script.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "r0V4hVXE7fDO"
   },
   "source": [
    "# Pruning on Fashion MNIST"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 122
    },
    "colab_type": "code",
    "id": "XOsKqRy_8fNU",
    "outputId": "e6aebdc1-bc5c-4a98-ca83-53dbd153f297"
   },
   "outputs": [],
   "source": [
    "# from google.colab import drive\n",
    "# drive.mount('/content/drive')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "u2o5bSZZ7fDS"
   },
   "source": [
    "### Imports"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 85
    },
    "colab_type": "code",
    "id": "zLedBPQb7fDT",
    "outputId": "c3368dcc-3ff2-444c-92b8-e05732d0d999"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.5/dist-packages/h5py/__init__.py:36: 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"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From <ipython-input-2-3e5edf6e41e0>:14: read_data_sets (from tensorflow.contrib.learn.python.learn.datasets.mnist) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Please use alternatives such as official/mnist/dataset.py from tensorflow/models.\n",
      "WARNING:tensorflow:From /usr/local/lib/python3.5/dist-packages/tensorflow/contrib/learn/python/learn/datasets/mnist.py:260: maybe_download (from tensorflow.contrib.learn.python.learn.datasets.base) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Please write your own downloading logic.\n",
      "WARNING:tensorflow:From /usr/local/lib/python3.5/dist-packages/tensorflow/contrib/learn/python/learn/datasets/mnist.py:262: extract_images (from tensorflow.contrib.learn.python.learn.datasets.mnist) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Please use tf.data to implement this functionality.\n",
      "Extracting input/data/train-images-idx3-ubyte.gz\n",
      "WARNING:tensorflow:From /usr/local/lib/python3.5/dist-packages/tensorflow/contrib/learn/python/learn/datasets/mnist.py:267: extract_labels (from tensorflow.contrib.learn.python.learn.datasets.mnist) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Please use tf.data to implement this functionality.\n",
      "Extracting input/data/train-labels-idx1-ubyte.gz\n",
      "WARNING:tensorflow:From /usr/local/lib/python3.5/dist-packages/tensorflow/contrib/learn/python/learn/datasets/mnist.py:110: dense_to_one_hot (from tensorflow.contrib.learn.python.learn.datasets.mnist) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Please use tf.one_hot on tensors.\n",
      "Extracting input/data/t10k-images-idx3-ubyte.gz\n",
      "Extracting input/data/t10k-labels-idx1-ubyte.gz\n",
      "WARNING:tensorflow:From /usr/local/lib/python3.5/dist-packages/tensorflow/contrib/learn/python/learn/datasets/mnist.py:290: DataSet.__init__ (from tensorflow.contrib.learn.python.learn.datasets.mnist) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Please use alternatives such as official/mnist/dataset.py from tensorflow/models.\n"
     ]
    }
   ],
   "source": [
    "# Import libraries\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "import tensorflow as tf\n",
    "from tensorflow.python.framework import ops\n",
    "from timeit import default_timer as timer\n",
    "\n",
    "%matplotlib inline\n",
    "%config InlineBackend.figure_format='retina'\n",
    "\n",
    "# Import Fashion MNIST data with one-hot encoding\n",
    "from tensorflow.examples.tutorials.mnist import input_data\n",
    "fashion_mnist = input_data.read_data_sets('input/data', one_hot=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "vJTB48A_7fDb"
   },
   "source": [
    "### Data format\n",
    "\n",
    "Now that we've imported our dataset, we can take a look at what format our data is in, and check the shapes of each example image and label array."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 51
    },
    "colab_type": "code",
    "id": "MYEue-Pw7fDc",
    "outputId": "16996e6a-2229-4ec0-e843-10100834ebd5"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fashion MNIST: <class 'tensorflow.contrib.learn.python.learn.datasets.base.Datasets'>\n"
     ]
    }
   ],
   "source": [
    "# Dataset type\n",
    "print(\"Fashion MNIST: {type}\".format(type=type(fashion_mnist)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 212
    },
    "colab_type": "code",
    "id": "TCHJl0bP7fDg",
    "outputId": "aae3368b-dfab-4f77-eaec-b2d635b33282"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fashion MNIST:\n",
      "Training set (images) shape: (55000, 784)\n",
