lab-10-6-mnist_nn_batchnorm.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Lab: Batchnormalization Layer\n",
    "\n",
    "## What is a batchnormalization layer?\n",
    "It is a layer that normalize the output before the activation layer. [The original paper](https://arxiv.org/abs/1502.03167) was proposed by Sergey Ioffe in 2015.\n",
    "\n",
    "Batch Normalization Layer looks like this: ![bn](https://kratzert.github.io/images/bn_backpass/bn_algorithm.PNG)\n",
    "\n",
    "## Why batchnormalization?\n",
    "The distribution of each layer's input changes because the weights of the previous layer change as we update weights by the gradient descent. This is called a covariance shift, which makes the network training difficult.\n",
    "\n",
    "For example, if the activation layer is a relu layer and the input of the activation layer is shifted to less than zeros, no weights will be activated!\n",
    "\n",
    "One thing also worth mentioning is that $\\gamma$ and $\\beta$ parameters in $$ y = \\gamma \\hat{x} + \\beta $$ are also trainable. \n",
    "\n",
    "**What it means is that if we don't need the batchnormalization, its parameters will be updated such that it offsets the normalization step.**\n",
    "\n",
    "For example, assume that\n",
    "\n",
    "\\begin{align}\n",
    "\\gamma &= \\sqrt{\\sigma^2_B + \\epsilon}\\\\\n",
    "\\beta &= \\mu_B\n",
    "\\end{align}\n",
    "\n",
    "then\n",
    "\n",
    "$$ y_i = \\gamma \\hat{x_i} + \\beta = x_i $$\n",
    "\n",
    "Also note that $\\mu$ and $\\sigma$ are computed using moving averages during the training step. However, during the test time, the computed $\\mu$ and $\\sigma$ will be used as fixed\n",
    "\n",
    "## Conclusion\n",
    "- Always use the batch normalization!"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Enough Talk: how to implement in Tensorflow"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 1. Load Library\n",
    "- We use the famous MNIST data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "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",
    "import matplotlib.pyplot as plt\n",
    "from tensorflow.examples.tutorials.mnist import input_data\n",
    "%matplotlib inline\n",
    "\n",
    "mnist = input_data.read_data_sets(\"MNIST_data/\", one_hot=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(55000, 784)"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "mnist.train.images.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2. Define Model & Solver Class\n",
    "- Object-Oriented-Programming allows to define multiple model easily\n",
    "- Why do we separate model and solver classes?\n",
    "  - We can just swap out the model class in the Solver class when we need a different network architecture\n",
    "  - Usually we need one solver class"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "class Model:\n",
    "    \"\"\"Network Model Class\n",
    "    \n",
    "    Note that this class has only the constructor.\n",
    "    The actual model is defined inside the constructor.\n",
    "    \n",
    "    Attributes\n",
    "    ----------\n",
    "    X : tf.float32\n",
    "        This is a tensorflow placeholder for MNIST images\n",
    "        Expected shape is [None, 784]\n",
    "        \n",
    "    y : tf.float32\n",
    "        This is a tensorflow placeholder for MNIST labels (one hot encoded)\n",
    "        Expected shape is [None, 10]\n",
    "        \n",
    "    mode : tf.bool\n",
    "        This is used for the batch normalization\n",
    "        It's `True` at training time and `False` at test time\n",
    "        \n",
    "    loss : tf.float32\n",
    "        The loss function is a softmax cross entropy\n",
    "        \n",
    "    train_op\n",
    "        This is simply the training op that minimizes the loss\n",
    "        \n",
    "    accuracy : tf.float32\n",
    "        The accuracy operation\n",
    "        \n",
    "    \n",
    "    Examples\n",
    "    ----------\n",
    "    >>> model = Model(\"Batch Norm\", 32, 10)\n",
    "\n",
    "    \"\"\"\n",
    "    def __init__(self, name, input_dim, output_dim, hidden_dims=[32, 32], use_batchnorm=True, activation_fn=tf.nn.relu, optimizer=tf.train.AdamOptimizer, lr=0.01):\n",
    "        \"\"\" Constructor\n",
    "        \n",
    "        Parameters\n",
    "        --------\n",
    "        name : str\n",
    "            The name of this network\n",
    "            The entire network will be created under `tf.variable_scope(name)`\n",
