Tensorflow.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "# Introducing Tensorflow\n",
    "\n",
    "Be sure to \"pip install tensorflow\" or \"pip install tensorflow-gpu\" first!\n",
    "\n",
    "## The world's simplest Tensorflow application\n",
    "\n",
    "Let's begin by writing a really simple program to illustrate Tensorflow's main concepts. We'll set up two Variables, named \"a\" and \"b\", which each contain a tensor which contains a single value - the number 1, and the number 2.\n",
    "\n",
    "We then create a graph \"f\" that adds these two tensors together. But \"f = a + b\" just creates the graph; it doesn't actually perform the addition yet.\n",
    "\n",
    "Next we need to initialize any global variables before we run the graph.\n",
    "\n",
    "And finally, we create a Tensorflow Session object, run our variable initializer, and execute the graph with eval(). \n",
    "\n",
    "This returns the sum of 1 + 2 in a rather complex, yet highly scalable manner :)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "ename": "ImportError",
     "evalue": "Traceback (most recent call last):\n  File \"C:\\Users\\ACER\\AppData\\Local\\Enthought\\Canopy\\edm\\envs\\User\\lib\\site-packages\\tensorflow\\python\\pywrap_tensorflow_internal.py\", line 14, in swig_import_helper\n    return importlib.import_module(mname)\n  File \"C:\\Users\\ACER\\AppData\\Local\\Enthought\\Canopy\\edm\\envs\\User\\lib\\importlib\\__init__.py\", line 126, in import_module\n    return _bootstrap._gcd_import(name[level:], package, level)\n  File \"<frozen importlib._bootstrap>\", line 986, in _gcd_import\n  File \"<frozen importlib._bootstrap>\", line 969, in _find_and_load\n  File \"<frozen importlib._bootstrap>\", line 958, in _find_and_load_unlocked\n  File \"<frozen importlib._bootstrap>\", line 666, in _load_unlocked\n  File \"<frozen importlib._bootstrap>\", line 577, in module_from_spec\n  File \"<frozen importlib._bootstrap_external>\", line 906, in create_module\n  File \"<frozen importlib._bootstrap>\", line 222, in _call_with_frames_removed\nImportError: DLL load failed: A dynamic link library (DLL) initialization routine failed.\n\nDuring handling of the above exception, another exception occurred:\n\nTraceback (most recent call last):\n  File \"C:\\Users\\ACER\\AppData\\Local\\Enthought\\Canopy\\edm\\envs\\User\\lib\\site-packages\\tensorflow\\python\\pywrap_tensorflow.py\", line 58, in <module>\n    from tensorflow.python.pywrap_tensorflow_internal import *\n  File \"C:\\Users\\ACER\\AppData\\Local\\Enthought\\Canopy\\edm\\envs\\User\\lib\\site-packages\\tensorflow\\python\\pywrap_tensorflow_internal.py\", line 17, in <module>\n    _pywrap_tensorflow_internal = swig_import_helper()\n  File \"C:\\Users\\ACER\\AppData\\Local\\Enthought\\Canopy\\edm\\envs\\User\\lib\\site-packages\\tensorflow\\python\\pywrap_tensorflow_internal.py\", line 16, in swig_import_helper\n    return importlib.import_module('_pywrap_tensorflow_internal')\n  File \"C:\\Users\\ACER\\AppData\\Local\\Enthought\\Canopy\\edm\\envs\\User\\lib\\importlib\\__init__.py\", line 126, in import_module\n    return _bootstrap._gcd_import(name[level:], package, level)\nImportError: No module named '_pywrap_tensorflow_internal'\n\n\nFailed to load the native TensorFlow runtime.\n\nSee https://www.tensorflow.org/install/install_sources#common_installation_problems\n\nfor some common reasons and solutions.  Include the entire stack trace\nabove this error message when asking for help.",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mImportError\u001b[0m                               Traceback (most recent call last)",
      "\u001b[1;32mC:\\Users\\ACER\\AppData\\Local\\Enthought\\Canopy\\edm\\envs\\User\\lib\\site-packages\\tensorflow\\python\\pywrap_tensorflow_internal.py\u001b[0m in \u001b[0;36mswig_import_helper\u001b[1;34m()\u001b[0m\n\u001b[0;32m     13\u001b[0m         \u001b[1;32mtry\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m---> 14\u001b[1;33m             \u001b[1;32mreturn\u001b[0m \u001b[0mimportlib\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mimport_module\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mmname\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m     15\u001b[0m         \u001b[1;32mexcept\u001b[0m \u001b[0mImportError\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32mC:\\Users\\ACER\\AppData\\Local\\Enthought\\Canopy\\edm\\envs\\User\\lib\\importlib\\__init__.py\u001b[0m in \u001b[0;36mimport_module\u001b[1;34m(name, package)\u001b[0m\n\u001b[0;32m    125\u001b[0m             \u001b[0mlevel\u001b[0m \u001b[1;33m+=\u001b[0m \u001b[1;36m1\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 126\u001b[1;33m     \u001b[1;32mreturn\u001b[0m \u001b[0m_bootstrap\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_gcd_import\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mname\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mlevel\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mpackage\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mlevel\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    127\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32mC:\\Users\\ACER\\AppData\\Local\\Enthought\\Canopy\\edm\\envs\\User\\lib\\importlib\\_bootstrap.py\u001b[0m in \u001b[0;36m_gcd_import\u001b[1;34m(name, package, level)\u001b[0m\n",
      "\u001b[1;32mC:\\Users\\ACER\\AppData\\Local\\Enthought\\Canopy\\edm\\envs\\User\\lib\\importlib\\_bootstrap.py\u001b[0m in \u001b[0;36m_find_and_load\u001b[1;34m(name, import_)\u001b[0m\n",
      "\u001b[1;32mC:\\Users\\ACER\\AppData\\Local\\Enthought\\Canopy\\edm\\envs\\User\\lib\\importlib\\_bootstrap.py\u001b[0m in \u001b[0;36m_find_and_load_unlocked\u001b[1;34m(name, import_)\u001b[0m\n",