      "Training set (labels) shape: (55000, 10)\n",
      "Test set (images) shape: (10000, 784)\n",
      "Test set (labels) shape: (10000, 10)\n"
     ]
    }
   ],
   "source": [
    "# Shapes of training set and test set\n",
    "print(\"Fashion MNIST:\")\n",
    "print(\"Training set (images) shape: {shape}\".format(shape=fashion_mnist.train.images.shape))\n",
    "print(\"Training set (labels) shape: {shape}\".format(shape=fashion_mnist.train.labels.shape))\n",
    "print(\"Test set (images) shape: {shape}\".format(shape=fashion_mnist.test.images.shape))\n",
    "print(\"Test set (labels) shape: {shape}\".format(shape=fashion_mnist.test.labels.shape))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "Y7RySPM97fDm",
    "outputId": "dce6b2d1-9c23-4c5b-cc81-c581d41694ca"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(784,)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Shape of first image in training set\n",
    "fashion_mnist.train.images[0].shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "B1uNunRr7fDn",
    "outputId": "7d89ba85-488e-492d-e5db-2b2911fce3cd"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0., 0., 0., 0., 1., 0., 0., 0., 0., 0.])"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Format of labels\n",
    "fashion_mnist.train.labels[0]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "d5cHZ7tw7fDq"
   },
   "source": [
    "### Target classes\n",
    "\n",
    "The target class labels in the dataset are arrays, and the index of the 1 signifies the piece of clothing to which the image corresponds.\n",
    "\n",
    "Let's create a dictionary of these integer labels with their associated text labels. \n",
    "\n",
    "**Note: ** Mapping between labels and text taken from the original [Fashion MNIST](https://github.com/zalandoresearch/fashion-mnist) repository."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "6qOeJ8nU7fDr"
   },
   "outputs": [],
   "source": [
    "# Dictionary of labels\n",
    "label_dict = {\n",
    "    0: 'T-shirt/top',\n",
    "    1: 'Trouser',\n",
    "    2: 'Pullover',\n",
    "    3: 'Dress',\n",
    "    4: 'Coat',\n",
    "    5: 'Sandal',\n",
    "    6: 'Shirt',\n",
    "    7: 'Sneaker',\n",
    "    8: 'Bag',\n",
    "    9: 'Ankle boot'\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 547
    },
    "colab_type": "code",
    "id": "Ehto8X2S7fDu",
    "outputId": "f29d90bf-6bcf-49ae-a422-ae28ed1f130d"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "y = 3 (Dress)\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fb9a2ef1e80>"
      ]
     },
     "metadata": {
      "image/png": {
       "height": 251,
       "width": 254
      }
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "y = 7 (Sneaker)\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fb9a0e81f60>"
      ]
     },
     "metadata": {
      "image/png": {
       "height": 251,
       "width": 254
      }
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Get random images and associated label\n",
    "sample_1 = fashion_mnist.train.images[47].reshape(28,28)\n",
    "sample_label_1 = np.where(fashion_mnist.train.labels[47] == 1)[0][0]\n",
    "\n",
    "sample_2 = fashion_mnist.train.images[23].reshape(28,28)\n",
    "sample_label_2 = np.where(fashion_mnist.train.labels[23] == 1)[0][0]\n",
    "\n",
    "# Show images and labels\n",
    "print(\"y = {label_index} ({label})\".format(label_index=sample_label_1, label=label_dict[sample_label_1]))\n",
    "plt.imshow(sample_1, cmap='Greys')\n",
    "plt.show()\n",
    "\n",
    "print(\"y = {label_index} ({label})\".format(label_index=sample_label_2, label=label_dict[sample_label_2]))\n",
    "plt.imshow(sample_2, cmap='Greys')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "xkTLAEdS7fDv"
   },
   "source": [
    "## Building our neural network"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "nmfeNOrR7fDw"
   },
   "source": [
    "### Setting network parameters\n",
    "\n",
    "First, let's set some parameters for our network. This includes information about the number of units in each layer, shape of inputs, number of output classes, and number of examples."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "jFK3baix7fDx"
   },
   "outputs": [],
   "source": [
    "# Network parameters\n",
    "n_hidden_1 = 1000 # Units in first hidden layer\n",
    "n_hidden_2 = 1000 # Units in second hidden layer\n",
    "n_hidden_3 = 500 # Units in third hidden layer\n",
    "n_hidden_4 = 200 # Units in fourth hidden layer\n",
