fashionmnist3.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/ilayda/Desktop/anaconda3/lib/python3.6/site-packages/h5py/__init__.py:34: FutureWarning: Conversion of the second argument of issubdtype from `float` to `np.floating` is deprecated. In future, it will be treated as `np.float64 == np.dtype(float).type`.\n",
      "  from ._conv import register_converters as _register_converters\n"
     ]
    }
   ],
   "source": [
    "import tensorflow as tf\n",
    "import numpy as np\n",
    "from matplotlib import pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Extracting data/fashion/train-images-idx3-ubyte.gz\n",
      "Extracting data/fashion/train-labels-idx1-ubyte.gz\n",
      "Extracting data/fashion/t10k-images-idx3-ubyte.gz\n",
      "Extracting data/fashion/t10k-labels-idx1-ubyte.gz\n"
     ]
    }
   ],
   "source": [
    "from tensorflow.examples.tutorials.mnist import input_data\n",
    "fashion_mnist = input_data.read_data_sets('data/fashion', source_url='http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/',one_hot=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def var(name, shape, init=None, std=None):\n",
    "    if init is None:\n",
    "        if std is None:\n",
    "            std = (2./shape[0])**0.5\n",
    "        init = tf.truncated_normal_initializer(stddev=std)\n",
    "    return tf.get_variable(name=name, shape=shape,\n",
    "                                  dtype=tf.float32, initializer=init)\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From <ipython-input-4-4bb44abac081>:29: softmax_cross_entropy_with_logits (from tensorflow.python.ops.nn_ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "\n",
      "Future major versions of TensorFlow will allow gradients to flow\n",
      "into the labels input on backprop by default.\n",
      "\n",
      "See tf.nn.softmax_cross_entropy_with_logits_v2.\n",
      "\n"
     ]
    }
   ],
   "source": [
    "g = tf.Graph()\n",
    "with g.as_default():\n",
    "    X = tf.placeholder(dtype=tf.float32, shape=[None, 784])\n",
    "    Y = tf.placeholder(dtype=tf.float32, shape=[None, 10])\n",
    "    \n",
    "    #hidden layer\n",
    "    W1 = var('W1',[784, 256])\n",
    "    b1 = var('b1',[256],tf.constant_initializer(0.1))\n",
    "    out1 = tf.nn.relu(tf.matmul(X, W1) + b1)\n",
    "    \n",
    "    #hidden layer\n",
    "    W2 = var('W2',[256, 256])\n",
    "    b2 = var('b2',[256],tf.constant_initializer(0.1))\n",
    "    out2 = tf.nn.relu(tf.matmul(out1, W2) + b2)\n",
    "    \n",
    "    #output layer\n",
    "    W3 = var('W3',[256,10])\n",
    "    b3 = var('b3',[10],tf.constant_initializer(0.1))\n",
    "    logits =tf.matmul(out2, W3) + b3\n",
    "    \n",
    "      \n",
    "    #accuracy\n",
    "    pred = tf.argmax(logits, axis=1)\n",
    "    truth = tf.argmax(Y, axis=1)\n",
    "    match = tf.cast(tf.equal(pred,truth), tf.float32)\n",
    "    acc = tf.reduce_mean(match)\n",
    "    \n",
    "    loss = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(\n",
    "    logits=logits, labels=Y))\n",
    "    step = tf.train.AdamOptimizer().minimize(loss)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "sess = tf.InteractiveSession(graph=g)\n",
    "tf.global_variables_initializer().run()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "9999) acc: 0.9600\r"
     ]
    }
   ],
   "source": [
    "bsize = 100\n",
    "for I in range(10000):\n",
    "    x,y = fashion_mnist.train.next_batch(bsize)\n",
    "    _, acc_ = sess.run([step, acc], feed_dict={X:x, Y:y})\n",
    "    print('%d) acc: %2.4f' % (I,acc_), end = '\\r')\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.9483455"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sess.run(acc, feed_dict={X:fashion_mnist.train.images, Y:fashion_mnist.train.labels})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.892"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sess.run(acc, feed_dict={X:fashion_mnist.test.images, Y:fashion_mnist.test.labels})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0) loss: 0.1078\n",
      "100) loss: 0.0009\n",
      "200) loss: 0.0004\n",
      "300) loss: 0.0002\n",
      "400) loss: 0.0001\n",
      "500) loss: 0.0001\n",
      "600) loss: 0.0001\n",
      "700) loss: 0.0001\n",
      "800) loss: 0.0000\n",
      "900) loss: 0.0000\n"
     ]
    }
   ],
   "source": [
    "loss_list = []\n",
    "for i in range(1000):\n",
    "    _,loss_ = sess.run([step, loss],feed_dict={X:x, Y:y})\n",
    "    loss_list.append(loss_)\n",
    "    if not i%100:\n",
    "        print('%d) loss: %2.4f'  %(i, loss_))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYQAAAD8CAYAAAB3u9PLAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQUREAPj/pyor2RGJUjAAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1c2d54bc18>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(loss_list[10:1000],c='r')\n",
    "plt.legend(['loss function'])\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": "3.6.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}