5SVFY+UsLfjXzR5Tf5P0/ay3zGjyFYxdJKDkgpA+WKAFZqb2x86j3ZFDxW2Lle1Wwe0YKJAtzBkxEzpXfaNtoWc8dyjp1R+KA6W0JoDirGL2691LtPWght38uclJyVJsK9wz4nFECqvpfDawECFhbvhY5KTldXlPLvIaOAgWp/s45VMHmgJB+yKk7cPvA67Q7MZNhMGAial75idOupx+b55pm7RgQyJx29Tdg7vx2p2rvQSuWKItqgUooAZ+4TBMor2FE0gjdnLMSBIslYAlpkiYcSb1S6wn8z5WlAiYVBofBgImofeUXP3Ys8PxzXcu7Wq2yy7v6GjDDdfVqFKEEfOIyTdHeIggQOgUEauQ1+DuXUI6j8JCTmKxkt0Gp9QTePFCNNw9UM6kwSAwGTESLKz+1p13DdfVqFKEGS4GWadTYqcEAz/jkbElVutugWHeg7vxFr/0KPIlJhWxmJA+3FpqM+B8tAK9XfkbYR26G96AWpbYwajndq/W2S1KWnC2pu6saVOk2KHYxBCApIBAA2Hr1wMfFd3PJQCLuJjAZrTLa1WSG96AWpbYwiss049PGY7RttKqDsNbbLklZUnM+Kuo+89ttEAg+6z9/SDI2/WoEbL2kLRmwmZF8nBkwqXC1CFbydcPd5jjcr++Pt/VbW4xNtal+JRjxnAnY+e1OFO8rDnjcrJuWY8P22IDH/WX2T4LO+hdzEf7naC3ePFAd8Pjnf3Erfn5rv6BeK9IwZ8Ck1Mxo90XpyofiexAH5V3f7dJsUFajiqOS1N7CqAYjnjNJz+VwtPeElEn8ULoNWqIEdyAhJRhgMyPpODNAivCVYBTqOn84BmVf70VUeGshZg+dzUGMIoLUnI8Vw/8TU/9fRcDnC2Vm4Oo5OXHn2g98JhUyZ0A+5gxQyNSqfBiOqnX+3oto4+GNGPvXsbosi0ykNKk5Hz9J66NZt0GxmZH4vJ6vAwBPTcxkICADgwEKmRo179UsrexPoPciarig3z4JpA67w46Kugrs/HYnKuoqdFkSWy1Sknq1HqB9JRXaevXgtsIgMGeAQqZGUZlwVa2TW/hGrTK64aLnpMlwKqsuw+pPV6PhQoP7tqRrk/DEbU/oIodEC1JyPsQB2rPOgE2lQkD5Q5JxT6ZNsQJHkYzBAIVMjaIy4apaJ+cc9d4nQS69J02GS1l1Gebvnd/l9oYLDZi/dz42/HRDxHw+UhKTtR6gOyYVUvC4TEAhk9qaWE7VwHBVrQv0XrwxQxndcORnGIHdYceK/Sv8HrNy/8qIWjKQQhygf35rP9w+8DpeqRsAgwEKmRpFZdQIMKTo+F6kMnoZ3XDlZxjBwfqDON923u8xTW1NOFh/UKMzIlIHgwFShNJVA8NZtU58L9YYq9/j1ApIOtIiaU2NBFCzKK8tV/Q4Ir0yVM6A025nX3IdU7qoTDia6nR87ZyUHLzyxSvY+PnGLvdrUUZXqzV8dhX0Q+rsNmfBTUvJDox6ZphgoLm0tGvrXJtNdutcUpfSlQ/DWbXOEmXBQ7c+hBsTb9Q8IJHTJS5U7Cro22jraPwH/kPScWQ+Sndg1DNDVCBsLi3F6cfmAZ6nKriis37PP8eAgGSRu4UuHF3+pHSJU+Ic2FXQN7vDjn96659w/pLvvIGE7gnYO2VvxH02Zid2SlS6A6Ne6T4YcNrtOD4mt9OMQCeCgGirFTfuKeOSAUmi9y10FXUVmLlrZsDjNudtVmwWhm2jffO1tVAUSVsLI4VY7rjjjEBHZix3rPsEwtaDn/kOBADA6UR7XR1aygPXxNaK025Hy6flOL/jfbR8Wg6nPfKysPUqlC10WlWgC8caPttG+5bbPxcbfrqhS0KpNcbKQMCkyk82+gwEAHO2SNZ9zkD7GWlfeKfnzUPyvz0d9uUC5jboV6AtdAIEnxUFtZxNCNcaPrsK+sbPJrJI7awYSgdGvdH9zEB0H4ntM8+fx+nH5qG5tFTlM/