    "n_input = 784 # Fashion MNIST data input (img shape: 28*28)\n",
    "n_classes = 10 # Fashion MNIST total classes (0-9 digits)\n",
    "n_samples = fashion_mnist.train.num_examples"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "CWOxujHU7fDz"
   },
   "source": [
    "### Creating Placeholders\n",
    "\n",
    "Next, let's set up a function that creates placeholders for `X` and `Y`, using which we will later pass training data while running our network."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "Dl2gq-6G7fD0"
   },
   "outputs": [],
   "source": [
    "# Create placeholders\n",
    "def create_placeholders(n_x, n_y):\n",
    "    \"\"\"\n",
    "    Creates the placeholders for the tensorflow session.\n",
    "    \n",
    "    Arguments:\n",
    "    n_x -- scalar, size of an image vector (28*28 = 784)\n",
    "    n_y -- scalar, number of classes (from 0 to 9, so -> 10)\n",
    "    \n",
    "    Returns:\n",
    "    X -- placeholder for the data input, of shape [n_x, None] and dtype \"float\"\n",
    "    Y -- placeholder for the input labels, of shape [n_y, None] and dtype \"float\"\n",
    "    \"\"\"\n",
    "    \n",
    "    X = tf.placeholder(tf.float32, [n_x, None], name=\"X\")\n",
    "    Y = tf.placeholder(tf.float32, [n_y, None], name=\"Y\")\n",
    "    \n",
    "    return X, Y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 51
    },
    "colab_type": "code",
    "id": "Axxv8cIx7fD2",
    "outputId": "f16dc2a4-30aa-4909-ee54-0c730afb7d8d"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape of X: (784, ?)\n",
      "Shape of Y: (10, ?)\n"
     ]
    }
   ],
   "source": [
    "# Test create_placeholders()\n",
    "X, Y = create_placeholders(n_input, n_classes)\n",
    "\n",
    "print(\"Shape of X: {shape}\".format(shape=X.shape))\n",
    "print(\"Shape of Y: {shape}\".format(shape=Y.shape))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "RH9tLTSm7fD7"
   },
   "source": [
    "### Initializing parameters\n",
    "\n",
    "After creating placeholders, we will have to initialize the weights and biases for each layer in our 4-layer neural network.\n",
    "\n",
    "This implementation uses the **Xavier initialization** for weights"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "4xVjc_6M7fD7"
   },
   "outputs": [],
   "source": [
    "def initialize_parameters():\n",
    "    \"\"\"\n",
    "    Initializes parameters to build a neural network with tensorflow. The shapes are:\n",
    "                        W1 : [n_hidden_1, n_input]\n",
    "                        W2 : [n_hidden_2, n_hidden_1]\n",
    "                        W3 : [n_hidden_3, n_hidden_2]\n",
    "                        W4 : [n_hidden_4, n_hidden_3]\n",
    "                        W4 : [n_classes, n_hidden_4]\n",
    "         \n",
    "    Returns:\n",
    "    parameters -- a dictionary of tensors containing W1, W2, W3, W4, W5\n",
    "    \"\"\"\n",
    "    \n",
    "    # Set random seed for reproducibility\n",
    "    tf.set_random_seed(42)\n",
    "    \n",
    "    # Initialize weights and biases for each layer\n",
    "    W1 = tf.get_variable(\"W1\", [n_hidden_1, n_input], initializer=tf.contrib.layers.xavier_initializer(seed=42))\n",
    "    \n",
    "    W2 = tf.get_variable(\"W2\", [n_hidden_2, n_hidden_1], initializer=tf.contrib.layers.xavier_initializer(seed=42))\n",
    "    \n",
    "    W3 = tf.get_variable(\"W3\", [n_hidden_3, n_hidden_2], initializer=tf.contrib.layers.xavier_initializer(seed=42))\n",
    "    \n",
    "    W4 = tf.get_variable(\"W4\", [n_hidden_4, n_hidden_3], initializer=tf.contrib.layers.xavier_initializer(seed=42))\n",
    "    \n",
    "    W5 = tf.get_variable(\"W5\", [n_classes, n_hidden_4], initializer=tf.contrib.layers.xavier_initializer(seed=42))\n",
    "    \n",
    "    # Store initializations as a dictionary of parameters\n",
    "    parameters = {\n",
    "        \"W1\": W1,\n",
    "        \"W2\": W2,\n",
    "        \"W3\": W3,\n",
    "        \"W4\": W4,\n",
    "        \"W5\": W5\n",
    "    }\n",
    "\n",
    "    return parameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 85
    },
    "colab_type": "code",
    "id": "9NC67sZV7fD_",
    "outputId": "87a766a5-3e17-4a7d-e920-9f3f850f2387"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "W1 = <tf.Variable 'W1:0' shape=(1000, 784) dtype=float32_ref>\n",
      "W2 = <tf.Variable 'W2:0' shape=(1000, 1000) dtype=float32_ref>\n",
