day2_MNIST_MLP_skeleton.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## IMPORT PACKAGES"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CURRENT TF VERSION IS [1.2.1]\n",
      "PACKAGES LOADED\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "import tensorflow as tf\n",
    "from tensorflow.examples.tutorials.mnist import input_data\n",
    "print (\"CURRENT TF VERSION IS [%s]\" % (tf.__version__))\n",
    "print (\"PACKAGES LOADED\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## LOAD MNIST DATA"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "mnist = input_data.read_data_sets('data/', one_hot=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## INVESTIGATE MNIST DATA"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def print_np(x):\n",
    "    print ('SHAPE OF is %s' % (x.shape,))\n",
    "    print ('VALUES LOOK LIKE \\n %s' % (x))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "print_np(mnist.train.images)\n",
    "print_np(mnist.train.labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "print_np(mnist.test.images)\n",
    "print_np(mnist.test.labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "print_np(mnist.validation.images)\n",
    "print_np(mnist.validation.labels)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## DRAW MNIST DATA "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "ntrain = mnist.train.images.shape[0]\n",
    "nsample = 3\n",
    "randidx = np.random.randint(ntrain, size=nsample)\n",
    "for i in randidx:\n",
    "    imgvec = mnist.train.images[i, :]\n",
    "    labelvec = mnist.train.labels[i, :]\n",
    "    img = np.reshape(imgvec, (28, 28))\n",
    "    label = np.argmax(labelvec)\n",
    "    plt.matshow(img, cmap=plt.get_cmap('gray'))\n",
    "    plt.title(\"[%d] DATA / LABEL IS [%d]\"%(i, label))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## GET RANDOM BATCH"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "ntrain = 10\n",
    "randindices = np.random.permutation(ntrain)\n",
    "print (randindices.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "ntrain = 10\n",
    "nbatch = 4\n",
    "niter  = ntrain // nbatch + 1\n",
    "for i in range(niter):\n",
    "    currindices = randindices[i*nbatch:(i+1)*nbatch]\n",
    "    print (\"ITER: [%d] BATCH INDEX: %s\" % (i, currindices))\n",
    "    # GET BATCH\n",
    "    xbatch = mnist.train.images[currindices, :]\n",
    "    ybatch = mnist.train.labels[currindices, :]\n",
    "    print (\"  - SHAPE OF 'XBATCH' IS %s\" % (xbatch.shape,))\n",
    "    print (\"  - SHAPE OF 'YBATCH' IS %s\" % (ybatch.shape,))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## DEFINE MODEL"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# NETWORK TOPOLOGIES\n",
    "n_hidden_1 = 256 \n",
    "n_hidden_2 = 128 \n",
    "n_input    = 784 \n",
    "n_classes  = 10\n",
    "\n",
    "# INPUTS AND OUTPUTS\n",
    "x = # [EXERCISE 1]\n",
    "y = # [EXERCISE 1]\n",
    "    \n",
    "# NETWORK PARAMETERS\n",
    "stddev = 0.1\n",
    "weights = {\n",
    "    'h1': # [EXERCISE 1],\n",
    "    'h2': # [EXERCISE 1],\n",
    "    'out': # [EXERCISE 1]\n",
    "}\n",
    "biases = {\n",
    "    'b1': # [EXERCISE 1],\n",
    "    'b2': # [EXERCISE 1],\n",
    "    'out': # [EXERCISE 1]\n",
    "}\n",
    "print (\"NETWORK READY\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## BUILD GRAPH"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# MODEL\n",
    "def multilayer_perceptron(_X, _weights, _biases):\n",
    "    # [EXERCISE 2]\n",
    "\n",
    "# PREDICTION\n",
    "pred = multilayer_perceptron(x, weights, biases)\n",
    "\n",
    "# LOSS AND OPTIMIZER\n",
    "cost = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(labels=# [EXERCISE 3], logits=# [EXERCISE 3]))\n",
    "optm = tf.train.AdamOptimizer(learning_rate=0.001).minimize(cost) \n",
    "corr = # [EXERCISE 3]\n",
    "accr = tf.reduce_mean(tf.cast(corr, \"float\"))\n",
    "\n",
    "# INITIALIZER\n",
    "init = tf.global_variables_initializer()\n",
    "print (\"FUNCTIONS READY\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## RUN"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# PARAMETERS\n",
    "training_epochs = 20\n",
    "batch_size      = 100\n",
    "display_step    = 4\n",
    "# LAUNCH THE GRAPH\n",
    "sess = tf.Session()\n",
    "sess.run(init)\n",
    "# OPTIMIZE\n",
    "for epoch in range(training_epochs):\n",
    "    avg_cost = 0.\n",
    "    total_batch = int(mnist.train.num_examples/batch_size)\n",
    "    # ITERATION\n",
    "    for i in range(total_batch):\n",
    "        batch_xs, batch_ys = mnist.train.next_batch(batch_size)\n",
    "        feeds = {x: batch_xs, y: batch_ys}\n",
    "        sess.run(optm, feed_dict=feeds)\n",
    "        avg_cost += sess.run(cost, feed_dict=feeds)\n",
    "    avg_cost = avg_cost / total_batch\n",
    "    # DISPLAY\n",
    "    if (epoch+1) % display_step == 0:\n",
    "        print (\"Epoch: %03d/%03d cost: %.9f\" % (epoch, training_epochs, avg_cost))\n",
    "        feeds = {x: batch_xs, y: batch_ys}\n",
    "        train_acc = sess.run(accr, feed_dict=feeds)\n",
    "        print (\"TRAIN ACCURACY: %.3f\" % (train_acc))\n",
    "        feeds = {x: mnist.test.images, y: mnist.test.labels}\n",
    "        test_acc = sess.run(accr, feed_dict=feeds)\n",
    "        print (\"TEST ACCURACY: %.3f\" % (test_acc))\n",
    "print (\"OPTIMIZATION FINISHED\")"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [conda env:tensorflow]",
   "language": "python",
   "name": "conda-env-tensorflow-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}