      "\u001b[1;32mC:\\Users\\ACER\\AppData\\Local\\Enthought\\Canopy\\edm\\envs\\User\\lib\\importlib\\_bootstrap.py\u001b[0m in \u001b[0;36m_load_unlocked\u001b[1;34m(spec)\u001b[0m\n",
      "\u001b[1;32mC:\\Users\\ACER\\AppData\\Local\\Enthought\\Canopy\\edm\\envs\\User\\lib\\importlib\\_bootstrap.py\u001b[0m in \u001b[0;36mmodule_from_spec\u001b[1;34m(spec)\u001b[0m\n",
      "\u001b[1;32mC:\\Users\\ACER\\AppData\\Local\\Enthought\\Canopy\\edm\\envs\\User\\lib\\importlib\\_bootstrap_external.py\u001b[0m in \u001b[0;36mcreate_module\u001b[1;34m(self, spec)\u001b[0m\n",
      "\u001b[1;32mC:\\Users\\ACER\\AppData\\Local\\Enthought\\Canopy\\edm\\envs\\User\\lib\\importlib\\_bootstrap.py\u001b[0m in \u001b[0;36m_call_with_frames_removed\u001b[1;34m(f, *args, **kwds)\u001b[0m\n",
      "\u001b[1;31mImportError\u001b[0m: DLL load failed: A dynamic link library (DLL) initialization routine failed.",
      "\nDuring handling of the above exception, another exception occurred:\n",
      "\u001b[1;31mImportError\u001b[0m                               Traceback (most recent call last)",
      "\u001b[1;32mC:\\Users\\ACER\\AppData\\Local\\Enthought\\Canopy\\edm\\envs\\User\\lib\\site-packages\\tensorflow\\python\\pywrap_tensorflow.py\u001b[0m in \u001b[0;36m<module>\u001b[1;34m()\u001b[0m\n\u001b[0;32m     57\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m---> 58\u001b[1;33m   \u001b[1;32mfrom\u001b[0m \u001b[0mtensorflow\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mpython\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mpywrap_tensorflow_internal\u001b[0m \u001b[1;32mimport\u001b[0m \u001b[1;33m*\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m     59\u001b[0m   \u001b[1;32mfrom\u001b[0m \u001b[0mtensorflow\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mpython\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mpywrap_tensorflow_internal\u001b[0m \u001b[1;32mimport\u001b[0m \u001b[0m__version__\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32mC:\\Users\\ACER\\AppData\\Local\\Enthought\\Canopy\\edm\\envs\\User\\lib\\site-packages\\tensorflow\\python\\pywrap_tensorflow_internal.py\u001b[0m in \u001b[0;36m<module>\u001b[1;34m()\u001b[0m\n\u001b[0;32m     16\u001b[0m             \u001b[1;32mreturn\u001b[0m \u001b[0mimportlib\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mimport_module\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;34m'_pywrap_tensorflow_internal'\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m---> 17\u001b[1;33m     \u001b[0m_pywrap_tensorflow_internal\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mswig_import_helper\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m     18\u001b[0m     \u001b[1;32mdel\u001b[0m \u001b[0mswig_import_helper\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32mC:\\Users\\ACER\\AppData\\Local\\Enthought\\Canopy\\edm\\envs\\User\\lib\\site-packages\\tensorflow\\python\\pywrap_tensorflow_internal.py\u001b[0m in \u001b[0;36mswig_import_helper\u001b[1;34m()\u001b[0m\n\u001b[0;32m     15\u001b[0m         \u001b[1;32mexcept\u001b[0m \u001b[0mImportError\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m---> 16\u001b[1;33m             \u001b[1;32mreturn\u001b[0m \u001b[0mimportlib\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mimport_module\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;34m'_pywrap_tensorflow_internal'\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m     17\u001b[0m     \u001b[0m_pywrap_tensorflow_internal\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mswig_import_helper\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32mC:\\Users\\ACER\\AppData\\Local\\Enthought\\Canopy\\edm\\envs\\User\\lib\\importlib\\__init__.py\u001b[0m in \u001b[0;36mimport_module\u001b[1;34m(name, package)\u001b[0m\n\u001b[0;32m    125\u001b[0m             \u001b[0mlevel\u001b[0m \u001b[1;33m+=\u001b[0m \u001b[1;36m1\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 126\u001b[1;33m     \u001b[1;32mreturn\u001b[0m \u001b[0m_bootstrap\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_gcd_import\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mname\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mlevel\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mpackage\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mlevel\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    127\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;31mImportError\u001b[0m: No module named '_pywrap_tensorflow_internal'",
      "\nDuring handling of the above exception, another exception occurred:\n",
      "\u001b[1;31mImportError\u001b[0m                               Traceback (most recent call last)",
      "\u001b[1;32m<ipython-input-1-6b0f4483b0d5>\u001b[0m in \u001b[0;36m<module>\u001b[1;34m()\u001b[0m\n\u001b[1;32m----> 1\u001b[1;33m \u001b[1;32mimport\u001b[0m \u001b[0mtensorflow\u001b[0m \u001b[1;32mas\u001b[0m \u001b[0mtf\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m      2\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m      3\u001b[0m \u001b[0ma\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mtf\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mVariable\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;36m1\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mname\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;34m\"a\"\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m      4\u001b[0m \u001b[0mb\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mtf\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mVariable\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;36m2\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mname\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;34m\"b\"\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m      5\u001b[0m \u001b[0mf\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0ma\u001b[0m \u001b[1;33m+\u001b[0m \u001b[0mb\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32mC:\\Users\\ACER\\AppData\\Local\\Enthought\\Canopy\\edm\\envs\\User\\lib\\site-packages\\tensorflow\\__init__.py\u001b[0m in \u001b[0;36m<module>\u001b[1;34m()\u001b[0m\n\u001b[0;32m     20\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     21\u001b[0m \u001b[1;31m# pylint: disable=g-bad-import-order\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m---> 22\u001b[1;33m \u001b[1;32mfrom\u001b[0m \u001b[0mtensorflow\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mpython\u001b[0m \u001b[1;32mimport\u001b[0m \u001b[0mpywrap_tensorflow\u001b[0m  \u001b[1;31m# pylint: disable=unused-import\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m     23\u001b[0m \u001b[1;32mfrom\u001b[0m \u001b[1;33m.\u001b[0m \u001b[1;32mimport\u001b[0m \u001b[0mapp\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     24\u001b[0m \u001b[1;32mfrom\u001b[0m \u001b[1;33m.\u001b[0m \u001b[1;32mimport\u001b[0m \u001b[0mbitwise\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32mC:\\Users\\ACER\\AppData\\Local\\Enthought\\Canopy\\edm\\envs\\User\\lib\\site-packages\\tensorflow\\python\\__init__.py\u001b[0m in \u001b[0;36m<module>\u001b[1;34m()\u001b[0m\n\u001b[0;32m     47\u001b[0m \u001b[1;32mimport\u001b[0m \u001b[0mnumpy\u001b[0m \u001b[1;32mas\u001b[0m \u001b[0mnp\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     48\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m---> 49\u001b[1;33m \u001b[1;32mfrom\u001b[0m \u001b[0mtensorflow\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mpython\u001b[0m \u001b[1;32mimport\u001b[0m \u001b[0mpywrap_tensorflow\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m     50\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     51\u001b[0m \u001b[1;31m# Protocol buffers\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32mC:\\Users\\ACER\\AppData\\Local\\Enthought\\Canopy\\edm\\envs\\User\\lib\\site-packages\\tensorflow\\python\\pywrap_tensorflow.py\u001b[0m in \u001b[0;36m<module>\u001b[1;34m()\u001b[0m\n\u001b[0;32m     72\u001b[0m \u001b[1;32mfor\u001b[0m \u001b[0msome\u001b[0m \u001b[0mcommon\u001b[0m \u001b[0mreasons\u001b[0m \u001b[1;32mand\u001b[0m \u001b[0msolutions\u001b[0m\u001b[1;33m.\u001b[0m  \u001b[0mInclude\u001b[0m \u001b[0mthe\u001b[0m \u001b[0mentire\u001b[0m \u001b[0mstack\u001b[0m \u001b[0mtrace\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     73\u001b[0m above this error message when asking for help.\"\"\" % traceback.format_exc()\n\u001b[1;32m---> 74\u001b[1;33m   \u001b[1;32mraise\u001b[0m \u001b[0mImportError\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mmsg\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m     75\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     76\u001b[0m \u001b[1;31m# pylint: enable=wildcard-import,g-import-not-at-top,unused-import,line-too-long\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;31mImportError\u001b[0m: Traceback (most recent call last):\n  File \"C:\\Users\\ACER\\AppData\\Local\\Enthought\\Canopy\\edm\\envs\\User\\lib\\site-packages\\tensorflow\\python\\pywrap_tensorflow_internal.py\", line 14, in swig_import_helper\n    return importlib.import_module(mname)\n  File \"C:\\Users\\ACER\\AppData\\Local\\Enthought\\Canopy\\edm\\envs\\User\\lib\\importlib\\__init__.py\", line 126, in import_module\n    return _bootstrap._gcd_import(name[level:], package, level)\n  File \"<frozen importlib._bootstrap>\", line 986, in _gcd_import\n  File \"<frozen importlib._bootstrap>\", line 969, in _find_and_load\n  File \"<frozen importlib._bootstrap>\", line 958, in _find_and_load_unlocked\n  File \"<frozen importlib._bootstrap>\", line 666, in _load_unlocked\n  File \"<frozen importlib._bootstrap>\", line 577, in module_from_spec\n  File \"<frozen importlib._bootstrap_external>\", line 906, in create_module\n  File \"<frozen importlib._bootstrap>\", line 222, in _call_with_frames_removed\nImportError: DLL load failed: A dynamic link library (DLL) initialization routine failed.\n\nDuring handling of the above exception, another exception occurred:\n\nTraceback (most recent call last):\n  File \"C:\\Users\\ACER\\AppData\\Local\\Enthought\\Canopy\\edm\\envs\\User\\lib\\site-packages\\tensorflow\\python\\pywrap_tensorflow.py\", line 58, in <module>\n    from tensorflow.python.pywrap_tensorflow_internal import *\n  File \"C:\\Users\\ACER\\AppData\\Local\\Enthought\\Canopy\\edm\\envs\\User\\lib\\site-packages\\tensorflow\\python\\pywrap_tensorflow_internal.py\", line 17, in <module>\n    _pywrap_tensorflow_internal = swig_import_helper()\n  File \"C:\\Users\\ACER\\AppData\\Local\\Enthought\\Canopy\\edm\\envs\\User\\lib\\site-packages\\tensorflow\\python\\pywrap_tensorflow_internal.py\", line 16, in swig_import_helper\n    return importlib.import_module('_pywrap_tensorflow_internal')\n  File \"C:\\Users\\ACER\\AppData\\Local\\Enthought\\Canopy\\edm\\envs\\User\\lib\\importlib\\__init__.py\", line 126, in import_module\n    return _bootstrap._gcd_import(name[level:], package, level)\nImportError: No module named '_pywrap_tensorflow_internal'\n\n\nFailed to load the native TensorFlow runtime.\n\nSee https://www.tensorflow.org/install/install_sources#common_installation_problems\n\nfor some common reasons and solutions.  Include the entire stack trace\nabove this error message when asking for help."
