{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "from sklearn import datasets, model_selection\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense\n",
    "from keras import optimizers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "mnist = datasets.fetch_mldata('MNIST original', data_home='data/src/download')\n",
    "\n",
    "X = mnist.data / 255\n",
    "y = mnist.target\n",
    "Y = np.identity(10)[y.astype(int)]\n",
    "\n",
    "train_size = 60000\n",
    "test_size = 10000\n",
    "\n",
    "X_train, X_test, Y_train, Y_test = model_selection.train_test_split(\n",
    "    X, Y, test_size=test_size, train_size=train_size)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "model = Sequential()\n",
    "model.add(Dense(units=10, input_dim=784, activation='softmax'))\n",
    "sgd = optimizers.SGD(lr=0.5)\n",
    "model.compile(\n",
    "    optimizer=sgd,\n",
    "    loss='categorical_crossentropy',\n",
    "    metrics=['accuracy']\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "batch_size = 100\n",
    "epochs = 20"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/20\n",
      "60000/60000 [==============================] - 1s - loss: 0.3934 - acc: 0.8884     \n",
      "Epoch 2/20\n",
      "60000/60000 [==============================] - 1s - loss: 0.3018 - acc: 0.9145     \n",
      "Epoch 3/20\n",
      "60000/60000 [==============================] - 1s - loss: 0.2870 - acc: 0.9192     \n",
      "Epoch 4/20\n",
      "60000/60000 [==============================] - 1s - loss: 0.2793 - acc: 0.9212     \n",
      "Epoch 5/20\n",
      "60000/60000 [==============================] - 1s - loss: 0.2739 - acc: 0.9235     \n",
      "Epoch 6/20\n",
      "60000/60000 [==============================] - 1s - loss: 0.2699 - acc: 0.9242     \n",
      "Epoch 7/20\n",
      "60000/60000 [==============================] - 1s - loss: 0.2666 - acc: 0.9252     \n",
      "Epoch 8/20\n",
      "60000/60000 [==============================] - 1s - loss: 0.2645 - acc: 0.9251     \n",
      "Epoch 9/20\n",
      "60000/60000 [==============================] - 1s - loss: 0.2621 - acc: 0.9269     \n",
      "Epoch 10/20\n",
      "60000/60000 [==============================] - 1s - loss: 0.2601 - acc: 0.9280     \n",
      "Epoch 11/20\n",
      "60000/60000 [==============================] - 1s - loss: 0.2588 - acc: 0.9274     \n",
      "Epoch 12/20\n",
      "60000/60000 [==============================] - 1s - loss: 0.2586 - acc: 0.9277     \n",
      "Epoch 13/20\n",
      "60000/60000 [==============================] - 1s - loss: 0.2558 - acc: 0.9285     \n",
      "Epoch 14/20\n",
      "60000/60000 [==============================] - 1s - loss: 0.2547 - acc: 0.9287     \n",
      "Epoch 15/20\n",
      "60000/60000 [==============================] - 1s - loss: 0.2537 - acc: 0.9297     \n",
      "Epoch 16/20\n",
      "60000/60000 [==============================] - 1s - loss: 0.2526 - acc: 0.9288     \n",
      "Epoch 17/20\n",
      "60000/60000 [==============================] - 1s - loss: 0.2524 - acc: 0.9287     \n",
      "Epoch 18/20\n",
      "60000/60000 [==============================] - 1s - loss: 0.2515 - acc: 0.9298     \n",
      "Epoch 19/20\n",
      "60000/60000 [==============================] - 1s - loss: 0.2499 - acc: 0.9303     \n",
      "Epoch 20/20\n",
      "60000/60000 [==============================] - 1s - loss: 0.2499 - acc: 0.9302     \n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x10d404d30>"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.fit(X_train, Y_train, epochs=epochs, batch_size=batch_size)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "10000/10000 [==============================] - 0s     \n",
      "[0.30260467473864555, 0.91790000000000005]\n"
     ]
    }
   ],
   "source": [
    "score = model.evaluate(X_test, Y_test)\n",
    "print(score)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}