Keras_Mnist.ipynb

Keras_Mnist.ipynb

{
  "cells": [
    {
      "metadata": {
        "ExecuteTime": {
          "start_time": "2018-05-18T07:04:17.691786Z",
          "end_time": "2018-05-18T07:04:32.575710Z"
        },
        "trusted": true
      },
      "cell_type": "code",
      "source": "import numpy as np\nimport matplotlib.pyplot as plt\nimport tensorflow as tf\nimport keras\nfrom keras.models import Sequential\nfrom keras.layers import *\nimport keras.backend as K",
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "text": "Using TensorFlow backend.\n",
          "name": "stderr"
        }
      ]
    },
    {
      "metadata": {
        "ExecuteTime": {
          "start_time": "2018-05-18T07:04:34.038329Z",
          "end_time": "2018-05-18T07:04:34.529171Z"
        },
        "trusted": true
      },
      "cell_type": "code",
      "source": "(x_train, y_train), (x_test, y_test) = keras.datasets.mnist.load_data()",
      "execution_count": 2,
      "outputs": []
    },
    {
      "metadata": {
        "ExecuteTime": {
          "start_time": "2018-05-18T07:04:35.066300Z",
          "end_time": "2018-05-18T07:04:35.070311Z"
        },
        "trusted": true
      },
      "cell_type": "code",
      "source": "print(x_train.shape)\nprint(x_test.shape)\nprint(y_train.shape)\nprint(y_test.shape)",
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "stream",
          "text": "(60000, 28, 28)\n(10000, 28, 28)\n(60000,)\n(10000,)\n",
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "ExecuteTime": {
          "start_time": "2018-05-18T07:04:35.964971Z",
          "end_time": "2018-05-18T07:04:36.216160Z"
        },
        "trusted": true
      },
      "cell_type": "code",
      "source": "x_train_1d = x_train.reshape((-1, 784)).astype(np.float32) / 255.0\nx_test_1d = x_test.reshape((-1, 784)).astype(np.float32) / 255.0",
      "execution_count": 4,
      "outputs": []
    },
    {
      "metadata": {
        "ExecuteTime": {
          "start_time": "2018-05-18T07:04:36.748704Z",
          "end_time": "2018-05-18T07:04:36.756225Z"
        },
        "trusted": true
      },
      "cell_type": "code",
      "source": "NUM_CLASS = 10\n\ny_train_onehot = np.eye(NUM_CLASS)[y_train]\ny_test_onehot = np.eye(NUM_CLASS)[y_test]",
      "execution_count": 5,
      "outputs": []
    },
    {
      "metadata": {
        "ExecuteTime": {
          "start_time": "2018-05-18T07:04:37.578079Z",
          "end_time": "2018-05-18T07:04:37.582091Z"
        },
        "trusted": true
      },
      "cell_type": "code",
      "source": "print(x_train_1d.shape)\nprint(x_test_1d.shape)\nprint(y_train_onehot.shape)\nprint(y_test_onehot.shape)",
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "stream",
          "text": "(60000, 784)\n(10000, 784)\n(60000, 10)\n(10000, 10)\n",
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "ExecuteTime": {
          "start_time": "2018-05-18T07:04:38.371721Z",
          "end_time": "2018-05-18T07:04:39.098131Z"
        },
        "trusted": true
      },
      "cell_type": "code",
      "source": "plt.imshow(x_train_1d[0].reshape(28, 28))",
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "execute_result",
          "execution_count": 7,
          "data": {
            "text/plain": "<matplotlib.image.AxesImage at 0x20d19d75fd0>"
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": "<Figure size 432x288 with 1 Axes>",
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "metadata": {
        "ExecuteTime": {
          "start_time": "2018-05-18T07:04:39.780961Z",
          "end_time": "2018-05-18T07:04:39.952448Z"
        },
        "code_folding": [],
        "trusted": true
      },
      "cell_type": "code",
      "source": "model = Sequential()\nmodel.add(InputLayer((784,)))\nmodel.add(Dense(16, activation='relu'))\nmodel.add(Dense(32, activation='relu'))\nmodel.add(Dense(64, activation='relu'))\nmodel.add(Dense(10, activation='softmax'))\nmodel.summary()\nmodel.compile(optimizer=\"adam\",\n              loss=\"categorical_crossentropy\",\n              metrics=[\"accuracy\"])",
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "stream",