JNzG3wnMlor68P+7lR8FvotC7IE64aC8DVBNDxaeMx2jaag10H/Gwih9TWx2Zqkaz7YCBm1EhE22zuZMFA6p9dreq0vK8lAKfdjvpnV3dNcgTct6l9buRfMNPv4SjIE84aC0QEZKX21qwDo17oPhgQLBZYlzwh7eAr+QOtBz9T5VyaS0txfEwuambMwA8LF6JmxgwcH5OL5tJSybkNap0bBRbM9Hu4CvJwDZ8ofKR2YASAAyfO4r8Pn8aBE2dhd+g6H98v3ecMAHCttT//HGqXPwnHef+lQQHpeQZy+NreKC4BJE6fJul51Di3cDBiAShx+j3QFrqO0+/hLMjDdWqi8AnUgRFAlx0HRq5BoNtgwHOwiRszBlFxcfj+14G3XEnNM5BzLn6XAAQB5999T9JzKX1u4WDUJElx+r1obxEECF630HlOv4e7II/SRZyISDpfHRh3V9V5rUFQd/4iHv5TpSFrEOiyzoCvwSZpcTEa1qxFe32994FZpZoDLZ+Wo2bGjIDHWRITYW9q0vTctGaGAlDedgbYYmxeKwqyIA8RdWTWGgS6yxnwl5H/w/wixE8Y77rBM6Hwys/WJU8oPthKndqPnzSp07m4qXhuWnLa7ah98inDJ0nm9s/FroJd2Jy3GWvvWovNeZtRUlDidR2eyXxE1JFZaxDoKhiQkpHf/P5O9NuwHtHWzgVAoq1W2VelUosDSZ3aj7v7bvR7/jlFzk2P/v7v/w5HU5PvAwyUJClnmxiT+YhIZNYaBLrKGZCakW9J7I0b95SFlMAmZ91b3N4YaHlCPIe4MWMMl1wXiNNuR+Obf5R0bPuZM4ZMMPSHyXxEBJi3BoGuggGp0/HtZ85AsFgQe1tWUK8TaGcAPK7ixe2Npx+b55ry7/g4L0sAoZybXrUe/EzSTg4AaKuu7tJPwggJhoEwmY+IxBoEdecveskicjFiDQJdLRNInY4PJSM/2OJA8WPHmnoJIBCpgZoQE4O//+EPEVuFUav+BUQUHh1rEPgy6ZZkQyUPAjqbGZAzHR8sOcWBPK/u48eONeUSgBRSAzAhKgpeN6hc2YJZ/+xqxI0ZY8rPTO/dEIlIGflDkvGv/ycVL3900uv9//HRSQy/IdFQ2wt1NTPQqdqgShn5cpYifJ1j7G1Z6PWzCYi9LcuUg5o3UspCC7GxcPz4o+8nMVCCoVxa9y9QA2c1iKSxO5x49/Nav8esfK/KUBUJdRUMAOpPx2uxFGFGfgO1KxImF0h6LrNUYRSFo3+B0sqqy5D3dh5m7pqJ4n3FmLlrJvLezjNEEEOkNTNuL9RdMAC4AoIb95ThhjfeQN/f/Q43vPEGbtxTpsi6fMArXEFAtM0W0lKEWfkM1Gw29HvhecTdPUbS85gt0ApX/wKlmGFWg0hLZtxeqKucgY7UysiXuzOAOvOXN+G021XP+dCjcPUvsDvsIW91DDSrIUDA2vK1yEnJ4TZKoivMuL1Qt8GAmsTGR13qDFitht/+pgVfgVqkBlrh6F+gVLKinFkNbqskcgm0vVAsSWyk7YW6XCbQgppLEZEsErdgit0QPcsViwQIsMXYOnVDDIWS0/rh7MpIZFRSWxwbaXuhLhsVkfGZrQJhIOIADcBrN0SlyhaLjZN8Xc3LbZxUUVeBmbsCdwLdnLeZMwNEHkqO1nZpcWzUNsYMBogUIqcbYrCUHrzZlZEoNHaHs0uLYyPNCIgiMmeASA1a9C9Qelpf7MpYtLcIAgSvsxrFWcUAXIEI+zIQdWaJEnD7wOvCfRohYzBApCC1+xeokawodmX0lpAoBgKeSxOhVFZUYhcEESmLywREBiJlWj8pJgmr7lyFsxfOyhpsvQ3SH37/IYr2FnV5rWBzIViymUifwhoMONracO7Pf0Hb99/jmpQUJP7yQURdc024TofIEPwlKzrhRK9reuF829UOk8EOtkonK4rnrVRgQUTKCVswUL9uHRpfex1wOK7eGBWF3