    "            \n",
    "        input_dim : int\n",
    "            The input dimension\n",
    "            In this example, 784\n",
    "        \n",
    "        output_dim : int\n",
    "            The number of output labels\n",
    "            There are 10 labels\n",
    "            \n",
    "        hidden_dims : list (default: [32, 32])\n",
    "            len(hidden_dims) = number of layers\n",
    "            each element is the number of hidden units\n",
    "            \n",
    "        use_batchnorm : bool (default: True)\n",
    "            If true, it will create the batchnormalization layer\n",
    "            \n",
    "        activation_fn : TF functions (default: tf.nn.relu)\n",
    "            Activation Function\n",
    "            \n",
    "        optimizer : TF optimizer (default: tf.train.AdamOptimizer)\n",
    "            Optimizer Function\n",
    "            \n",
    "        lr : float (default: 0.01)\n",
    "            Learning rate\n",
    "        \n",
    "        \"\"\"\n",
    "        with tf.variable_scope(name):\n",
    "            # Placeholders are defined\n",
    "            self.X = tf.placeholder(tf.float32, [None, input_dim], name='X')\n",
    "            self.y = tf.placeholder(tf.float32, [None, output_dim], name='y')\n",
    "            self.mode = tf.placeholder(tf.bool, name='train_mode')            \n",
    "            \n",
    "            # Loop over hidden layers\n",
    "            net = self.X\n",
    "            for i, h_dim in enumerate(hidden_dims):\n",
    "                with tf.variable_scope('layer{}'.format(i)):\n",
    "                    net = tf.layers.dense(net, h_dim)\n",
    "                    \n",
    "                    if use_batchnorm:\n",
    "                        net = tf.layers.batch_normalization(net, training=self.mode)\n",
    "                        \n",
    "                    net = activation_fn(net)\n",
    "            \n",
    "            # Attach fully connected layers\n",
    "            net = tf.contrib.layers.flatten(net)\n",
    "            net = tf.layers.dense(net, output_dim)\n",
    "            \n",
    "            self.loss = tf.nn.softmax_cross_entropy_with_logits(logits=net, labels=self.y)\n",
    "            self.loss = tf.reduce_mean(self.loss, name='loss')    \n",
    "            \n",
    "            # When using the batchnormalization layers,\n",
    "            # it is necessary to manually add the update operations\n",
    "            # because the moving averages are not included in the graph            \n",
    "            update_ops = tf.get_collection(tf.GraphKeys.UPDATE_OPS, scope=name)\n",
    "            with tf.control_dependencies(update_ops):                     \n",
    "                self.train_op = optimizer(lr).minimize(self.loss)\n",
    "            \n",
    "            # Accuracy etc \n",
    "            softmax = tf.nn.softmax(net, name='softmax')\n",
    "            self.accuracy = tf.equal(tf.argmax(softmax, 1), tf.argmax(self.y, 1))\n",
    "            self.accuracy = tf.reduce_mean(tf.cast(self.accuracy, tf.float32))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "class Solver:\n",
    "    \"\"\"Solver class\n",
    "    \n",
    "    This class will contain the model class and session\n",
    "    \n",
    "    Attributes\n",
    "    ----------\n",
    "    model : Model class\n",
    "    sess : TF session\n",
    "        \n",
    "    Methods\n",
    "    ----------\n",
    "    train(X, y)\n",
    "        Run the train_op and Returns the loss\n",
    "        \n",
    "    evalulate(X, y, batch_size=None)\n",
    "        Returns \"Loss\" and \"Accuracy\"\n",
    "        If batch_size is given, it's computed using batch_size\n",
    "        because most GPU memories cannot handle the entire training data at once\n",
    "            \n",
    "    Example\n",
    "    ----------\n",
    "    >>> sess = tf.InteractiveSession()\n",
    "    >>> model = Model(\"BatchNorm\", 32, 10)\n",
    "    >>> solver = Solver(sess, model)\n",
    "    \n",
    "    # Train\n",
    "    >>> solver.train(X, y)\n",
    "    \n",
    "    # Evaluate\n",
    "    >>> solver.evaluate(X, y)\n",
    "    \"\"\"\n",
    "    def __init__(self, sess, model):\n",
    "        self.model = model\n",
    "        self.sess = sess\n",
    "        \n",
    "    def train(self, X, y):\n",
    "        feed = {\n",
    "            self.model.X: X,\n",