     ]
    }
   ],
   "source": [
    "import tensorflow as tf\n",
    "\n",
    "a = tf.Variable(1, name=\"a\")\n",
    "b = tf.Variable(2, name=\"b\")\n",
    "f = a + b\n",
    "\n",
    "init = tf.global_variables_initializer()\n",
    "with tf.Session() as s:\n",
    "    init.run()\n",
    "    print( f.eval() )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true,
    "deletable": true,
    "editable": true
   },
   "source": [
    "## And now for something more interesting: Handwriting recognition\n",
    "\n",
    "The standard example for machine learning these days is the MNIST data set, a collection of 70,000 handwriting samples of the numbers 0-9. Our challenge - to predict which number each handwritten image represents.\n",
    "\n",
    "Although we'll talk about neural networks that are specifically well suited for image recognition later, we actually don't need to go there for this relatively simple task. We can achieve decent without a whole lot of code.\n",
    "\n",
    "Each image is 28x28 grayscale pixels, so we can treat each image as just a 1D array, or tensor, of 768 numbers. As long as we're consistent in how we flatten each image into an array, it'll still work. Yes, it would be even better if we could preserve the 2D structure of the data while training - but we'll get there later.\n",
    "\n",
    "Let's start by importing the data set, which conveniently is part of tensorflow itself:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "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": [
    "import tensorflow as tf\n",
    "from tensorflow.examples.tutorials.mnist import input_data\n",
    "\n",
    "sess = tf.InteractiveSession()\n",
    "\n",
    "mnist = input_data.read_data_sets(\"MNIST_data/\", one_hot=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "MNIST provides 55,000 samples in a training data set, 10,000 samples in a test data set, and 5,000 samples in a \"validation\" data set. We haven't talked about validation sets before, but their intent is to be used for model selection. So you'd use validation data to select your model, train the model with the training set, and then evaluate the model using the test data set.\n",
    "\n",
    "If you're new to the concept of train/test - it's important to evaluate the performance of our neural network using data it's never seen before. Otherwise it's kinda like giving students a math test for problems they already have the answers for. So, we use a completely different set of images to train our neural network from the images used for testing its accuracy.\n",
    "\n",
    "The training data is therefore a tensor of shape [55,000, 784] - 55,000 instances of 784 numbers that represent each image.\n",
    "\n",
    "The test data is encoded as \"one_hot\" when we loaded it above. Think of one_hot as a binary representation of the label data - that is, which number each handwriting sample was intended to represent. Mathematically one_hot represents a dimension for every possible label value. Every dimension is set to the value 0, except for the \"correct\" one which is set to 1. For example, the label vector representing the number 1 would be [0, 1, 0, 0, 0, 0, 0, 0, 0, 0] (remember we start counting at 0.) It's just a format that's optimized for how the labels are applied during training.\n",
    "\n",
    "So the test data is a tensor of shape [55,000, 10] - 55,000 test images each associated with 10 binary values that indicate whether or not the image represents a given number from 0-9.\n",
    "\n",
    "Let's define a little function to let us visualize what the input data looks like, and pick some random training image to see what it is we're up against:\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[ 0.  0.  0.  0.  0.  0.  0.  0.  0.  1.]\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAEICAYAAACQ6CLfAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nOfxmiXL4zRL1/87IS889LY/lAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x19a72d7d828>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "\n",
    "def display_sample(num):\n",
    "    #Print the one-hot array of this sample's label \n",
    "    print(mnist.train.labels[num])  \n",
    "    #Print the label converted back to a number\n",
    "    label = mnist.train.labels[num].argmax(axis=0)\n",
    "    #Reshape the 768 values to a 28x28 image\n",
    "    image = mnist.train.images[num].reshape([28,28])\n",
    "    plt.title('Sample: %d  Label: %d' % (num, label))\n",
    "    plt.imshow(image, cmap=plt.get_cmap('gray_r'))\n",
    "    plt.show()\n",
    "    \n",