r/+Z1gXLQrHKREZhrcr7ITuCWi61NTl2GAHWyWTFZUOLIhIWWHJGahftw6Nr27ueofD4b6dAQGRb57Jir179MbSj5d6PTbYSoJKJiuyuBGRvmledMjR1uaaEfCj8bXX4Whr0+aEiAxKTFYcnzYeligLGi40+Dw2mP4ISiYrsrgRkb5pHgyc+/NfOi8NeONwuI4jIknUGGzFyoqBnLt4LuAx4SjZTETSaR4MtH3/vaLHEZE6g60lyoLHRz8e8Lh1FesCtmfWumQzkZbsDicOnDiL/z58GgdOnIXdYbxNeprnDFyTkqLocUR0dbANVElQ7mCb2CMx4DFS1vqlFjdi8iAZjVlKEms+M5D4yweBqAAvGxXlOo6IJBEHWwBdrr5DGWyVXH4QixslxSR1ut0aY+W2QjKkkqO1ePhPlZ0CAQCoO38RD/+pEiVHa8N0ZvJpPjMQdc016P3rf/a+m+CK3r/+Z9YbIJIpUCXBYAZbpZcftCjZTKQFu8OJle9VeW1h7ISre+HK96pwT6bNEL0KwrK1UNw2yDoD6mHJ18ik9GCrxvKD2iWbibRQfrKxy4xAR04AtecvovxkoyF6F4StN4F10SL0eewxViBUAUu+ukRaG2WRkoNtpK/1M6gmXxr+4TsQCOa4cGNvApNhyVeX5tJS1D+7Gu11de7bom02WJc8gfixY8N4ZsakRXtmvWFQTf4cOHEWD77yScDj/jL7J4aYGWAwYCIs+erSXFqK04/NAzz/tAVXQNTv+ecYEAQhkq6SGVRTIHaHE3eu/QB15y96zRsQANh69cDHxXcbImdA890EpB45JV/Nymm3o/7Z1V0DAcB9W/2zq+G0+98XT111rHg42jbatIGA3WHHmvI1XnMkxNvWlq8NWFuBzM0SJeCpiZkA0KV6hvjzUxMzDREIAAwGTIUlX+HKEeiwNNCF04n2ujq0HvxMu5MiQ2FQTVLlD0nGpl+NgK1Xj06323r1wKZfjTBUnYGwJRCS8ljyFWg/Iy3QkXocRR4G1SRH/pBk3JNpQ/nJRjT84yKS4nogK7W3YWYERAwGTEStKnRGEt1HWqAj9TiKPFKD5bMXz8LusJt2uYSks0QJhkgS9IfLBCaiVhU6I4kZNRLRNps7WbALQUC0zYaYUSO1PTEyjEB9FES/rfgt8t7OQ1l1mUZnRqQeBgMmE+klXwWLBdYlT1z5wePL/MrP1iVPRES9AQqOv6DaU0NrA4r2FjEgIK+M1MCIWwtNKpK2gXnDOgMUKm91BnyxXmvFrsm7Iuq/MfLPaA2MGAzoSKQP4EozcgVCu8Np+IQkM7A77Pjz13/Gbyt+G/DYwlsL8dAtD2lwVqR3YgMjz8FV/C9YjzsNmECoE6x2pjzBYkHsbVnhPg3ZjHZFYWaWKAuu6yEtMWzj4Y24MeFG/vca4YzawIg5AzogVjvznI7kemTkMVNLVLOQsxWXxYhITgMjPWEwEGZt7W14+sDTrHamELvDjoq6Cuz8dicq6ioM9bkFuqIAXFcUaiUhGfmzU5O4u0AKFiMyL6nJgEZtYMRlgjAqqy7D0weexrlL53we07HaGdu++qeXpZZgcz/C2RJVL5+dHom7C+bvnS/peBYjMh85S3dJcT08H+6V1OO0wpmBMBGXBvwFAh3xC8Y/vSy1lFWXIe/tPMzcNRPF+4oxc9dMyXvRw3VFoZfPTs9y++ei8JZCSceaucJnJJK7dJeV2hvJvXr43JQqwBVIZKX2VueEg8RgIAz8NULxhV8wvumlsUyog2o4rij08tkZwexhs/0uFwgQYIuxmbrCZ6QJZunOqA2MGAyEQaBGKB3xCyYwPTSWUWJQDccVhR4+O6MQlwuEK//rKFIqfAbLSMV3OpK6dPfJt2c7vb97Mm2Ga2DEnIEwkDvlzy8Y//TQWEbOoOor90O8onj4T5UQgE5hhVpXFHr47IxErPDpLb+iOKs44vMrvDHyVlmpS3KF/1mJpguX3T+L7+/j4rsD1gvRS00RBgNhIHXKP7F7Ip68/Ul+wQSgh26NSg2qYktUzy9Pm0pfnnr47Iwmt38uclJyWCBMAl/Fd8T1dr1eJYukLsl1DAQA6e9PT4ESg4EwCNRdEAB69+iN3QW7cU30NRqfnfGcuxg4CVPtpRYlB1UtW6Ky02VwLFEW7u4JwKjFdzoSl+7qzl+UkeEl7f3pLVBizkAYBOouKEDA8p8sN0QgEO696XaHXVKp2EWjF6l65Rao053c3A+xJerPb+2H2wdep9qXJTtdklqMWnynI3/JgIH4e3/hriniDYOBMDFDd8FQttEpRWoyZmKPRFXPw8iDqhn+Fkl/jFp8x5O4dOeZDJgQ003S4729Pz0GSlwmCCMjrz2K2+g8p5bFbXRaDSJ6SoAzcnKZkf8WSZ+MWnzHG29Ldw6HE1Nf/TTgY729Pz0GSgwGwsyIa49St9HlpOSoPpjoLQHOyIOqEf8WSb8CrbcLcCXGStkqq4eMe3HpruM5Bfv+9BgoMRgg2aRMzWtVQllKMmbCNQmaJsBxUCVSbqusVhn3cgOOju/PU6D3p2SgpBTmDJBs9S3SCiZJPS4U4lq9v2qOTW1N+PD7D1U/FyLqzNd6e6+YbpiXm4F7Mm1+H69VF8+So7W4c+0HePCVT/DYlsN48JVPcOfaDyQ9fy8vuQMJMd387gbQY5VCBgMkm9R+ClKPC1VOSg56de/l834BAkvqEoVJ/pBkfFx8N+bnpiPhWtfA2dR6GRvK/uZ3wNUq4z7YgEN8XFPr5S73nfNymydfgVK4qhRymYBkS+wuLTNf6nGhqmyoxPlL533ez86PytPDGi5pQ4nf9e6qOjxXdkzWnnotungGWwvB3+Pg53GetKwpEgiDAZLNGiutt7vU40Klpx0FRib1S19PVdNIXUr8roMdcLXIuA824FAyUPFMTAwXBgMkm5i05y+JUMvmSnrbUWBEUr/09VY1jdSj1O862IFTi4z7YAMOPW4NDBVzBki2jt3bvBEgaFpgR+nqf5FG6pqpHqumkTqU/F1LHRD/9/iZTs+nRRfPYAMOPW4NDBWDAQqKWGDHs7+7LcamedU6I1f/Czc5X/p6rJpG6lDydy11QPzDhyc6JRRqkXEfbMARjnbjamMwQEHL7Z+LXQW7sDlvM9betRab8zajpKAkLJX2WFI3OHK+9M04NUreKfm7DjRwduQ5G6V2xn2wAYcetwaGijkDFBI9FdgxcvW/cJHzpW/GqVHyTsnftb/iQ568JRSqnXEfbNtwrduNq43BAJmKnoITI5Dzpa/HqmmkDqV/174GTm+8JRSqnXEfbMChp62BoeIyAVEEk7P2acapUfJOjd+1WHxoTs6Nko7Xerkp2LbhWrUbVxuDAaIIJuVL/8kJg1B3rAl/q6jD0G498NIvh+umahqpR431ekuUgOwbr5d0LJebtCU4nU7uAyKKcL7qDCzI/P/w44G/o6Xpkvv22ITuyL4/HX9PsBh+apQCU7rapN3hxJ1rPwi4BPFx8d38m9IQgwFShN1hZ+KewXl+6V/fZEfpK0d9Hp//myEYODzJ5/1Evoi1LQDv3Qw5y6Q9BgMUsrLqMqwpX9OpIqE1xorFWYu5pc+gHA4n3lyyv9OMgKeeid0xbdUdiOLVGwWBZa31hcEAhaSsugxFe4u6tBAWi/1wj78xnf7mHLZvOBTwuHvnD0e/m7RpSEXmw4ZX+sGthRQ0u8OONeVrugQCgKtToNg6OCclh0sGBtPS7HtGIJjjiLzRS5Me4m4CCkFlQ6XfZkUdWweTscTGd1f0OCLSNwYDFDS2Djav5PQExCb4H+h7JnZHcnqCNidERKpiMEBBY+tg84qKEnDXA+l+j7lzSjqTB4lMgsEABY2tg81t4PAk5P9mSJcZgp6J3bmtkMhkuJuAQiLuJgDQKZGQuwnMw+FwovZYE1qaLyE23rU0wBkBInNhMEAh81ZnwBZjQ3FWMQMBIiIDYDBAimAFQiIi42IwQEREFOGYQEhERBThGAwQERFFOAYDREREEY7BABERUYRjMEBERBThGAwQERFFOAYDREREEY7BABERUYSLDvcJENFVdocT5Scb0fCPi0iK64Gs1N6wsA8AEamMwQAFjSWIlVVytBYr36tC7fmL7tuSe/XAUxMzkT8kOYxnFjoGOUT6xnLEFBRvzYmsMVYszlrM5kRB2PlFLR75c2WX28XhctOvRhg2IDBzkENkFgwGSDaxbXHHlsUA2xYHa+cXP2DOXw7B4eO/RAGArVcPfFx8t+GupkuO1uLhP1XC862ZIcghMhMmEJIsbe1tePrA010CAQDu29aWr4XdYdf61Ayp5GgtHvmz70AAAJwAas9fRPnJRs3OSwl2hxMr36vy8pcC920r36uC3d+bJyJNMGfAoDqu1/fu0RuCIODshbOqrt2XVZfh6QNP49ylcz6PccKJutY6VDZUYrRttOLnYCbiYClVwz8uBj5IR8pPNnZaGvDUMci5feB12p0YEXXBYMCAvK3Xd6TG2r2vpQFfzrSeUey1zSrQYOkpKa6HimejPKnBi9GCHCIz4jKBwYiDsq9AAAAaWhtQtLcIZdVlirym3WHHmvI1kgMBAOgT00eR1zYzOYNgci9XBr6RSA1ero/trvKZEFEgDAYMROqgrPTafWVDpd/goyMBAmwxNoxIGhHy65qdnCv9pyZmGi55MCu1N5J79UCgs350yyHs/KJWk3MiIu8YDBiInEG549p9qORO+RdnFbPegARSBssoAXjpl8bMuLdECXhqYiYA+H2PjS1teOTPlVi9U3r+BBEpi8GAgQSzDq/E2r3UKf/E7oncViiDlMHyDw8Ox/hhxgsERPlDkrHpVyNgjQ+8FPDyRyex84sfNDgrIvLEYMBAglmHV2LtfkTSCFhjrO46At707tEbZZPLGAjIJA6Wtl6dlwySe/XAv/9qBMYP6xumM1NO/pBk/H7KrZKOXfbfR7nVkCgMuJvAQMRBuaG1IWDegAAB1hirImv3ligLFmctRtHeIggQOr22GCAs/8lyXBN9TcivFYnyhyTjnkybqcv1/v3HS5KOa2y5zK2GRGHAmQEDEQdlAH6v0sX7lFy7z+2fi/U/XY+kmKROt1tjrFwaUIAlSsDtA6/Dz2/th9sHXmeqQACQlyzJrYZE2mM5YgMKVGfAFmNDcVaxKgM0mxNRMOwOJ0avKkNjS1vAY/8y+yecGSDSGIMBgwpHBUKiUPhqxtRRskF7MBAZHYMBItLM6p1VePmjk17vE8DGRUThwmCAiDS184sfsOy/j6Kx5bL7NrY0JgovBgNEpDm7w2nq3RNERsNggIiIKMJxayEREVGEYzBAREQU4RgMEBERRTgGA0RERBGOwQAREVGEYzBAREQU4RgMEBERRTgGA0RERBGOwQAREVGEYzBAREQU4RgMEBERRTgGA0RERBGOwQAREVGEYzBAREQU4RgMEBERRTgGA0RERBGOwQAREVGEYzBAREQU4RgMEBERRTgGA0RERBHu/wdUBS9nBcB+nAAAAABJRU5ErkJggg==\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "data_70[\"topics\"][0].sum()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ZEBr2jhbTpSl",
        "outputId": "0b0d54c0-89e7-4706-cbaf-97623b0b9a9b"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0.45757685861297015"
            ]
          },
          "metadata": {},
          "execution_count": 49
        }
      ]
    }
  ]
}