multiclass.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "multiclass.ipynb",
      "version": "0.3.2",
      "provenance": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "metadata": {
        "id": "Mi6x0HzyM24W",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "from keras.datasets import reuters\n",
        "import numpy as np\n",
        "from keras.utils import to_categorical\n",
        "from keras import models\n",
        "from keras import layers\n",
        "import matplotlib.pyplot as plt\n",
        "import copy"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "ZHZd2uNSgmrt",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 71
        },
        "outputId": "23f87928-5ccb-4319-e249-2f01ccb9f477"
      },
      "cell_type": "code",
      "source": [
        "(train_data, train_labels), (test_data, test_labels) = reuters.load_data(num_words=10000)"
      ],
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Downloading data from https://s3.amazonaws.com/text-datasets/reuters.npz\n",
            "2113536/2110848 [==============================] - 0s 0us/step\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "s5Di07Yfgs74",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 88
        },
        "outputId": "4c0590cc-a426-44cf-b4cf-b5a57976bc4f"
      },
      "cell_type": "code",
      "source": [
        "word_index = reuters.get_word_index()\n",
        "reverse_word_index = dict([(value, key) for (key, value) in word_index.items()])\n",
        "decoded_newswire = ' '.join([reverse_word_index.get(i - 3, '?') for i in train_data[0]])\n",
        "decoded_newswire"
      ],
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Downloading data from https://s3.amazonaws.com/text-datasets/reuters_word_index.json\n",
            "557056/550378 [==============================] - 0s 1us/step\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "'? ? ? said as a result of its december acquisition of space co it expects earnings per share in 1987 of 1 15 to 1 30 dlrs per share up from 70 cts in 1986 the company said pretax net should rise to nine to 10 mln dlrs from six mln dlrs in 1986 and rental operation revenues to 19 to 22 mln dlrs from 12 5 mln dlrs it said cash flow per share this year should be 2 50 to three dlrs reuter 3'"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 6
        }
      ]
    },
    {
      "metadata": {
        "id": "LNC8YVjKhEOI",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "def vectorize_sequences(sequences, dimension=10000):\n",
        "  results = np.zeros((len(sequences), dimension))\n",
        "  for i, sequence in enumerate(sequences):\n",
        "    results[i, sequence] = 1\n",
        "  return results\n",
        "\n",
        "x_train = vectorize_sequences(train_data)\n",
        "x_test = vectorize_sequences(test_data)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "3Nc0ds8ChRJg",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "def to_one_hot(labels, dimension=46):\n",
        "  results = np.zeros((len(labels), dimension))\n",
        "  for i, label in enumerate(labels):\n",
        "    results[i, label] = 1\n",
        "  return results\n",
        "\n",
        "#one_hot_train_labels = to_one_hot(train_labels)\n",
        "#one_hot_test_labels = to_one_hot(test_labels)\n",
        "\n",
        "one_hot_train_labels = to_categorical(train_labels)\n",
        "one_hot_test_labels = to_categorical(test_labels)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "D0dple9uhoQy",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "model = models.Sequential()\n",
        "model.add(layers.Dense(64, activation='relu', input_shape=(10000,)))\n",
        "model.add(layers.Dense(128, activation='relu'))\n",
        "model.add(layers.Dense(128, activation='relu'))\n",
        "model.add(layers.Dense(46, activation='softmax'))\n",
        "\n",
        "model.compile(optimizer='rmsprop', loss='categorical_crossentropy', metrics=['accuracy'])"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "5yNv4LS4j_zr",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "x_val = x_train[:1000]\n",
        "partial_x_train = x_train[1000:]\n",
        "y_val = one_hot_train_labels[:1000]\n",
        "partial_y_train = one_hot_train_labels[1000:]"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "CX1G3JphkAD6",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 340
        },
        "outputId": "e18c2b96-a28a-4017-8b40-1f1fa0083f2e"
      },
      "cell_type": "code",
      "source": [
        "history = model.fit(partial_x_train, partial_y_train, epochs=9, batch_size=512, validation_data=(x_val, y_val))"
      ],
      "execution_count": 25,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train on 7982 samples, validate on 1000 samples\n",
            "Epoch 1/9\n",
            "7982/7982 [==============================] - 1s 113us/step - loss: 2.3902 - acc: 0.5049 - val_loss: 1.5406 - val_acc: 0.6560\n",
            "Epoch 2/9\n",
