Project 2 Handwritten Number Recognition with TFLearn and MNIST.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import tensorflow as tf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "curses is not supported on this machine (please install/reinstall curses for an optimal experience)\n"
     ]
    }
   ],
   "source": [
    "import tflearn\n",
    "import tflearn.datasets.mnist as mnist"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "## Shape of datsets - 55000 (image) , 28 X 28 ( each image is 28 X 28)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Extracting mnist/train-images-idx3-ubyte.gz\n",
      "Extracting mnist/train-labels-idx1-ubyte.gz\n",
      "Extracting mnist/t10k-images-idx3-ubyte.gz\n",
      "Extracting mnist/t10k-labels-idx1-ubyte.gz\n"
     ]
    }
   ],
   "source": [
    "trainX, trainY, testX, testY = mnist.load_data(one_hot=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(55000, 784)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "trainX.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAEICAYAAACQ6CLfAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nsGgAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x210070b62e8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "def display_digit(index):\n",
    "    label = trainY[index].argmax(axis=0)\n",
    "    image = trainX[index].reshape([28,28])\n",
    "    plt.title(label)\n",
    "    plt.imshow(image, cmap='gray_r')\n",
    "    \n",
    "display_digit(0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainY.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Build neural network\n",
    "* Build input layer - number of features - 28 X 28\n",
    "* Build hidden layer - Layers for feature identification\n",
    "* Build output layer - Predict the label of given image"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#creating a input layer of 100 features\n",
    "#net = tflearn.input_data([None, 100]) \n",
    "\n",
    "#Adding layers\n",
    "#n_units = 32\n",
    "#This adds fully connected layer where every unit in the previous layer is connected to each unit of this layer\n",
    "#net = tflearn.fully_connected(net, n_units, activation='ReLU')\n",
    "\n",
    "#net = tflearn.regression(net, optimizer='sgd', learning_rate=0.1, loss='categorical_cross_entropy')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def build_model():\n",
    "    tf.reset_default_graph()\n",
    "    \n",
    "    #Inputs\n",
    "    net = tflearn.input_data([None, trainX.shape[1]])\n",
    "    \n",
    "    #Hidden layers\n",
    "    net = tflearn.fully_connected(net, 128, activation='ReLU')\n",
    "    net = tflearn.fully_connected(net, 32, activation='ReLU')\n",
    "    \n",
    "    #Output Layer & training model\n",
    "    #activation for output layer is softmax\n",
    "    #\n",
    "    net = tflearn.fully_connected(net,10, activation='softmax')\n",
    "    net = tflearn.regression(net, optimizer='sgd', learning_rate=0.01, loss='categorical_crossentropy')\n",
    "    \n",
    "    model = tflearn.DNN(net)\n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "model = build_model()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training Step: 49499  | total loss: \u001b[1m\u001b[32m0.05478\u001b[0m\u001b[0m | time: 6.939s\n",
      "| SGD | epoch: 100 | loss: 0.05478 - acc: 0.9854 -- iter: 49400/49500\n",
      "Training Step: 49500  | total loss: \u001b[1m\u001b[32m0.05754\u001b[0m\u001b[0m | time: 7.992s\n",
      "| SGD | epoch: 100 | loss: 0.05754 - acc: 0.9838 | val_loss: 0.11520 - val_acc: 0.9685 -- iter: 49500/49500\n",
      "--\n"
     ]
    }
   ],
   "source": [
    "model.fit(trainX,trainY, validation_set=0.1, show_metric=True, batch_size=100, n_epoch=100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "pred = np.argmax(model.predict(testX),axis=1)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "act = np.argmax(testY,axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "9715"
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sum(pred == act)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<module 'tflearn.layers' from 'C:\\\\Users\\\\ZekeLabs\\\\Anaconda3-N\\\\lib\\\\site-packages\\\\tflearn\\\\layers\\\\__init__.py'>"
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Help on function fully_connected in module tflearn.layers.core:\n",
      "\n",
      "fully_connected(incoming, n_units, activation='linear', bias=True, weights_init='truncated_normal', bias_init='zeros', regularizer=None, weight_decay=0.001, trainable=True, restore=True, reuse=False, scope=None, name='FullyConnected')\n",
      "    Fully Connected.\n",
      "    \n",
      "    A fully connected layer.\n",
      "    \n",
      "    Input:\n",
      "        (2+)-D Tensor [samples, input dim]. If not 2D, input will be flatten.\n",
      "    \n",
      "    Output:\n",
      "        2D Tensor [samples, n_units].\n",
      "    \n",
      "    Arguments:\n",
      "        incoming: `Tensor`. Incoming (2+)D Tensor.\n",
      "        n_units: `int`, number of units for this layer.\n",
      "        activation: `str` (name) or `function` (returning a `Tensor`).\n",
      "            Activation applied to this layer (see tflearn.activations).\n",
      "            Default: 'linear'.\n",
      "        bias: `bool`. If True, a bias is used.\n",
      "        weights_init: `str` (name) or `Tensor`. Weights initialization.\n",
      "            (see tflearn.initializations) Default: 'truncated_normal'.\n",
      "        bias_init: `str` (name) or `Tensor`. Bias initialization.\n",
      "            (see tflearn.initializations) Default: 'zeros'.\n",
      "        regularizer: `str` (name) or `Tensor`. Add a regularizer to this\n",
      "            layer weights (see tflearn.regularizers). Default: None.\n",
      "        weight_decay: `float`. Regularizer decay parameter. Default: 0.001.\n",
      "        trainable: `bool`. If True, weights will be trainable.\n",
      "        restore: `bool`. If True, this layer weights will be restored when\n",
      "            loading a model.\n",
      "        reuse: `bool`. If True and 'scope' is provided, this layer variables\n",
      "            will be reused (shared).\n",
      "        scope: `str`. Define this layer scope (optional). A scope can be\n",
      "            used to share variables between layers. Note that scope will\n",
      "            override name.\n",
      "        name: A name for this layer (optional). Default: 'FullyConnected'.\n",
      "    \n",
      "    Attributes:\n",
      "        scope: `Scope`. This layer scope.\n",
      "        W: `Tensor`. Variable representing units weights.\n",
      "        b: `Tensor`. Variable representing biases.\n",
      "\n"
     ]
    }
   ],
   "source": [
    "help(tflearn.fully_connected)"
   ]
  },
  {
   "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": "3.5.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}