    "            self.model.y: y,\n",
    "            self.model.mode: True\n",
    "        }\n",
    "        train_op = self.model.train_op\n",
    "        loss = self.model.loss\n",
    "        \n",
    "        return self.sess.run([train_op, loss], feed_dict=feed)\n",
    "    \n",
    "    def evaluate(self, X, y, batch_size=None):\n",
    "        if batch_size:\n",
    "            N = X.shape[0]\n",
    "            \n",
    "            total_loss = 0\n",
    "            total_acc = 0\n",
    "            \n",
    "            for i in range(0, N, batch_size):\n",
    "                X_batch = X[i:i + batch_size]\n",
    "                y_batch = y[i:i + batch_size]\n",
    "                \n",
    "                feed = {\n",
    "                    self.model.X: X_batch,\n",
    "                    self.model.y: y_batch,\n",
    "                    self.model.mode: False\n",
    "                }\n",
    "                \n",
    "                loss = self.model.loss\n",
    "                accuracy = self.model.accuracy\n",
    "                \n",
    "                step_loss, step_acc = self.sess.run([loss, accuracy], feed_dict=feed)\n",
    "                \n",
    "                total_loss += step_loss * X_batch.shape[0]\n",
    "                total_acc += step_acc * X_batch.shape[0]\n",
    "            \n",
    "            total_loss /= N\n",
    "            total_acc /= N\n",
    "            \n",
    "            return total_loss, total_acc\n",
    "            \n",
    "            \n",
    "        else:\n",
    "            feed = {\n",
    "                self.model.X: X,\n",
    "                self.model.y: y,\n",
    "                self.model.mode: False\n",
    "            }\n",
    "            \n",
    "            loss = self.model.loss            \n",
    "            accuracy = self.model.accuracy\n",
    "\n",
    "            return self.sess.run([loss, accuracy], feed_dict=feed)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3. Instantiate Model/Solver classes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "input_dim = 784\n",
    "output_dim = 10\n",
    "N = 55000\n",
    "\n",
    "tf.reset_default_graph()\n",
    "sess = tf.InteractiveSession()\n",
    "\n",
    "# We create two models: one with the batch norm and other without\n",
    "bn = Model('batchnorm', input_dim, output_dim, use_batchnorm=True)\n",
    "nn = Model('no_norm', input_dim, output_dim, use_batchnorm=False)\n",
    "\n",
    "# We create two solvers: to train both models at the same time for comparison\n",
    "# Usually we only need one solver class\n",
    "bn_solver = Solver(sess, bn)\n",
    "nn_solver = Solver(sess, nn)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "epoch_n = 10\n",
    "batch_size = 32\n",
    "\n",
    "# Save Losses and Accuracies every epoch\n",
    "# We are going to plot them later\n",
    "train_losses = []\n",
    "train_accs = []\n",
    "\n",
    "valid_losses = []\n",
    "valid_accs = []"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4. Run the train step"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[Epoch 0-TRAIN] Batchnorm Loss(Acc): 0.18456(94.19%) vs No Batchnorm Loss(Acc): 0.31917(91.01%)\n",
      "[Epoch 0-VALID] Batchnorm Loss(Acc): 0.19054(94.10%) vs No Batchnorm Loss(Acc): 0.31920(91.00%)\n",
      "\n",
      "[Epoch 1-TRAIN] Batchnorm Loss(Acc): 0.10349(96.78%) vs No Batchnorm Loss(Acc): 0.16142(95.34%)\n",
      "[Epoch 1-VALID] Batchnorm Loss(Acc): 0.11720(96.48%) vs No Batchnorm Loss(Acc): 0.18348(94.96%)\n",
      "\n",
      "[Epoch 2-TRAIN] Batchnorm Loss(Acc): 0.11239(96.43%) vs No Batchnorm Loss(Acc): 0.17737(94.79%)\n",
      "[Epoch 2-VALID] Batchnorm Loss(Acc): 0.12829(96.30%) vs No Batchnorm Loss(Acc): 0.20401(94.34%)\n",
      "\n",
      "[Epoch 3-TRAIN] Batchnorm Loss(Acc): 0.07526(97.69%) vs No Batchnorm Loss(Acc): 0.15240(95.65%)\n",
      "[Epoch 3-VALID] Batchnorm Loss(Acc): 0.09549(97.12%) vs No Batchnorm Loss(Acc): 0.20025(95.16%)\n",
      "\n",
      "[Epoch 4-TRAIN] Batchnorm Loss(Acc): 0.07339(97.68%) vs No Batchnorm Loss(Acc): 0.15641(95.53%)\n",
      "[Epoch 4-VALID] Batchnorm Loss(Acc): 0.10588(96.96%) vs No Batchnorm Loss(Acc): 0.19816(94.86%)\n",
      "\n",
      "[Epoch 5-TRAIN] Batchnorm Loss(Acc): 0.08164(97.38%) vs No Batchnorm Loss(Acc): 0.15969(95.67%)\n",
      "[Epoch 5-VALID] Batchnorm Loss(Acc): 0.11476(96.52%) vs No Batchnorm Loss(Acc): 0.22123(95.10%)\n",
      "\n",
