tensor_flow-example10-softmax_classifier_for_MNIST.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# lecture 10 slide 7\n",
    "import tensorflow as tf\n",
    "import numpy as np\n",
    "import random\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Extracting MNIST_data/train-images-idx3-ubyte.gz\n",
      "Extracting MNIST_data/train-labels-idx1-ubyte.gz\n",
      "Extracting MNIST_data/t10k-images-idx3-ubyte.gz\n",
      "Extracting MNIST_data/t10k-labels-idx1-ubyte.gz\n"
     ]
    }
   ],
   "source": [
    "from tensorflow.examples.tutorials.mnist import input_data\n",
    "# reproduceability\n",
    "tf.set_random_seed(777)\n",
    "mnist = input_data.read_data_sets('MNIST_data/',one_hot = True)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# input placeholder\n",
    "X = tf.placeholder(tf.float32, [None,784])\n",
    "Y = tf.placeholder(tf.float32,[None,10])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "W = tf.Variable(tf.random_normal([784,10]))\n",
    "b = tf.Variable(tf.random_normal([10]))\n",
    "hypothesis = tf.matmul(X,W) + b"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# define parameters\n",
    "\n",
    "learning_rate = 0.001\n",
    "batch_size = 100\n",
    "num_epochs = 50\n",
    "num_iterations = int(mnist.train.num_examples/batch_size)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# define cost/loss & optimizer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "cost = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits=hypothesis, labels =Y))\n",
    "train = tf.train.AdamOptimizer(learning_rate = learning_rate).minimize(cost)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "cost = tf.reduce_mean(\n",
    "        tf.nn.softmax_cross_entropy_with_logits(\n",
    "            logits = hypothesis, labels = tf.stop_gradient(Y)\n",
    "    )\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "correct_prediction = tf.equal(tf.argmax(hypothesis,axis = 1), tf.argmax(Y, axis = 1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "accuracy = tf.reduce_mean(tf.cast(correct_prediction,tf.float32))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 0001 cost = 5.743060071\n",
      "Epoch: 0002 cost = 1.773943864\n",
      "Epoch: 0003 cost = 1.119321038\n",
      "Epoch: 0004 cost = 0.868712265\n",
      "Epoch: 0005 cost = 0.736231768\n",
      "Epoch: 0006 cost = 0.652670781\n",
      "Epoch: 0007 cost = 0.594188317\n",
      "Epoch: 0008 cost = 0.550967683\n",
      "Epoch: 0009 cost = 0.516835602\n",
      "Epoch: 0010 cost = 0.489964912\n",
      "Epoch: 0011 cost = 0.467128218\n",
      "Epoch: 0012 cost = 0.448298602\n",
      "Epoch: 0013 cost = 0.432575236\n",
      "Epoch: 0014 cost = 0.417959033\n",
      "Epoch: 0015 cost = 0.405480357\n",
      "Epoch: 0016 cost = 0.394785488\n",
      "Epoch: 0017 cost = 0.385329806\n",
      "Epoch: 0018 cost = 0.376084003\n",
      "Epoch: 0019 cost = 0.368119199\n",
      "Epoch: 0020 cost = 0.361015388\n",
      "Epoch: 0021 cost = 0.354496846\n",
      "Epoch: 0022 cost = 0.348092492\n",
      "Epoch: 0023 cost = 0.342500602\n",
      "Epoch: 0024 cost = 0.337264904\n",
      "Epoch: 0025 cost = 0.332164492\n",
      "Epoch: 0026 cost = 0.327897823\n",
      "Epoch: 0027 cost = 0.323723156\n",
      "Epoch: 0028 cost = 0.319248884\n",
      "Epoch: 0029 cost = 0.315582290\n",
      "Epoch: 0030 cost = 0.312147272\n",
      "Epoch: 0031 cost = 0.309228776\n",
      "Epoch: 0032 cost = 0.305559514\n",
      "Epoch: 0033 cost = 0.302600884\n",
      "Epoch: 0034 cost = 0.300022253\n",
      "Epoch: 0035 cost = 0.297138732\n",
      "Epoch: 0036 cost = 0.294461078\n",
      "Epoch: 0037 cost = 0.292577450\n",
      "Epoch: 0038 cost = 0.289750245\n",
      "Epoch: 0039 cost = 0.287941169\n",
      "Epoch: 0040 cost = 0.284938286\n",
      "Epoch: 0041 cost = 0.283750568\n",
      "Epoch: 0042 cost = 0.282085486\n",
      "Epoch: 0043 cost = 0.279926878\n",
      "Epoch: 0044 cost = 0.278170597\n",
      "Epoch: 0045 cost = 0.276369286\n",
      "Epoch: 0046 cost = 0.274539205\n",
      "Epoch: 0047 cost = 0.273366167\n",
      "Epoch: 0048 cost = 0.271711297\n",
      "Epoch: 0049 cost = 0.270217787\n",
      "Epoch: 0050 cost = 0.268990613\n",
      "learning finished\n",
      "accuracy 0.9206\n",
      "label: [6]\n",
      "prediction: [6]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x243801566d8>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "with tf.Session() as sess:\n",
    "    sess.run(tf.global_variables_initializer())\n",
    "    \n",
    "    for epoch in range(num_epochs):\n",
    "        avg_cost = 0\n",
    "        \n",
    "        for iteration in range(num_iterations):\n",
    "            batch_xs,batch_ys = mnist.train.next_batch(batch_size)\n",
    "            _,cost_val = sess.run([train,cost],feed_dict = {X:batch_xs,Y:batch_ys})\n",
    "            avg_cost += cost_val/num_iterations\n",
    "        print('Epoch:','%04d' % (epoch +1), 'cost =', '{:.9f}'.format(avg_cost))\n",
    "    print('learning finished')\n",
    "    print(\n",
    "        'accuracy',sess.run(accuracy,feed_dict = {X:mnist.test.images,Y:mnist.test.labels}))\n",
    "    r = random.randint(0,mnist.test.num_examples -1)\n",
    "    print('label:',sess.run(tf.argmax(mnist.test.labels[r:r+1],axis = 1)))\n",
    "    print(\n",
    "    'prediction:',sess.run(tf.argmax(hypothesis,axis =1),feed_dict = {X:mnist.test.images[r:r+1]}))\n",
    "    plt.imshow(mnist.test.images[r:r+1].reshape(28,28),cmap ='Greys',interpolation='nearest')\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.5.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}