imdb_binary_classification.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "imdb-binary-classification.ipynb",
      "version": "0.3.2",
      "provenance": [],
      "collapsed_sections": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "[View in Colaboratory](https://colab.research.google.com/github/michalMalujdy/machine-learning-colab/blob/master/imdb_binary_classification.ipynb)"
      ]
    },
    {
      "metadata": {
        "id": "WqU9lOWV67wN",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "# Binary classification with Keras\n",
        "Learning binary classification of film reviews (positive or negative) from IMDB using Keras framework"
      ]
    },
    {
      "metadata": {
        "id": "kdQHSHqH5-Ps",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "from keras.datasets import imdb\n",
        "\n",
        "(train_data, train_labels), (test_data, test_labels) = imdb.load_data(num_words=10000)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "w-QcKMVv66tX",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "\n",
        "\n",
        "---\n",
        "\n"
      ]
    },
    {
      "metadata": {
        "id": "W66wRcI-qWds",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "Function preprocessing a single review by transforming a list of integers to numpy one demensional array (a vector) where all values are zeros except for the indexes that corresponds to the list numbers (keys for words), where the value is set to 1."
      ]
    },
    {
      "metadata": {
        "id": "DxDHXxHfqWjy",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "import numpy as np\n",
        "\n",
        "def preprocessReview(review):\n",
        "  preprocessedVector = np.zeros(10000)\n",
        "  \n",
        "  for index in review:\n",
        "    preprocessedVector[index] = 1\n",
        "  \n",
        "  return preprocessedVector"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "OgAodAtNqWpt",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "\n",
        "\n",
        "---\n",
        "\n"
      ]
    },
    {
      "metadata": {
        "id": "BtVTo2rHyoaz",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "Function that does the trick mentioned above for all reviews in passed list of lists, so the return type is numpy array."
      ]
    },
    {
      "metadata": {
        "id": "yOnFCqUGyoi3",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "def preprocessReviews(reviews):\n",
        "\n",
        "  input_array = np.empty([len(train_data), 10000])\n",
        "\n",
        "  for i, review in enumerate(reviews):\n",
        "    input_array[i] = preprocessReview(review)\n",
        "    \n",
        "  return input_array"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "ELvXb3EDyopX",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "\n",
        "\n",
        "---\n",
        "\n"
      ]
    },
    {
      "metadata": {
        "id": "aa3OneEhslAG",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "Preprocess input data and labels data. The latter needs to be float32 numpay array, so the transformation is simple from int to float. But the input is preprocessed by adequate functions stated previously."
      ]
    },
    {
      "metadata": {
        "id": "LBTboCrNslGM",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "input_train = preprocessReviews(train_data)\n",
        "input_test = preprocessReviews(test_data)\n",
        "\n",
        "train_labels = np.array(train_labels).astype('float32')\n",
        "test_labels = np.array(test_labels).astype('float32')"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "KP3NYJpXslSW",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "\n",
        "\n",
        "---\n",
        "\n"
      ]
    },
    {
      "metadata": {
        "id": "0jttZS_QtCn7",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "Divide the train data into two sets: the actual training set and a validation set used during the training (**do not mistake with test set that is used AFTER the training**)"
      ]
    },
    {
      "metadata": {
        "id": "3yzPUBJ-tCx4",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "validation_set_size = 10000\n",
        "\n",
        "input_train_validation = input_train[:validation_set_size]\n",
        "input_train_partial = input_train[validation_set_size:]\n",
        "\n",
        "train_labels_validation = train_labels[:validation_set_size]\n",
        "train_labels_partial = train_labels[validation_set_size:]"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "ekCeDUGhtC7d",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "\n",
        "\n",
        "---\n",
        "\n"
      ]
    },
    {
      "metadata": {
        "id": "QtcJZO0V0Nwo",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "Create the network model"
      ]
    },
    {
      "metadata": {
        "id": "13CESLET0N18",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "from keras.models import Sequential\n",
        "from keras.layers import Dense, Activation\n",
        "\n",
        "net = Sequential()\n",
        "net.add(Dense(4, activation = 'sigmoid', input_dim = 10000))\n",
        "net.add(Dense(6, activation = 'sigmoid'))\n",
        "net.add(Dense(4, activation = 'sigmoid'))\n",
        "net.add(Dense(1, activation = 'sigmoid'))\n",
        "\n",
        "epochs_count = 25;\n",
        "\n",
        "net.compile(optimizer = 'rmsprop',\n",
        "           loss = 'binary_crossentropy',\n",
        "           metrics = ['accuracy'])"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "DZ_CHD6W0N8d",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "\n",
        "\n",
        "---\n",
        "\n"
      ]
    },
    {
      "metadata": {
        "id": "CTuZWx4g3p5S",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "Actual network learning."