      "W3 = <tf.Variable 'W3:0' shape=(500, 1000) dtype=float32_ref>\n",
      "W4 = <tf.Variable 'W4:0' shape=(200, 500) dtype=float32_ref>\n",
      "W5 = <tf.Variable 'W5:0' shape=(10, 200) dtype=float32_ref>\n"
     ]
    }
   ],
   "source": [
    "# Test initialize_parameters()\n",
    "tf.reset_default_graph()\n",
    "with tf.Session() as sess:\n",
    "    parameters = initialize_parameters()\n",
    "    print(\"W1 = {w1}\".format(w1=parameters[\"W1\"]))\n",
    "    print(\"W2 = {w2}\".format(w2=parameters[\"W2\"]))\n",
    "    print(\"W3 = {w3}\".format(w3=parameters[\"W3\"]))\n",
    "    print(\"W4 = {w4}\".format(w4=parameters[\"W4\"]))\n",
    "    print(\"W5 = {w5}\".format(w5=parameters[\"W5\"]))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "pbXJiXxr7fED"
   },
   "source": [
    "As expected, our parameters haven't been initialized yet."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "9UQGoZhI7fEE"
   },
   "source": [
    "### Model\n",
    "\n",
    "`LINEAR -> RELU -> LINEAR -> RELU -> LINEAR -> SOFTMAX`\n",
    "\n",
    "`LINEAR`: $$ Z^{[l]} = W^{[l]}A^{[l-1]} $$\n",
    "\n",
    "`RELU`: $$ A^{[l]} = max ( 0, Z^{[l]} ) $$\n",
    "\n",
    "### Forward Propagation\n",
    "\n",
    "Next, let's implement forward propagation. This function will take in a dictionary of parameters and the input (`X`), to complete the forward pass. It will return the output from the last layer."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "aUEEk27C7fEF"
   },
   "outputs": [],
   "source": [
    "def forward_propagation(X, parameters):\n",
    "    \"\"\"\n",
    "    Implements the forward propagation for the model: LINEAR -> RELU -> LINEAR -> RELU -> LINEAR -> SOFTMAX\n",
    "    \n",
    "    Arguments:\n",
    "    X -- input dataset placeholder, of shape (input size, number of examples)\n",
    "    parameters -- python dictionary containing your parameters \"W1\", \"W2\", \"W3\", \"W4\", \"W5\"\n",
    "                  the shapes are given in initialize_parameters\n",
    "\n",
    "    Returns:\n",
    "    Z5 -- the output of the last LINEAR unit\n",
    "    \"\"\"\n",
    "    \n",
    "    # Retrieve parameters from dictionary\n",
    "    W1 = parameters['W1']\n",
    "    W2 = parameters['W2']\n",
    "    W3 = parameters['W3']\n",
    "    W4 = parameters['W4']\n",
    "    W5 = parameters['W5']\n",
    "    \n",
    "    # Carry out forward propagation      # Numpy equivalents:\n",
    "    Z1 = tf.matmul(W1,X)                 # Z1 = np.dot(W1, X)\n",
    "    A1 = tf.nn.relu(Z1)                  # A1 = relu(Z1)\n",
    "    Z2 = tf.matmul(W2,A1)                # Z2 = np.dot(W2, a1) \n",
    "    A2 = tf.nn.relu(Z2)                  # A2 = relu(Z2)\n",
    "    Z3 = tf.matmul(W3,A2)                # Z3 = np.dot(W3,Z2)\n",
    "    A3 = tf.nn.relu(Z3)\n",
    "    Z4 = tf.matmul(W4,A3)\n",
    "    A4 = tf.nn.relu(Z4)\n",
    "    Z5 = tf.matmul(W5,A4)\n",
    "    \n",
    "    return Z5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "def sparse_forward_propagation(X, parameters):\n",
    "    \"\"\"\n",
    "    Implements the forward propagation for the pruned model: LINEAR -> RELU -> LINEAR -> RELU -> LINEAR -> SOFTMAX\n",
    "    \"\"\"\n",
    "    \n",
    "    # Retrieve parameters from dictionary\n",
    "    W1 = parameters['W1']\n",
    "    W2 = parameters['W2']\n",
    "    W3 = parameters['W3']\n",
    "    W4 = parameters['W4']\n",
    "    W5 = parameters['W5']\n",
    " \n",
    "    #Converting Weight matrices to a sparse tensor since most of the values are 0\n",
    "    W1 = tf.contrib.layers.dense_to_sparse(W1)\n",
    "    W2 = tf.contrib.layers.dense_to_sparse(W2)\n",
    "    W3 = tf.contrib.layers.dense_to_sparse(W3)\n",
    "    W4 = tf.contrib.layers.dense_to_sparse(W4)\n",
    "    W5 = tf.contrib.layers.dense_to_sparse(W5)\n",
    "    \n",
    "    #Using sparse matrix multiplication to take advantage of the sparsity\n",
    "    Z1 = tf.sparse.matmul(W1,X)    \n",
    "    A1 = tf.nn.relu(Z1)                 \n",
    "    Z2 = tf.sparse.matmul(W2,A1) \n",
    "    A2 = tf.nn.relu(Z2)                \n",
    "    Z3 = tf.sparse.matmul(W3,A2)\n",
    "    A3 = tf.nn.relu(Z3)\n",
    "    Z4 = tf.sparse.matmul(W4,A3)\n",
    "    A4 = tf.nn.relu(Z4)\n",
    "    Z5 = tf.sparse.matmul(W5,A4)\n",
    "    \n",
    "    return Z5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "1_x2IMlu7fEH",