    "display_sample(1234)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "So, you can see the \"one_hot\" representation of the number this image represents in the array that was printed out from the training label for image# 1234, as well as what this particular sample looks like. You can tell that some of the training data would even be challenging for a human to classify!\n",
    "\n",
    "Go ahead and try different input images to get a feel of the data we're given. Any value between 0 and 55,000 will work.\n",
    "\n",
    "As a reminder, we're flattening each image to a 1D array of 784 (28 x 28) numerical values. Each one of those values will be an input node into our deep neural network. Let's visualize how the data is being fed into it just to drive that point home:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXoAAAD4CAYAAADiry33AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x19a7084a588>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import numpy as np\n",
    "\n",
    "images = mnist.train.images[0].reshape([1,784])\n",
    "for i in range(1, 500):\n",
    "    images = np.concatenate((images, mnist.train.images[i].reshape([1,784])))\n",
    "plt.imshow(images, cmap=plt.get_cmap('gray_r'))\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "This is showing the first 500 training samples, one on each row. Imagine each pixel on each row getting fed into the bottom layer of a neural network 768 neurons (or \"units\") wide as we train our neural network.\n",
    "\n",
    "So let's start setting up that artificial neural network. We'll start by creating \"placeholders\" for the input images and for the \"correct\" labels for each. Think of these as parameters - we build up our neural network model without knowledge of the actual data that will be fed into it; we just need to construct it in such a way that our data will fit in.\n",
    "\n",
    "So our \"input_images\" placeholder will be set up to hold an array of values that consist of 784 floats (28x28), and our \"target_labels\" placeholder will be set up to hold an array of values that consist of 10 floats (one-hot format for 10 digits.)\n",
    "\n",
    "While training, we'll assign input_images to the training images and target_labels to the training lables. While testing, we'll use the test images and test labels instead."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "input_images = tf.placeholder(tf.float32, shape=[None, 784])\n",
    "target_labels = tf.placeholder(tf.float32, shape=[None, 10])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "So let's set up our deep neural network. We'll need an input layer with one node per input pixel per image, or 784 nodes. That will feed into a hidden layer of some arbitrary size - let's pick 512. That hidden layer will output 10 values, corresonding to scores for each classification to be fed into softmax.\n",
    "\n",
    "We'll need to reserve variables to keep track of the all the weights and biases for both layers:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "hidden_nodes = 512\n",
    "\n",
    "input_weights = tf.Variable(tf.truncated_normal([784, hidden_nodes]))\n",
    "input_biases = tf.Variable(tf.zeros([hidden_nodes]))\n",
    "\n",
    "hidden_weights = tf.Variable(tf.truncated_normal([hidden_nodes, 10]))\n",
    "hidden_biases = tf.Variable(tf.zeros([10]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "Now let's set up the neural network itself. We'll define the input layer and associate it with our placeholder for input data. All this layer does is multiply these inputs by our input_weight tensor which will be learned over time.\n",
    "\n",
    "Then we'll feed that into our hidden layer, which applies the ReLU activation function to the weighted inputs with our learned biases added in as well.\n",
    "\n",
    "Finally our output layer, called digit_weights, multiplies in the learned weights of the hidden layer and adds in the hidden layer's bias term."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "input_layer = tf.matmul(input_images, input_weights)\n",
    "hidden_layer = tf.nn.relu(input_layer + input_biases)\n",
    "digit_weights = tf.matmul(hidden_layer, hidden_weights) + hidden_biases"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "Make sure you noodle on the above block. This sets up a deep neural network like the one we talked about in our slides.\n",
    "\n",
    "output layer\n",
    "\n",
    "hidden layer\n",
    "\n",
    "input layer\n",
    "\n",