          "text": "_________________________________________________________________\nLayer (type)                 Output Shape              Param #   \n=================================================================\ninput_1 (InputLayer)         (None, 784)               0         \n_________________________________________________________________\ndense_1 (Dense)              (None, 16)                12560     \n_________________________________________________________________\ndense_2 (Dense)              (None, 32)                544       \n_________________________________________________________________\ndense_3 (Dense)              (None, 64)                2112      \n_________________________________________________________________\ndense_4 (Dense)              (None, 10)                650       \n=================================================================\nTotal params: 15,866\nTrainable params: 15,866\nNon-trainable params: 0\n_________________________________________________________________\n",
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "ExecuteTime": {
          "start_time": "2018-05-18T07:04:50.011106Z",
          "end_time": "2018-05-18T07:08:43.165644Z"
        },
        "trusted": true
      },
      "cell_type": "code",
      "source": "history = model.fit(x_train_1d, y_train_onehot,\n                    validation_split=0.05, epochs=25)",
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "stream",
          "text": "Train on 57000 samples, validate on 3000 samples\nEpoch 1/25\n57000/57000 [==============================] - 13s 225us/step - loss: 0.3741 - acc: 0.8885 - val_loss: 0.1619 - val_acc: 0.9537\nEpoch 2/25\n57000/57000 [==============================] - 9s 160us/step - loss: 0.1952 - acc: 0.9416 - val_loss: 0.1421 - val_acc: 0.9600\nEpoch 3/25\n57000/57000 [==============================] - 9s 159us/step - loss: 0.1633 - acc: 0.9509 - val_loss: 0.1107 - val_acc: 0.9693\nEpoch 4/25\n57000/57000 [==============================] - 9s 161us/step - loss: 0.1433 - acc: 0.9564 - val_loss: 0.1160 - val_acc: 0.9653\nEpoch 5/25\n57000/57000 [==============================] - 9s 160us/step - loss: 0.1324 - acc: 0.9599 - val_loss: 0.1177 - val_acc: 0.9670\nEpoch 6/25\n57000/57000 [==============================] - 9s 164us/step - loss: 0.1197 - acc: 0.9632 - val_loss: 0.1041 - val_acc: 0.9707\nEpoch 7/25\n57000/57000 [==============================] - 9s 159us/step - loss: 0.1106 - acc: 0.9662 - val_loss: 0.0996 - val_acc: 0.9723\nEpoch 8/25\n57000/57000 [==============================] - 9s 158us/step - loss: 0.1040 - acc: 0.9674 - val_loss: 0.1033 - val_acc: 0.9703\nEpoch 9/25\n57000/57000 [==============================] - 9s 160us/step - loss: 0.0972 - acc: 0.9693 - val_loss: 0.1058 - val_acc: 0.9693\nEpoch 10/25\n57000/57000 [==============================] - 9s 162us/step - loss: 0.0901 - acc: 0.9727 - val_loss: 0.1235 - val_acc: 0.9640\nEpoch 11/25\n57000/57000 [==============================] - 9s 161us/step - loss: 0.0870 - acc: 0.9725 - val_loss: 0.1093 - val_acc: 0.9703\nEpoch 12/25\n57000/57000 [==============================] - 9s 161us/step - loss: 0.0833 - acc: 0.9739 - val_loss: 0.0981 - val_acc: 0.9713\nEpoch 13/25\n57000/57000 [==============================] - 9s 161us/step - loss: 0.0796 - acc: 0.9750 - val_loss: 0.1108 - val_acc: 0.9720\nEpoch 14/25\n57000/57000 [==============================] - 9s 158us/step - loss: 0.0760 - acc: 0.9758 - val_loss: 0.0998 - val_acc: 0.9770\nEpoch 15/25\n57000/57000 [==============================] - 9s 161us/step - loss: 0.0718 - acc: 0.9773 - val_loss: 0.0988 - val_acc: 0.9733\nEpoch 16/25\n57000/57000 [==============================] - 9s 160us/step - loss: 0.0699 - acc: 0.9775 - val_loss: 0.1018 - val_acc: 0.9737\nEpoch 17/25\n57000/57000 [==============================] - 9s 161us/step - loss: 0.0669 - acc: 0.9788 - val_loss: 0.1071 - val_acc: 0.9733\nEpoch 18/25\n57000/57000 [==============================] - 9s 159us/step - loss: 0.0660 - acc: 0.9788 - val_loss: 0.1074 - val_acc: 0.9727\nEpoch 19/25\n57000/57000 [==============================] - 9s 162us/step - loss: 0.0618 - acc: 0.9798 - val_loss: 0.1141 - val_acc: 0.9723\nEpoch 20/25\n57000/57000 [==============================] - 9s 161us/step - loss: 0.0615 - acc: 0.9802 - val_loss: 0.1066 - val_acc: 0.9733\nEpoch 21/25\n57000/57000 [==============================] - 9s 161us/step - loss: 0.0585 - acc: 0.9814 - val_loss: 0.1012 - val_acc: 0.9733\nEpoch 22/25\n57000/57000 [==============================] - 9s 160us/step - loss: 0.0567 - acc: 0.9820 - val_loss: 0.1062 - val_acc: 0.9723\nEpoch 23/25\n57000/57000 [==============================] - 9s 163us/step - loss: 0.0568 - acc: 0.9817 - val_loss: 0.1193 - val_acc: 0.9723\nEpoch 24/25\n57000/57000 [==============================] - 9s 164us/step - loss: 0.0530 - acc: 0.9827 - val_loss: 0.1062 - val_acc: 0.9717\nEpoch 25/25\n57000/57000 [==============================] - 9s 162us/step - loss: 0.0517 - acc: 0.9827 - val_loss: 0.1176 - val_acc: 0.9713\n",