      "[Epoch 6-TRAIN] Batchnorm Loss(Acc): 0.05879(98.10%) vs No Batchnorm Loss(Acc): 0.18191(94.92%)\n",
      "[Epoch 6-VALID] Batchnorm Loss(Acc): 0.09402(97.30%) vs No Batchnorm Loss(Acc): 0.25907(94.50%)\n",
      "\n",
      "[Epoch 7-TRAIN] Batchnorm Loss(Acc): 0.05014(98.38%) vs No Batchnorm Loss(Acc): 0.23831(93.59%)\n",
      "[Epoch 7-VALID] Batchnorm Loss(Acc): 0.08446(97.58%) vs No Batchnorm Loss(Acc): 0.28310(93.46%)\n",
      "\n",
      "[Epoch 8-TRAIN] Batchnorm Loss(Acc): 0.04956(98.41%) vs No Batchnorm Loss(Acc): 0.12616(96.48%)\n",
      "[Epoch 8-VALID] Batchnorm Loss(Acc): 0.08479(97.48%) vs No Batchnorm Loss(Acc): 0.18636(95.44%)\n",
      "\n",
      "[Epoch 9-TRAIN] Batchnorm Loss(Acc): 0.04351(98.61%) vs No Batchnorm Loss(Acc): 0.12277(96.54%)\n",
      "[Epoch 9-VALID] Batchnorm Loss(Acc): 0.08275(97.66%) vs No Batchnorm Loss(Acc): 0.19641(95.74%)\n",
      "\n"
     ]
    }
   ],
   "source": [
    "init = tf.global_variables_initializer()\n",
    "sess.run(init)\n",
    "\n",
    "for epoch in range(epoch_n):\n",
    "    for _ in range(N//batch_size):\n",
    "        X_batch, y_batch = mnist.train.next_batch(batch_size)\n",
    "        \n",
    "        _, bn_loss = bn_solver.train(X_batch, y_batch)\n",
    "        _, nn_loss = nn_solver.train(X_batch, y_batch)       \n",
    "    \n",
    "    b_loss, b_acc = bn_solver.evaluate(mnist.train.images, mnist.train.labels, batch_size)\n",
    "    n_loss, n_acc = nn_solver.evaluate(mnist.train.images, mnist.train.labels, batch_size)\n",
    "    \n",
    "    # Save train losses/acc\n",
    "    train_losses.append([b_loss, n_loss])\n",
    "    train_accs.append([b_acc, n_acc])\n",
    "    print(f'[Epoch {epoch}-TRAIN] Batchnorm Loss(Acc): {b_loss:.5f}({b_acc:.2%}) vs No Batchnorm Loss(Acc): {n_loss:.5f}({n_acc:.2%})')\n",
    "    \n",
    "    b_loss, b_acc = bn_solver.evaluate(mnist.validation.images, mnist.validation.labels)\n",
    "    n_loss, n_acc = nn_solver.evaluate(mnist.validation.images, mnist.validation.labels)\n",
    "    \n",
    "    # Save valid losses/acc\n",
    "    valid_losses.append([b_loss, n_loss])\n",
    "    valid_accs.append([b_acc, n_acc])\n",
    "    print(f'[Epoch {epoch}-VALID] Batchnorm Loss(Acc): {b_loss:.5f}({b_acc:.2%}) vs No Batchnorm Loss(Acc): {n_loss:.5f}({n_acc:.2%})')\n",
    "    print()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5. Performance Comparison\n",
    "* With the batchnormalization, the loss is lower and it's more accurate too!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[0.089340471, 0.97370011]"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "bn_solver.evaluate(mnist.test.images, mnist.test.labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[0.20733583, 0.95130014]"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "nn_solver.evaluate(mnist.test.images, mnist.test.labels)  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "def plot_compare(loss_list: list, ylim=None, title=None) -> None:\n",
    "    \n",
    "    bn = [i[0] for i in loss_list]\n",
    "    nn = [i[1] for i in loss_list]\n",
    "    \n",
    "    plt.figure(figsize=(15, 10))\n",
    "    plt.plot(bn, label='With BN')\n",
    "    plt.plot(nn, label='Without BN')\n",
    "    if ylim:\n",
    "        plt.ylim(ylim)\n",
    "        \n",
    "    if title:\n",
    "        plt.title(title)\n",
    "    plt.legend()\n",
    "    plt.grid('on')\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA3QAAAJOCAYAAAD/BkXEAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f4d9e3cf828>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_compare(train_losses, title='Training Loss at Epoch')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA20AAAJOCAYAAAAkve/mAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n/wBT7+ZSA8RytwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f4d990b1a20>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_compare(train_accs, [0, 1.0], title=\"Training Acc at Epoch\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA3QAAAJOCAYAAAD/BkXEAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f4d990b1b00>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_compare(valid_losses, title='Validation Loss at Epoch')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA20AAAJOCAYAAAAkve/mAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f4d96592c50>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_compare(valid_accs, [0, 1.], title='Validation Acc at Epoch')"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}