{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "from keras.layers.core import Dense, Activation\n",
    "from keras.models import Sequential\n",
    "from keras.optimizers import Adam\n",
    "from sklearn import datasets, preprocessing\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# MNIST - Keras (Backend: TensorFlow)\n",
    "\n",
    "以下のTensorFlowのTutorialと見比べてみて下さい  \n",
    "http://www.tensorflow.org/tutorials/mnist/beginners/index.html"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## MNIST Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# データセットの用意\n",
    "mnist = datasets.fetch_mldata('MNIST original')\n",
    "\n",
    "X = mnist.data.astype(float)\n",
    "X /= 255\n",
    "\n",
    "y = preprocessing.LabelBinarizer().fit_transform(mnist.target)\n",
    "\n",
    "# 訓練用とテスト用にデータを分割\n",
    "train_size = 60000\n",
    "\n",
    "X_train, X_test = np.split(X, [train_size])\n",
    "y_train, y_test = np.split(y, [train_size])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Implementing the Regression"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "model = Sequential()\n",
    "model.add(Dense(10, input_shape=(784,), init='zero'))\n",
    "model.add(Activation(\"softmax\"))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# モデルをコンパイル\n",
    "model.compile(loss='categorical_crossentropy', optimizer=Adam(lr=0.01))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/10\n",
      "60000/60000 [==============================] - 4s - loss: 0.3547     \n",
      "Epoch 2/10\n",
      "60000/60000 [==============================] - 3s - loss: 0.2955     \n",
      "Epoch 3/10\n",
      "60000/60000 [==============================] - 3s - loss: 0.2883     \n",
      "Epoch 4/10\n",
      "60000/60000 [==============================] - 3s - loss: 0.2847     \n",
      "Epoch 5/10\n",
      "60000/60000 [==============================] - 3s - loss: 0.2812     \n",
      "Epoch 6/10\n",
      "60000/60000 [==============================] - 3s - loss: 0.2777     \n",
      "Epoch 7/10\n",
      "60000/60000 [==============================] - 3s - loss: 0.2739     \n",
      "Epoch 8/10\n",
      "60000/60000 [==============================] - 4s - loss: 0.2730     \n",
      "Epoch 9/10\n",
      "60000/60000 [==============================] - 4s - loss: 0.2691     \n",
      "Epoch 10/10\n",
      "60000/60000 [==============================] - 4s - loss: 0.2715     \n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x13045ce50>"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.fit(X_train, y_train, nb_epoch=10, batch_size=100)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Evaluating"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "10000/10000 [==============================] - 0s     \n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[0.32320401608347893, 0.9204]"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.evaluate(X_test, y_test, show_accuracy=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Accuracy: 0.9204"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}