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "ExecuteTime": {
          "start_time": "2018-05-18T07:08:54.921538Z",
          "end_time": "2018-05-18T07:08:55.506606Z"
        },
        "trusted": true
      },
      "cell_type": "code",
      "source": "score = model.evaluate(x_test_1d, y_test_onehot)",
      "execution_count": 10,
      "outputs": [
        {
          "output_type": "stream",
          "text": "10000/10000 [==============================] - 1s 58us/step\n",
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "ExecuteTime": {
          "start_time": "2018-05-18T07:08:56.262112Z",
          "end_time": "2018-05-18T07:08:56.265622Z"
        },
        "trusted": true
      },
      "cell_type": "code",
      "source": "print(\"Test Data Loss:\", score[0])\nprint(\"Test Data Accuracy:\", score[1])",
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "stream",
          "text": "Test Data Loss: 0.173282636549\nTest Data Accuracy: 0.9589\n",
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "ExecuteTime": {
          "start_time": "2018-05-18T07:08:58.030058Z",
          "end_time": "2018-05-18T07:08:58.277197Z"
        },
        "code_folding": [],
        "trusted": true
      },
      "cell_type": "code",
      "source": "def plot_history(history):\n    # print(history.history.keys())\n\n    # 精度の履歴をプロット\n    plt.plot(history.history['acc'])\n    plt.plot(history.history['val_acc'])\n    plt.title('model accuracy')\n    plt.xlabel('epoch')\n    plt.ylabel('accuracy')\n    plt.legend(['acc', 'val_acc'], loc='lower right')\n    plt.show()\n\n    # 損失の履歴をプロット\n    plt.plot(history.history['loss'])\n    plt.plot(history.history['val_loss'])\n    plt.title('model loss')\n    plt.xlabel('epoch')\n    plt.ylabel('loss')\n    plt.legend(['loss', 'val_loss'], loc='lower right')\n    plt.show()\n\n\n# 学習履歴をプロット\nplot_history(history)",
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": "<Figure size 432x288 with 1 Axes>",
            "image/png": "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\n"
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": "<Figure size 432x288 with 1 Axes>",
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "metadata": {
        "trusted": true
      },
      "cell_type": "code",
      "source": "",
      "execution_count": null,
      "outputs": []
    }
  ],
  "metadata": {
    "language_info": {
      "file_extension": ".py",
      "pygments_lexer": "ipython3",
      "version": "3.5.5",
      "nbconvert_exporter": "python",
      "codemirror_mode": {
        "version": 3,
        "name": "ipython"
      },
      "mimetype": "text/x-python",
      "name": "python"
    },
    "varInspector": {
      "window_display": false,
      "cols": {
        "lenName": 16,
        "lenType": 16,
        "lenVar": 40
      },
      "kernels_config": {
        "python": {
          "library": "var_list.py",
          "delete_cmd_prefix": "del ",
          "delete_cmd_postfix": "",
          "varRefreshCmd": "print(var_dic_list())"
        },
        "r": {
          "library": "var_list.r",
          "delete_cmd_prefix": "rm(",
          "delete_cmd_postfix": ") ",
          "varRefreshCmd": "cat(var_dic_list()) "
        }
      },
      "types_to_exclude": [
        "module",
        "function",
        "builtin_function_or_method",
        "instance",
        "_Feature"
      ]
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3",
      "language": "python"
    },
    "anaconda-cloud": {},
    "gist": {
      "id": "bae300451aabc33fc952a11865938edd",
      "data": {
        "description": "Keras_Mnist.ipynb",
        "public": true
      }
    },
    "_draft": {
      "nbviewer_url": "https://gist.github.com/bae300451aabc33fc952a11865938edd"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 2
}