    "outputId": "b0c87c4f-5d98-429e-bbfb-89c108ebc223"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Z5 = Tensor(\"MatMul_4:0\", shape=(10, ?), dtype=float32)\n"
     ]
    }
   ],
   "source": [
    "# Test forward_propagation()\n",
    "tf.reset_default_graph()\n",
    "\n",
    "with tf.Session() as sess:\n",
    "    X, Y = create_placeholders(n_input, n_classes)\n",
    "    parameters = initialize_parameters()\n",
    "    Z5 = forward_propagation(X, parameters)\n",
    "    print(\"Z5 = {final_Z}\".format(final_Z=Z5))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "pJQD_rrC7fEK"
   },
   "source": [
    "### Compute Cost\n",
    "\n",
    "Next, let's create a function which computes the cost based on the difference between our outputs (`Z3`) and the correct labels (`Y`).\n",
    "\n",
    "`SOFTMAX: `\n",
    "\n",
    "`tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(...))` "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "_gzzfRPV7fEK"
   },
   "outputs": [],
   "source": [
    "def compute_cost(Z5, Y):\n",
    "    \"\"\"\n",
    "    Computes the cost\n",
    "    \n",
    "    Arguments:\n",
    "    Z5 -- output of forward propagation (output of the last LINEAR unit), of shape (10, number_of_examples)\n",
    "    Y -- \"true\" labels vector placeholder, same shape as Z5\n",
    "    \n",
    "    Returns:\n",
    "    cost - Tensor of the cost function\n",
    "    \"\"\"\n",
    "    \n",
    "    # Get logits (predictions) and labels\n",
    "    logits = tf.transpose(Z5)\n",
    "    labels = tf.transpose(Y)\n",
    "    \n",
    "    # Compute cost\n",
    "    cost = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits=logits, labels=labels))\n",
    "    \n",
    "    return cost"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "Xn1sHk8-7fEM",
    "outputId": "d662d58c-61f3-402a-ebea-652f14bd9360"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From <ipython-input-17-5f295e25dc76>:18: 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",
      "cost = Tensor(\"Mean:0\", shape=(), dtype=float32)\n"
     ]
    }
   ],
   "source": [
    "# Test compute_cost()\n",
    "tf.reset_default_graph()\n",
    "\n",
    "with tf.Session() as sess:\n",
    "    X, Y = create_placeholders(n_input, n_classes)\n",
    "    parameters = initialize_parameters()\n",
    "    Z5 = forward_propagation(X, parameters)\n",
    "    cost = compute_cost(Z5, Y)\n",
    "    print(\"cost = {cost}\".format(cost=cost))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "zV6Fvh_b7fEO"
   },
   "source": [
    "### Backpropagation\n",
    "\n",
    "We will use the Adam optimizer in this implementation.\n",
    "\n",
    "`optimizer = tf.train.AdamOptimizer(learning_rate).minimize(cost)`\n",
    "\n",
    "\n",
    "### Building the model\n",
    "\n",
    "Finally, let's combine our previously created functions into one single function called `model()` and run our network."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "HNPFIp867fEP"
   },
   "outputs": [],
   "source": [
    "def model(train, test, learning_rate=0.0001, num_epochs=16, minibatch_size=32, print_cost=True, graph_filename='costs'):\n",
    "    \"\"\"\n",
    "    Implements a four-layer tensorflow neural network: LINEAR->RELU->LINEAR->RELU->LINEAR->RELU->LINEAR->SOFTMAX.\n",
    "    \n",
    "    Arguments:\n",
    "    train -- training set\n",
    "    test -- test set\n",
    "    learning_rate -- learning rate of the optimization\n",
    "    num_epochs -- number of epochs of the optimization loop\n",
    "    minibatch_size -- size of a minibatch\n",
    "    print_cost -- True to print the cost every epoch\n",
    "    \n",
    "    Returns:\n",
    "    parameters -- parameters learnt by the model. They can then be used to predict.\n",
    "    \"\"\"\n",
    "    \n",
    "    # Ensure that model can be rerun without overwriting tf variables\n",
    "    ops.reset_default_graph()\n",
    "    # For reproducibility\n",
    "    tf.set_random_seed(42)\n",
    "    seed = 42\n",
    "    # Get input and output shapes\n",
    "    (n_x, m) = train.images.T.shape\n",
    "    n_y = train.labels.T.shape[0]\n",
    "    \n",
    "    costs = []\n",
    "    \n",
    "    # Create placeholders of shape (n_x, n_y)\n",
    "    X, Y = create_placeholders(n_x, n_y)\n",
    "    # Initialize parameters\n",
    "    parameters = initialize_parameters()\n",
    "    \n",
    "    # Forward propagation\n",
    "    Z5 = forward_propagation(X, parameters)\n",
    "    # Cost function\n",