            "7982/7982 [==============================] - 1s 78us/step - loss: 1.3300 - acc: 0.7005 - val_loss: 1.2796 - val_acc: 0.7300\n",
            "Epoch 3/9\n",
            "7982/7982 [==============================] - 1s 77us/step - loss: 1.0282 - acc: 0.7671 - val_loss: 1.1336 - val_acc: 0.7480\n",
            "Epoch 4/9\n",
            "7982/7982 [==============================] - 1s 83us/step - loss: 0.8069 - acc: 0.8215 - val_loss: 1.0795 - val_acc: 0.7570\n",
            "Epoch 5/9\n",
            "7982/7982 [==============================] - 1s 82us/step - loss: 0.6355 - acc: 0.8544 - val_loss: 1.0112 - val_acc: 0.7880\n",
            "Epoch 6/9\n",
            "7982/7982 [==============================] - 1s 82us/step - loss: 0.4791 - acc: 0.8943 - val_loss: 0.9722 - val_acc: 0.8090\n",
            "Epoch 7/9\n",
            "7982/7982 [==============================] - 1s 81us/step - loss: 0.3939 - acc: 0.9169 - val_loss: 0.9576 - val_acc: 0.8020\n",
            "Epoch 8/9\n",
            "7982/7982 [==============================] - 1s 81us/step - loss: 0.3034 - acc: 0.9320 - val_loss: 0.9856 - val_acc: 0.8120\n",
            "Epoch 9/9\n",
            "7982/7982 [==============================] - 1s 78us/step - loss: 0.2475 - acc: 0.9421 - val_loss: 1.0712 - val_acc: 0.7780\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "r0r9R-nEj6_v",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 376
        },
        "outputId": "f3907570-2f82-44fb-8fc6-c8d4283584b3"
      },
      "cell_type": "code",
      "source": [
        "loss = history.history['loss']\n",
        "val_loss = history.history['val_loss']\n",
        "epochs = range(1, len(loss) + 1)\n",
        "plt.plot(epochs, loss, 'bo', label='Training loss')\n",
        "plt.plot(epochs, val_loss, '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": 26,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAe8AAAFnCAYAAACPasF4AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ng1F4i4iIGIzCW0RExGAU3iIiIgbz/wE2Al8oPXBXIwAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<matplotlib.figure.Figure at 0x7f6919752630>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "id": "QsGvLDs3khKB",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 376
        },
        "outputId": "ab5a9572-bf1d-428f-c277-e03577fa07bd"
      },
      "cell_type": "code",
      "source": [
        "acc = history.history['acc']\n",
        "val_acc = history.history['val_acc']\n",
        "plt.plot(epochs, acc, 'bo', label='Training acc')\n",
        "plt.plot(epochs, val_acc, 'b', label='Validation acc')\n",
        "plt.title('Training and validation accuracy')\n",
        "plt.xlabel('Epochs')\n",
        "plt.ylabel('Loss')\n",
        "plt.legend()\n",
        "plt.show()"
      ],
      "execution_count": 27,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAe8AAAFnCAYAAACPasF4AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\niIiIwfw/nkL/F8FzO4EAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<matplotlib.figure.Figure at 0x7f6919775cc0>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "id": "wFRiTs4TkoC3",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "5ec181bf-29ec-46be-d9ec-5f98491a96e9"
      },
      "cell_type": "code",
      "source": [
        "results = model.evaluate(x_test, one_hot_test_labels)"
      ],
      "execution_count": 28,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "2246/2246 [==============================] - 0s 152us/step\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "MEuPcwMalB1a",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "7dc32c0c-50db-4679-8ba4-7fc396ef8f1a"
      },
      "cell_type": "code",
      "source": [
        "results"
      ],
      "execution_count": 29,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[1.205018838175895, 0.7577916295902068]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 29
        }
      ]
    },
    {
      "metadata": {
        "id": "Dq7ECuVslEj1",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "e46cf10d-2ab9-43fc-9a07-10ec7ca7075d"
      },
      "cell_type": "code",
      "source": [
        "test_labels_copy = copy.copy(test_labels)\n",
        "np.random.shuffle(test_labels_copy)\n",
        "hits_array = np.array(test_labels) == np.array(test_labels_copy)\n",
        "float(np.sum(hits_array)/ len(test_labels))"
      ],
      "execution_count": 30,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0.1772039180765806"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 30
        }
      ]
    },
    {
      "metadata": {
        "id": "8rWiOY2mRD9l",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "predictions = model.predict(x_test)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "41_x5-dhRbpU",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "#when integer label use:\n",
        "#model.compile(...., loss=\"sparse_categorical_crossentropy\", ....)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "SCf0SdH6SIYN",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        ""
      ],
      "execution_count": 0,
      "outputs": []
    }
  ]
}