    "Next we will define our loss function for use in measuring our progress in gradient descent: cross-entropy, which applies a logarithmic scale to penalize incorrect classifications much more than ones that are close. Remember digit_weights is the output of our final layer, and we're comparing that against the target labels used for training."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "loss_function = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits=digit_weights, labels=target_labels))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "Now we will set up our gradient descent optimizer, initializing it with an aggressive learning rate (0.5) and our loss function defined above.\n",
    "\n",
    "That learning rate is an example of a hyperparameter that may be worth experimenting with and tuning."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "optimizer = tf.train.GradientDescentOptimizer(0.5).minimize(loss_function)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "Next we'll want to train our neural network and measure its accuracy. First let's define some methods for measuring the accuracy of our trained model. \n",
    "\n",
    "correct_prediction will look at the output of our neural network (in digit_weights) and choose the label with the highest value, and see if that agrees with the target label given. During testing, digit_weights will be our prediction based on the test data we give the network, and target_labels is a placeholder that we will assign to our test labels. Ultimately this gives us a 1 for every correct classification, and a 0 for every incorrect classification.\n",
    "\n",
    "\"accuracy\" then takes the average of all the classifications to produce an overall score for our model's accuracy."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": true,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "correct_prediction = tf.equal(tf.argmax(digit_weights,1), tf.argmax(target_labels,1))\n",
    "accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "Let's train this thing and see how it works! \n",
    "\n",
    "We'll set up a Tensorflow session, and initialize our variables. Next we will train our network in 2000 steps (or \"epochs\") with batches of 100 samples from our training data. At each step, we assign the input_images placeholder to the current batch of training images, and the target_labels placeholder to the current batch of training labels.\n",
    "\n",
    "Once training is complete, we'll measure the accuracy of our model using the accuracy graph we defined above. While measuring accuracy, we assign the input_images placeholder to our test images, and the target_labels placeholder to our test labels."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training epoch 100\n",
      "Accuracy: 0.854\n",
      "Training epoch 200\n",
      "Accuracy: 0.8682\n",
      "Training epoch 300\n",
      "Accuracy: 0.8826\n",
      "Training epoch 400\n",
      "Accuracy: 0.8905\n",
      "Training epoch 500\n",
      "Accuracy: 0.8997\n",
      "Training epoch 600\n",
      "Accuracy: 0.9034\n",
      "Training epoch 700\n",
      "Accuracy: 0.9105\n",
      "Training epoch 800\n",
      "Accuracy: 0.9089\n",
      "Training epoch 900\n",
      "Accuracy: 0.9122\n",
      "Training epoch 1000\n",
      "Accuracy: 0.9119\n",
      "Training epoch 1100\n",
      "Accuracy: 0.9157\n",
      "Training epoch 1200\n",
      "Accuracy: 0.9224\n",
      "Training epoch 1300\n",
      "Accuracy: 0.9199\n",
      "Training epoch 1400\n",
      "Accuracy: 0.9207\n",
      "Training epoch 1500\n",
      "Accuracy: 0.9195\n",
      "Training epoch 1600\n",
      "Accuracy: 0.9194\n",
      "Training epoch 1700\n",
      "Accuracy: 0.9272\n",
      "Training epoch 1800\n",
      "Accuracy: 0.9257\n",
      "Training epoch 1900\n",
      "Accuracy: 0.9252\n",
      "Training epoch 2000\n",
      "Accuracy: 0.9267\n"
     ]
    }
   ],
   "source": [
    "tf.global_variables_initializer().run()\n",
    "\n",
    "for x in range(2000):\n",
    "    batch = mnist.train.next_batch(100)\n",
    "    optimizer.run(feed_dict={input_images: batch[0], target_labels: batch[1]})\n",
    "    if ((x+1) % 100 == 0):\n",
    "        print(\"Training epoch \" + str(x+1))\n",
    "        print(\"Accuracy: \" + str(accuracy.eval(feed_dict={input_images: mnist.test.images, target_labels: mnist.test.labels})))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "You should have about 92% accuracy. Not bad! But hey, we're just starting.\n",
    "\n",