    "    cost = compute_cost(Z5, Y)\n",
    "    # Backpropagation (using Adam optimizer)\n",
    "    optimizer = tf.train.AdamOptimizer(learning_rate).minimize(cost)\n",
    "    \n",
    "    # Initialize variables\n",
    "    init = tf.global_variables_initializer()\n",
    "    \n",
    "    # Start session to compute Tensorflow graph\n",
    "    with tf.Session() as sess:\n",
    "        \n",
    "        # Run initialization\n",
    "        sess.run(init)\n",
    "        \n",
    "        # Training loop\n",
    "        for epoch in range(num_epochs):\n",
    "            \n",
    "            epoch_cost = 0.\n",
    "            num_minibatches = int(m / minibatch_size)\n",
    "            seed = seed + 1\n",
    "            \n",
    "            for i in range(num_minibatches):\n",
    "                \n",
    "                # Get next batch of training data and labels\n",
    "                minibatch_X, minibatch_Y = train.next_batch(minibatch_size)\n",
    "                \n",
    "                # Execute optimizer and cost function\n",
    "                _, minibatch_cost = sess.run([optimizer, cost], feed_dict={X: minibatch_X.T, Y: minibatch_Y.T})\n",
    "                \n",
    "                # Update epoch cost\n",
    "                epoch_cost += minibatch_cost / num_minibatches\n",
    "                \n",
    "            # Print the cost every epoch\n",
    "            if print_cost == True:\n",
    "                print(\"Cost after epoch {epoch_num}: {cost}\".format(epoch_num=epoch, cost=epoch_cost))\n",
    "                costs.append(epoch_cost)\n",
    "        \n",
    "        # Plot costs\n",
    "        plt.figure(figsize=(16,5))\n",
    "        plt.plot(np.squeeze(costs), color='#2A688B')\n",
    "        plt.xlim(0, num_epochs-1)\n",
    "        plt.ylabel(\"cost\")\n",
    "        plt.xlabel(\"iterations\")\n",
    "        plt.title(\"learning rate = {rate}\".format(rate=learning_rate))\n",
    "        plt.savefig(graph_filename, dpi=300)\n",
    "        plt.show()\n",
    "        \n",
    "        # Save parameters\n",
    "        parameters = sess.run(parameters)\n",
    "        print(\"Parameters have been trained!\")\n",
    "        \n",
    "        # Calculate correct predictions\n",
    "        correct_prediction = tf.equal(tf.argmax(Z5), tf.argmax(Y))\n",
    "        \n",
    "        # Calculate accuracy on test set\n",
    "        accuracy = tf.reduce_mean(tf.cast(correct_prediction, \"float\"))\n",
    "        \n",
    "        print (\"Train Accuracy:\", accuracy.eval({X: train.images.T, Y: train.labels.T}))\n",
    "        print (\"Test Accuracy:\", accuracy.eval({X: test.images.T, Y: test.labels.T}))\n",
    "        #Saving the model\n",
    "        save_graph = tf.train.Saver()\n",
    "        save_graph.save(sess,'./model.pb')\n",
    "        return parameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 436
    },
    "colab_type": "code",
    "id": "qOl0vTCK7fEQ",
    "outputId": "b3750d9f-a7e2-4a8a-c8af-92b8568cb94d",
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Cost after epoch 0: 0.4848288826132663\n",
      "Cost after epoch 1: 0.37299732537777586\n",
      "Cost after epoch 2: 0.3317094536777907\n",
      "Cost after epoch 3: 0.30719639480599664\n",
      "Cost after epoch 4: 0.28736659362738726\n",
      "Cost after epoch 5: 0.27186093573719783\n",
      "Cost after epoch 6: 0.2602449401462835\n",
      "Cost after epoch 7: 0.2461389622578402\n",
      "Cost after epoch 8: 0.23742645514350516\n",
      "Cost after epoch 9: 0.22411346343390434\n",
      "Cost after epoch 10: 0.21664811436269013\n",
      "Cost after epoch 11: 0.2055310311334747\n",
      "Cost after epoch 12: 0.19826382283098007\n",
      "Cost after epoch 13: 0.1905522949704134\n",
      "Cost after epoch 14: 0.1843779995238976\n",
      "Cost after epoch 15: 0.1791839527792366\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fb9a0e3be80>"
      ]
     },
     "metadata": {
      "image/png": {
       "height": 335,
       "width": 948
      }
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Parameters have been trained!\n",
      "Train Accuracy: 0.9425091\n",
      "Test Accuracy: 0.8958\n"
     ]
    }
   ],
   "source": [
    "# Running our model on Fashion MNIST\n",
    "train = fashion_mnist.train\n",
    "test = fashion_mnist.test\n",
    "\n",
    "parameters = model(train, test, learning_rate=0.0005, graph_filename='fashion_mnist_costs')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_acc = []\n",