      ]
    },
    {
      "metadata": {
        "id": "D_Y1ah-63p9R",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 884
        },
        "outputId": "def000c4-2a1b-4da6-a853-b5046d3429d7"
      },
      "cell_type": "code",
      "source": [
        "training_history = net.fit(\n",
        "    input_train_partial, \n",
        "    train_labels_partial, \n",
        "    epochs = epochs_count, \n",
        "    batch_size = 512,\n",
        "    validation_data=(input_train_validation, train_labels_validation))"
      ],
      "execution_count": 124,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train on 15000 samples, validate on 10000 samples\n",
            "Epoch 1/25\n",
            "15000/15000 [==============================] - 2s 134us/step - loss: 0.7093 - acc: 0.4965 - val_loss: 0.6999 - val_acc: 0.5053\n",
            "Epoch 2/25\n",
            "15000/15000 [==============================] - 1s 85us/step - loss: 0.6974 - acc: 0.4965 - val_loss: 0.6913 - val_acc: 0.5053\n",
            "Epoch 3/25\n",
            "15000/15000 [==============================] - 1s 87us/step - loss: 0.6890 - acc: 0.4965 - val_loss: 0.6853 - val_acc: 0.5053\n",
            "Epoch 4/25\n",
            "15000/15000 [==============================] - 1s 87us/step - loss: 0.6830 - acc: 0.4979 - val_loss: 0.6806 - val_acc: 0.5361\n",
            "Epoch 5/25\n",
            "15000/15000 [==============================] - 1s 87us/step - loss: 0.6772 - acc: 0.7222 - val_loss: 0.6750 - val_acc: 0.8165\n",
            "Epoch 6/25\n",
            "15000/15000 [==============================] - 1s 85us/step - loss: 0.6701 - acc: 0.8639 - val_loss: 0.6680 - val_acc: 0.8584\n",
            "Epoch 7/25\n",
            "15000/15000 [==============================] - 1s 87us/step - loss: 0.6617 - acc: 0.8909 - val_loss: 0.6597 - val_acc: 0.8587\n",
            "Epoch 8/25\n",
            "15000/15000 [==============================] - 1s 86us/step - loss: 0.6516 - acc: 0.8957 - val_loss: 0.6499 - val_acc: 0.8686\n",
            "Epoch 9/25\n",
            "14848/15000 [============================>.] - ETA: 0s - loss: 0.6401 - acc: 0.9028"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "15000/15000 [==============================] - 1s 85us/step - loss: 0.6400 - acc: 0.9033 - val_loss: 0.6388 - val_acc: 0.8701\n",
            "Epoch 10/25\n",
            "15000/15000 [==============================] - 1s 86us/step - loss: 0.6271 - acc: 0.9069 - val_loss: 0.6267 - val_acc: 0.8762\n",
            "Epoch 11/25\n",
            "15000/15000 [==============================] - 1s 86us/step - loss: 0.6127 - acc: 0.9135 - val_loss: 0.6132 - val_acc: 0.8757\n",
            "Epoch 12/25\n",
            "15000/15000 [==============================] - 1s 87us/step - loss: 0.5967 - acc: 0.9169 - val_loss: 0.5985 - val_acc: 0.8800\n",
            "Epoch 13/25\n",
            "15000/15000 [==============================] - 1s 87us/step - loss: 0.5795 - acc: 0.9211 - val_loss: 0.5828 - val_acc: 0.8819\n",
            "Epoch 14/25\n",
            "15000/15000 [==============================] - 1s 86us/step - loss: 0.5609 - acc: 0.9248 - val_loss: 0.5662 - val_acc: 0.8838\n",
            "Epoch 15/25\n",
            "15000/15000 [==============================] - 1s 88us/step - loss: 0.5413 - acc: 0.9278 - val_loss: 0.5491 - val_acc: 0.8854\n",