    "Let's take a look at some of the misclassified images and see just how good or bad our model is, compared to what your own brain can do. We'll go through the first 100 test images and look at the ones that are misclassified:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAEICAYAAACQ6CLfAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x19a70851940>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAEICAYAAACQ6CLfAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x19a00956780>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAEICAYAAACQ6CLfAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nrkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x19a00d5c4e0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAEICAYAAACQ6CLfAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nB8e3DVSdAUFYAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x19a009317f0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAEICAYAAACQ6CLfAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\njPISgIXeAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x19a00bf6d68>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAEICAYAAACQ6CLfAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x19a00c48e48>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAEICAYAAACQ6CLfAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAEQVJREFUeJzt3X2QVfV9x/H3R8GmA5pA2DKrgmjCH2XSFtKVhsQmWlNF\n2gbpWCf8odhRSa3RRJNJHZIZaUcrQzUZO1HTNTpiTIymwUo6UirWhzFNjasliFrrw6yBZYUFYsSJ\nT8C3f9yz6WXZ+7D3nnvPXX6f18ydPff8zjm/757dzz1P996jiMDM0nNE0QWYWTEcfrNEOfxmiXL4\nzRLl8JslyuE3S5TDXzBJsySFpAnZ8/WSljWwnJmS3pR0ZP5Vtp+kCyQ93u55U+Lw10FSv6S3snDt\nkHSHpMmt6CsizoqINXXW9Omy+X4eEZMjYn8r6hrR98cl/VTSXkmbJZ1SZdqVku5qdU15kHR+9kJ8\nUdG1tIPDX78/i4jJwEeBHuBrIydQyWG9TiVNBX4E/APwAWA18CNJUwotrElZ/SuAZ4uupV0O63/U\nVoiIAWA98BEASY9IulbSj4FfASdJer+k2yQNShqQdM3w7rikIyVdL2mXpFeAPylffra8i8qeXyzp\n+Wwr+5ykj0r6DjCTUujelPSVUQ4fjpW0TtIeSS9JurhsmSsl3Svpzmy5z0rqqXMVfBzYERE/iIj9\nEXEXMAT8+VjXpaSrJL1c9rstOXQSfVPSLyX9j6TTyxoqruMGXQf8I7CriWWMKw7/GEmaASwC/rts\n9HnAcuBo4FXgDmAf8GFgHnAGMBzoi4E/zcb3AOdU6esvgJXA+cAxwGeA3RFxHvBzsr2RiFg9yuzf\nB7YBx2Z9/L2kPypr/0w2zQeAdcA3y/q9WdLN1dfEwaWSvRiO0cvAHwLvB/4WuEtSd1n7H2TTTAOu\nBtZmex5QfR0fXJz0r5Kuqli8NJ/S3+JbDfwO41dE+FHjAfQDbwKvUwr3zcBvZm2PAH9XNu104J3h\n9mzcUuDhbPg/gL8qazsDCGBC2fIuyoY3AF+oUtOny57PGl4OMAPYDxxd1n4dcEc2vBLYWNY2B3ir\nznXxQeAXwGeBicAy4ADwTxWmXwncVeeyNwGLs+ELgO2Aytp/SumFttY6vgB4vM4+jwT6gI+NXP+H\n+2NC7ZcHy5wdERsrtG0tGz6BUigGJQ2PO6JsmmNHTP9qlT5nUNryjdWxwJ6I2Duin/Jd+9fKhn8F\nvE/ShIjYV23BEbFb0tnA9ZReBDcAGyntZYyJpPOBKym9cAFMprSVHzYQWSLLfodjqb2Ox+Kvgc0R\n8V8NzDuuOfz5KP8H3UppqzStQpAGKYV62Mwqy90KfKiOPkfaDkyVdHTZC8BMYKDKPHWLiEeBkwGy\ncwyvADeMZRmSTgBuBU4HfhIR+yVtonQIMew4SSp7AZhJ6RCl1joei9OBT0lalD2fCsyTNDciPt/k\nsjuaj/lzFhGDwL8DN0g6RtIRkj4k6VPZJPcCl0s6PjvDXPFYFPg28GVJv59dSfhwFhqAHcBJFWrY\nCvwncJ2k90n6XeBCIJdLbpLmSZoo6RhKewBbI2JDlVmOyOoYfvwGMInSC9hQtsy/5NDzBr9FaV1N\nzM5//DbwQB3reCwuyJY7N3v0UTr/8NUGljWuOPytcT5wFPAcpePjfwaGT2TdSmlX+WfA08DaSguJ\niB8A1wLfA/YC/0JpywSlY/ivSXpd0pdHmX0ppd3p7cB9wNVVDlsOIulbkqqd/PoKpbPiW7Pfa+RZ\n+tFqeavs8XJEPEdpb+EnlF7Ifgf48Yj5ngBmZ31dC5wTEbuztmrreOTvs17SitHaIuL1iHht+AG8\nC7wREb+s8TuNezr4kMrMUuEtv1miHH6zRDn8Zoly+M0S1dbr/NOmTYtZs2a1s0uzpPT397Nr1y7V\nnrLJ8EtaCNxI6S2S346IVdWmnzVrFn19fc10aWZV9PTU+/msJnb7s09Q3QScRem94UslzWl0eWbW\nXs0c888HXoqIVyLiXUqfEFucT1lm1mrNhP84Dv4gxbZs3EEkLZfUJ6lvaGioie7MLE8tP9sfEb0R\n0RMRPV1dXa3uzszq1Ez4Bzj402nHk9Onxsys9ZoJ/5PAbEknSjqK0pc7rMunLDNrtYYv9UXEPkmf\np/QJtSOB2yMimS8/NBvvmrrOHxEPAA/kVIuZtZHf3muWKIffLFEOv1miHH6zRDn8Zoly+M0S5fCb\nJcrhN0uUw2+WKIffLFEOv1miHH6zRDn8Zoly+M0S5fCbJcrhN0uUw2+WKIffLFEOv1miHH6zRDn8\nZoly+M0S5fCbJcrhN0uUw2+WKIffLFEOv1miHH6zRDn8Zolq6i69ZrXMnj27Ytvq1aurzrtkyZK8\ny7EyTYVfUj+wF9gP7IuInjyKMrPWy2PLf1pE7MphOWbWRj7mN0tUs+EPYKOkpyQtH20CScsl9Unq\nGxoaarI7M8tLs+E/JSLmAmcBl0r65MgJIqI3Inoioqerq6vJ7swsL02FPyIGsp87gfuA+XkUZWat\n13D4JU2SdPTwMHAGsCWvwsystZo52z8duE/S8HK+FxH/lktVdtjI/j9G9eijj1ad19f5W6vh8EfE\nK8Dv5ViLmbWRL/WZJcrhN0uUw2+WKIffLFEOv1mi/JHew8Du3bsrtvX29lad9/LLL6/aPmnSpIZq\nss7nLb9Zohx+s0Q5/GaJcvjNEuXwmyXK4TdLlMNvlihf5z8M7Nmzp2LbihUrqs47b968qu0LFy5s\nqCbrfN7ymyXK4TdLlMNvliiH3yxRDr9Zohx+s0Q5/GaJcvjNEuXwmyXK4TdLlMNvliiH3yxRDr9Z\nohx+s0Q5/GaJ8uf5DwMPP/xww/MODAzkWMmhuru7C+vbqqu55Zd0u6SdkraUjZsq6UFJL2Y/p7S2\nTDPLWz27/XcAI7/O5SrgoYiYDTyUPTezcaRm+CPiMWDk90QtBtZkw2uAs3Ouy8xarNETftMjYjAb\nfg2YXmlCScsl9UnqGxoaarA7M8tb02f7IyKAqNLeGxE9EdHT1dXVbHdmlpNGw79DUjdA9nNnfiWZ\nWTs0Gv51wLJseBlwfz7lmFm71LzOL+lu4FRgmqRtwNXAKuBeSRcCrwLntrJIq+6FF15oeN5p06bl\nWMmhTjvttIptN910U0v7tupqhj8illZoOj3nWsysjfz2XrNEOfxmiXL4zRLl8JslyuE3S5Q/0jsO\nvP3221Xb165dW7FtwoTqf+IpU/yBzFR5y2+WKIffLFEOv1miHH6zRDn8Zoly+M0S5fCbJcrX+ceB\na665pmp7f39/xbaJEydWnXf9+vVV2zdt2lS1vZZ77rmnYtuZZ57Z1LKtOd7ymyXK4TdLlMNvliiH\n3yxRDr9Zohx+s0Q5/GaJ8nX+ceCWW25peN733nuvavuqVasaXnazfJ2/WN7ymyXK4TdLlMNvliiH\n3yxRDr9Zohx+s0Q5/