    "def eval_pruning(pruning_type,threshold=0,pruning_perc=0):\n",
    "    \"\"\"\n",
    "    Implements a function to test our weight and unit pruning methods\n",
    "    \n",
    "    Arguments:\n",
    "    pruning_type -- weight_pruing / unit_pruning\n",
    "    threshold -- used to zero out values above threshold\n",
    "    pruning_perc -- sparsity percentage\n",
    "    \"\"\"\n",
    "    #prune all layers except the last one\n",
    "    total_ops = len(tf.trainable_variables()) -1\n",
    "    i=0\n",
    "    with tf.Session() as sess:\n",
    "        #Restore the model\n",
    "        save_graph = tf.train.Saver()\n",
    "        save_graph.restore(sess,'./model.pb')\n",
    "        for v in tf.trainable_variables():\n",
    "            mask = np.ones(v.shape)\n",
    "            if(i < total_ops):\n",
    "                if(pruning_type == 'weight_pruing'):\n",
    "                    #getting the indices of weight values above threshold\n",
    "                    ids = (sess.run(np.abs(v))) > threshold\n",
    "                    mask = (ids.astype(float))\n",
    "                if(pruning_type == 'unit_pruing'):\n",
    "                    #computing L2-norm of weight matrix\n",
    "                    unit_norm = np.linalg.norm(sess.run(v),axis=0)\n",
    "                    #calculate threshold based on sparsity percentage\n",
    "                    threshold = np.percentile(unit_norm, pruning_perc)\n",
    "                    ids = unit_norm<threshold\n",
    "                    #setting the columns to zero\n",
    "                    mask[:,ids]=0\n",
    "                #setting the mask onto the weight matrix\n",
    "                sess.run(v.assign(tf.multiply(v,mask)))\n",
    "                i+=1\n",
    "                \n",
    "        #updating parameters in the dict\n",
    "        parameters = {\n",
    "            \"W1\": tf.trainable_variables()[0],\n",
    "            \"W2\": tf.trainable_variables()[1],\n",
    "            \"W3\": tf.trainable_variables()[2],\n",
    "            \"W4\": tf.trainable_variables()[3],\n",
    "            \"W5\": tf.trainable_variables()[4]\n",
    "        }\n",
    "        #Calculating test accuracy\n",
    "        (n_x, m) = train.images.T.shape\n",
    "        n_y = train.labels.T.shape[0]\n",
    "        X, Y = create_placeholders(n_x, n_y)\n",
    "        Z5 = sparse_forward_propagation(X, parameters)\n",
    "        correct_prediction = tf.equal(tf.argmax(Z5), tf.argmax(Y))     \n",
    "        accuracy = tf.reduce_mean(tf.cast(correct_prediction, \"float\"))\n",
    "        print (\"Test Accuracy:\", accuracy.eval({X: test.images.T, Y: test.labels.T}))  \n",
    "        test_acc.append(accuracy.eval({X: test.images.T, Y: test.labels.T}))     \n",
    "        save_graph = tf.train.Saver()\n",
    "        save_graph.save(sess,'./sparse_model.pb')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Restoring parameters from ./model.pb\n",
      "Test Accuracy: 0.8456\n",
      "[0.8456]\n"
     ]
    }
   ],
   "source": [
    "def unit_pruning(p_percentage):\n",
    "    \"\"\"\n",
    "    Implements Unit-pruning based on the L2-norm of column of weight matrix\n",
    "    \n",
    "    Arguments:\n",
    "    p_percentage: pruning percentage\n",
    "    \"\"\"\n",
    "    \n",
    "    eval_pruning('unit_pruing',pruning_perc=p_percentage)\n",
    "    \n",
    "\n",
    "unit_pruning(10)\n",
    "print(test_acc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 247
    },
    "colab_type": "code",
    "id": "QL40rs0meyGq",
    "outputId": "a67d5d80-9b4a-4bd7-eed7-2748f6362628",
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Restoring parameters from ./model.pb\n",
      "Test Accuracy: 0.8952\n"
     ]
    }
   ],
   "source": [
    "def weight_pruing(pruning_perc):\n",
    "    \"\"\"\n",
    "    Implements weight pruning, sets individual weights in the weight matrix to zero\n",
    "    \n",
    "    Arguments:\n",
    "    pruning_perc: pruning percentage\n",
    "    \"\"\"\n",
    "    \n",
    "    weights = []\n",
    "    for key,value in parameters.items():\n",
    "        weights += list(value.flatten())\n",
    "\n",
    "    threshold = np.percentile(np.abs(weights), pruning_perc)\n",
    "    eval_pruning('weight_pruing',threshold)\n",
    "\n",
    "weight_pruing(50)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Restoring parameters from ./model.pb\n",