            "Epoch 16/25\n",
            "15000/15000 [==============================] - 1s 85us/step - loss: 0.5211 - acc: 0.9317 - val_loss: 0.5318 - val_acc: 0.8869\n",
            "Epoch 17/25\n",
            "15000/15000 [==============================] - 1s 85us/step - loss: 0.5003 - acc: 0.9343 - val_loss: 0.5140 - val_acc: 0.8881\n",
            "Epoch 18/25\n",
            " 5632/15000 [==========>...................] - ETA: 0s - loss: 0.4856 - acc: 0.9343"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "15000/15000 [==============================] - 1s 86us/step - loss: 0.4790 - acc: 0.9361 - val_loss: 0.4961 - val_acc: 0.8905\n",
            "Epoch 19/25\n",
            "15000/15000 [==============================] - 1s 87us/step - loss: 0.4575 - acc: 0.9393 - val_loss: 0.4784 - val_acc: 0.8911\n",
            "Epoch 20/25\n",
            "15000/15000 [==============================] - 1s 87us/step - loss: 0.4360 - acc: 0.9415 - val_loss: 0.4612 - val_acc: 0.8899\n",
            "Epoch 21/25\n",
            "15000/15000 [==============================] - 1s 86us/step - loss: 0.4146 - acc: 0.9457 - val_loss: 0.4441 - val_acc: 0.8902\n",
            "Epoch 22/25\n",
            "15000/15000 [==============================] - 1s 87us/step - loss: 0.3933 - acc: 0.9477 - val_loss: 0.4278 - val_acc: 0.8907\n",
            "Epoch 23/25\n",
            "15000/15000 [==============================] - 1s 87us/step - loss: 0.3727 - acc: 0.9505 - val_loss: 0.4124 - val_acc: 0.8914\n",
            "Epoch 24/25\n",
            "15000/15000 [==============================] - 1s 86us/step - loss: 0.3527 - acc: 0.9527 - val_loss: 0.3981 - val_acc: 0.8925\n",
            "Epoch 25/25\n",
            "15000/15000 [==============================] - 1s 86us/step - loss: 0.3333 - acc: 0.9551 - val_loss: 0.3844 - val_acc: 0.8923\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "V8wL1L2i3qBZ",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "\n",
        "\n",
        "---\n",
        "\n"
      ]
    },
    {
      "metadata": {
        "id": "up-kas066O_l",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "Display the accuraccy on test data"
      ]
    },
    {
      "metadata": {
        "id": "azG4vzqc6PFN",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        },
        "outputId": "0923bc6f-0b76-48d4-fa5a-0b8269e181c7"
      },
      "cell_type": "code",
      "source": [
        "(_, accuracy) = net.evaluate(input_test, test_labels)\n",
        "print(accuracy)"
      ],
      "execution_count": 125,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "25000/25000 [==============================] - 3s 133us/step\n",
            "0.88548\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "E4ozzKii6PKQ",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "\n",
        "\n",
        "---\n",
        "\n"
      ]
    },
    {
      "metadata": {
        "id": "AHF6OVn7ee-v",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "# Optional"
      ]
    },
    {
      "metadata": {
        "id": "Gp4Be7YE7uEB",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "Decode the review from a list of integers indicating the word in dictionary to a readable text. The words are not stored in a concise order, so the output may not be understandable."