GaJ8nX+cSAiGp53wYIFVduvuOKKqu3d3d0N9w1wzjnnVGxr5vey5tXc8ku6\nXdJOSVvKxq2UNCBpU/ZY1NoyzSxv9ez23wEsHGX8NyJibvZ4IN+yzKzVaoY/Ih4D9rShFjNro2ZO\n+F0maXN2WFDxhm+Slkvqk9Q3NDTURHdmlqdGw38LcBIwFxgEbqg0YUT0RkRPRPR0dXU12J2Z5a2h\n8EfEjojYHxEHgFuB+fmWZWat1lD4JZVf/1kCbKk0rZl1pprX+SXdDZwKTJO0DbgaOFXSXCCAfuBz\nLawxeSeffHLV9ssuu6xi28KFo12o+X8TJjT3Vo9q9wwAeOeddyq2SWqqb2tOzb98RCwdZfRtLajF\nzNrIb+81S5TDb5Yoh98sUQ6/WaIcfrNE+SO948CGDRuKLqGiffv2VW0/cOBAmyqxsfKW3yxRDr9Z\nohx+s0Q5/GaJcvjNEuXwmyXK4TdLlK/zW1N2795dtf2NN95oUyU2Vt7ymyXK4TdLlMNvliiH3yxR\nDr9Zohx+s0Q5/GaJcvjNEuXwmyXK4TdLlMNvliiH3yxRDr9Zohx+s0Q5/GaJqucW3TOAO4HplG7J\n3RsRN0qaCtwDzKJ0m+5zI+IXrSvVOtH27dsbnnfBggU5VmJjVc+Wfx/wpYiYA3wMuFTSHOAq4KGI\nmA08lD03s3GiZvgjYjAins6G9wLPA8cBi4E12WRrgLNbVaSZ5W9Mx/ySZgHzgCeA6RExmDW9Rumw\nwMzGibrDL2ky8EPgixFx0BezRURQOh8w2nzLJfVJ6hsaGmqqWDPLT13hlzSRUvC/GxFrs9E7JHVn\n7d3AztHmjYjeiOiJiJ6urq48ajazHNQMvyQBtwHPR8TXy5rWAcuy4WXA/fmXZ2atUs9Xd38COA94\nRtKmbNwKYBVwr6QLgVeBc1tTonWyRx55pOF5TzzxxPwKsTGrGf6IeBxQhebT8y3HzNrF7/AzS5TD\nb5Yoh98sUQ6/WaIcfrNEOfxmiXL4zRLl8JslyuE3S5TDb5Yoh98sUQ6/WaIcfrNEOfxmiarn8/xm\nDZs5c2bFtjlz5rSxEhvJW36zRDn8Zoly+M0S5fCbJcrhN0uUw2+WKIffLFG+zm9NOeGEE6q2X3LJ\nJRXbJk+enHc5Ngbe8pslyuE3S5TDb5Yoh98sUQ6/WaIcfrNEOfxmiap5nV/SDOBOYDoQQG9E3Chp\nJXAxMJRNuiIiHmhVodaZrrzyyqJLsAbV8yaffcCXIuJpSUcDT0l6MGv7RkRc37ryzKxVaoY/IgaB\nwWx4r6TngeNaXZiZtdaYjvklzQLmAU9koy6TtFnS7ZKmVJhnuaQ+SX1DQ0OjTWJmBag7/JImAz8E\nvhgRbwC3ACcBcyntGdww2nwR0RsRPRHR09XVlUPJZpaHusIvaSKl4H83ItYCRMSOiNgfEQeAW4H5\nrSvTzPJWM/ySBNwGPB8RXy8b31022RJgS/7lmVmr1HO2/xPAecAzkjZl41YASyXNpXT5rx/4XEsq\nNLOWqOds/+OARmnyNX2zcczv8DNLlMNvliiH3yxRDr9Zohx+s0Q5/GaJcvjNEuXwmyXK4TdLlMNv\nliiH3yxRDr9Zohx+s0Q5/GaJUkS0rzNpCHi1bNQ0YFfbChibTq2tU+sC19aoPGs7ISLq+r68tob/\nkM6lvojoKayAKjq1tk6tC1xbo4qqzbv9Zoly+M0SVXT4ewvuv5pOra1T6wLX1qhCaiv0mN/MilP0\nlt/MCuLwmyWqkPBLWijpBUkvSbqqiBoqkdQv6RlJmyT1FVzL7ZJ2StpSNm6qpAclvZj9HPUeiQXV\ntlLSQLbuNklaVFBtMyQ9LOk5Sc9K+kI2vtB1V6WuQtZb24/5JR0J/C/wx8A24ElgaUQ819ZCKpDU\nD/REROFvCJH0SeBN4M6I+Eg2bjWwJyJWZS+cUyLibzqktpXAm0Xftj27m1R3+W3lgbOBCyhw3VWp\n61wKWG9FbPnnAy9FxCsR8S7wfWBxAXV0vIh4DNgzYvRiYE02vIbSP0/bVaitI0TEYEQ8nQ3vBYZv\nK1/ouqtSVyGKCP9xwNay59socAWMIoCNkp6StLzoYkYxPSIGs+HXgOlFFjOKmrdtb6cRt5XvmHXX\nyO3u8+YTfoc6JSLmAmcBl2a7tx0pSsdsnXSttq7btrfLKLeV/7Ui112jt7vPWxHhHwBmlD0/PhvX\nESJiIPu5E7iPzrv1+I7hOyRnP3cWXM+vddJt20e7rTwdsO466Xb3RYT/SWC2pBMlHQV8FlhXQB2H\nkDQpOxGDpEnAGXTercfXAcuy4WXA/QXWcpBOuW17pdvKU/C667jb3UdE2x/AIkpn/F8GvlpEDRXq\nOgn4WfZ4tujagLsp7Qa+R+ncyIXAB4GHgBeBjcDUDqrtO8AzwGZKQesuqLZTKO3SbwY2ZY9FRa+7\nKnUVst789l6zRPmEn1miHH6zRDn8Zoly+M0S5fCbJcrhN0uUw2+WqP8D1NBlCv110rgAAAAASUVO\nRK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x19a00d67e48>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAEICAYAAACQ6CLfAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nKIffLFEOv1miHH6zRDn8Zoly+M0S9f+3vMoXGcsn9QAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x19a00c51f60>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAEICAYAAACQ6CLfAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x19a00cc39e8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAEICAYAAACQ6CLfAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n5fCbZcrhN8vU/wMiV90j6MATtQAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x19a02e7be48>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "for x in range(100):\n",
    "    # Load a single test image and its label\n",
    "    x_train = mnist.test.images[x,:].reshape(1,784)\n",
    "    y_train = mnist.test.labels[x,:]\n",
    "    # Convert the one-hot label to an integer\n",
    "    label = y_train.argmax()\n",
    "    # Get the classification from our neural network's digit_weights final layer, and convert it to an integer\n",
    "    prediction = sess.run(digit_weights, feed_dict={input_images: x_train}).argmax()\n",
    "    # If the prediction does not match the correct label, display it\n",
    "    if (prediction != label) :\n",
    "        plt.title('Prediction: %d Label: %d' % (prediction, label))\n",
    "        plt.imshow(x_train.reshape([28,28]), cmap=plt.get_cmap('gray_r'))\n",
    "        plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "To be honest, I'd be a little unsure about some of those myself!\n",
    "\n",
    "## Exercise\n",
    "\n",
    "See if you can improve upon the accuracy. Try using more hidden neurons (nodes). Try using fewer! Try a different learning rate. Try adding another hidden layer. Try different batch sizes. What's the best accuracy you can get from this multi-layer perceptron?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true,
    "deletable": true,
    "editable": 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
}