      "Test Accuracy: 0.8958\n",
      "INFO:tensorflow:Restoring parameters from ./model.pb\n",
      "Test Accuracy: 0.8958\n",
      "INFO:tensorflow:Restoring parameters from ./model.pb\n",
      "Test Accuracy: 0.8945\n",
      "INFO:tensorflow:Restoring parameters from ./model.pb\n",
      "Test Accuracy: 0.6922\n",
      "INFO:tensorflow:Restoring parameters from ./model.pb\n",
      "Test Accuracy: 0.8952\n",
      "INFO:tensorflow:Restoring parameters from ./model.pb\n",
      "Test Accuracy: 0.5674\n",
      "INFO:tensorflow:Restoring parameters from ./model.pb\n",
      "Test Accuracy: 0.8933\n",
      "INFO:tensorflow:Restoring parameters from ./model.pb\n",
      "Test Accuracy: 0.4736\n",
      "INFO:tensorflow:Restoring parameters from ./model.pb\n",
      "Test Accuracy: 0.8874\n",
      "INFO:tensorflow:Restoring parameters from ./model.pb\n",
      "Test Accuracy: 0.434\n",
      "INFO:tensorflow:Restoring parameters from ./model.pb\n",
      "Test Accuracy: 0.8675\n",
      "INFO:tensorflow:Restoring parameters from ./model.pb\n",
      "Test Accuracy: 0.3547\n",
      "INFO:tensorflow:Restoring parameters from ./model.pb\n",
      "Test Accuracy: 0.7742\n",
      "INFO:tensorflow:Restoring parameters from ./model.pb\n",
      "Test Accuracy: 0.2096\n",
      "INFO:tensorflow:Restoring parameters from ./model.pb\n",
      "Test Accuracy: 0.4891\n",
      "INFO:tensorflow:Restoring parameters from ./model.pb\n",
      "Test Accuracy: 0.1285\n",
      "INFO:tensorflow:Restoring parameters from ./model.pb\n",
      "Test Accuracy: 0.2979\n",
      "INFO:tensorflow:Restoring parameters from ./model.pb\n",
      "Test Accuracy: 0.146\n",
      "INFO:tensorflow:Restoring parameters from ./model.pb\n",
      "Test Accuracy: 0.1157\n",
      "INFO:tensorflow:Restoring parameters from ./model.pb\n",
      "Test Accuracy: 0.055\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fb9a02b3898>"
      ]
     },
     "metadata": {
      "image/png": {
       "height": 335,
       "width": 951
      }
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "sparcity_perc = [0, 25, 50, 60, 70, 80, 90, 95, 97, 99]\n",
    "unit_pruning_acc = []\n",
    "weight_pruing_acc = []\n",
    "test_acc = []\n",
    "for k in sparcity_perc:\n",
    "    weight_pruing(k)\n",
    "    weight_pruing_acc.append(test_acc[0])\n",
    "    test_acc = []\n",
    "    unit_pruning(k)\n",
    "    unit_pruning_acc.append(test_acc[0])\n",
    "plt.figure(figsize=(16,5))\n",
    "plt.plot(sparcity_perc, weight_pruing_acc,  '-b', label='weight_pruing')\n",
    "plt.plot(sparcity_perc, unit_pruning_acc,  '-r', label='unit_pruning')\n",
    "plt.xlim(0, 100)\n",
    "plt.ylim(0,1.0)\n",
    "plt.ylabel(\"Accuracy\")\n",
    "plt.xlabel(\"k\")\n",
    "plt.title(\"Pruning\")\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Conclusion"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "From the above graph, we can draw the conclusion that weight pruning is better than unit pruning as by zeroing out the columns we could be losing critical features\n",
    "Even after pruning 90% of the weights the accuracy dropped from 89% to just 81%, and also reducing the model size and gaining computational speed-ups. \n",
    "\n",
    "The results hint at the possibility that the baseline models in these experiments are perhaps severely over-parameterized at the outset and a viable alternative for model compression might be to simply reduce the number of hidden units while maintaining the model's dense connection structure, exposing a similar trade-off in model size and accuracy.\n",
    "\n",
    "Also, I feel large-sparse is typically better than small-dense, for a fixed number of non-zero parameters or memory footprint.\n",
    "\n",
    "When you are time-constrained, it is better to train a simple, smaller network from scratch than prune a large network. It is the architectures obtained through the pruning process, not the learned weights that prove valuable. Such architectures are powerful when trained from scratch. Furthermore, these architectures are easy to approximate without any further pruning: we can prune once and obtain a family of new, scalable network architectures for different memory requirements."
   ]
  }
 ],
 "metadata": {
  "colab": {
   "name": "script.ipynb",
   "provenance": [],
   "toc_visible": true,
   "version": "0.3.2"
  },
  "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": 1
}