      ]
    },
    {
      "metadata": {
        "id": "N14nVur57uoi",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 54
        },
        "outputId": "0a54f5a5-bf62-4cd3-d412-f397b3d3c516"
      },
      "cell_type": "code",
      "source": [
        "from random import randint\n",
        "\n",
        "word_index_dict = imdb.get_word_index()\n",
        "index_word_dict = {value: key for (key, value) in word_index_dict.items()}\n",
        "\n",
        "random_review_index = randint(0, len(test_data) - 1)\n",
        "review_ints_list = test_data[random_review_index];\n",
        "review_words_list = [index_word_dict[key] for key in review_ints_list]\n",
        "review_text = ' '.join(review_words_list)\n",
        "\n",
        "print(review_text)"
      ],
      "execution_count": 126,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "the as ever there in at only starts some br an of wild fact day worth just interpretation or of and other or of every spielberg as ever it is sam americans what have plain film we wonder rubbish as and was body in also of apparently some but numerous to also that some are year scene it of and tomorrow misguided and teenagers and and good br only starts rating story one in at to dozen third was other several of today replies film of apart whole for short in kay we some by br of seems was story one in air of year spielberg in adults story at of and totally says times new know of steven gay story at was then does what think br have start for lot not series by br of watched seems for total those heroine in we was let saw are black br of repeat of day original example there will scary are year her was why numerous to other interpretation or of and to false of resist of original real at scary movies bad of beginning kids in died comes emma bad kids as disturbing ever are year her wonder poetry of apparently\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "LgsWLK3pzVjp",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "\n",
        "\n",
        "---\n",
        "\n"
      ]
    },
    {
      "metadata": {
        "id": "BBLwPfWdqTud",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "Chart showing the relation between training loss and epochs and also validation loss and epochs"
      ]
    },
    {
      "metadata": {
        "id": "hwH7wLnVqT1N",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 376
        },
        "outputId": "2f3b15d9-47e6-464f-d1ec-beb1e8bf3b76"
      },
      "cell_type": "code",
      "source": [
        "import matplotlib.pyplot as plt\n",
        "\n",
        "history_dict = training_history.history\n",
        "loss_values = history_dict['loss']\n",
        "val_loss_values = history_dict['val_loss']\n",
        "epochs = range(1, epochs_count + 1)\n",
        "\n",
        "plt.plot(epochs, loss_values, 'bo', label='Training loss')\n",
        "plt.plot(epochs, val_loss_values, 'b', label='Validation loss')\n",
        "plt.title('Training and validation loss')\n",
        "plt.xlabel('Epochs')\n",
        "plt.ylabel('Loss')\n",
        "plt.legend()\n",
        "plt.show()"
      ],
      "execution_count": 127,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfUAAAFnCAYAAAC/5tBZAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nLEKhLiIiYhH/BdDSGvf1hE0AAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<matplotlib.figure.Figure at 0x7f39c40c8438>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "id": "Rn4Pojm9qT9I",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "\n",
        "\n",
        "---\n",
        "\n"
      ]
    },
    {
      "metadata": {
        "id": "fNBE2bdasHZW",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 376
        },
        "outputId": "2cf24900-be6f-4195-83ad-1262ca026fa8"
      },
      "cell_type": "code",
      "source": [
        "plt.clf()\n",
        "acc_values = history_dict['acc']\n",
        "val_acc_values = history_dict['val_acc']\n",
        "plt.plot(epochs, val_acc_values, 'bo', label='Training acc')\n",
        "plt.plot(epochs, acc_values, 'b', label='Validation acc')\n",
        "plt.title('Training and validation accuracy')\n",
        "plt.xlabel('Epochs')\n",
        "plt.ylabel('Accuraccy')\n",
        "plt.legend()\n",
        "plt.show()"
      ],
      "execution_count": 128,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAe8AAAFnCAYAAACPasF4AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<matplotlib.figure.Figure at 0x7f39c40c8470>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "id": "fUug4KcczzON",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "\n",
        "\n",
        "---\n",
        "\n"
      ]
    }
  ]
}