{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div align='left' style=\"width:400px;height:120px;overflow:hidden;\">\n",
    "<a href='http://www.uff.br'>\n",
    "<img align='left' style='display: block;height: 92%' src='imgs/UFF.png' alt='UFF logo' title='UFF logo'/>\n",
    "</a>\n",
    "<a href='http://www.ic.uff.br'>\n",
    "<img align='left' style='display: block;height: 100%' src='imgs/logo-ic.png' alt='IC logo' title='IC logo'/>\n",
    "</a>\n",
    "</div>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Machine Learning\n",
    "# 5. Programming Neural Networks with Theano"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### [Luis Martí](http://lmarti.com)\n",
    "#### [Instituto de Computação](http://www.ic.uff)\n",
    "#### [Universidade Federal Fluminense](http://www.uff.br)\n",
    "$\\newcommand{\\vec}[1]{\\boldsymbol{#1}}$"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os, sys, gzip\n",
    "import six.moves.cPickle as pickle\n",
    "\n",
    "import numpy\n",
    "\n",
    "import theano\n",
    "import theano.tensor as T\n",
    "import theano.d3viz as d3viz\n",
    "\n",
    "from IPython.display import Image, SVG, IFrame\n",
    "\n",
    "import matplotlib as mpl\n",
    "import matplotlib.pyplot as plt\n",
    "import matplotlib.cm as cm\n",
    "\n",
    "from mpl_toolkits.mplot3d import Axes3D"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "plt.rc('text', usetex=True); plt.rc('font', family='serif')\n",
    "plt.rcParams['text.latex.preamble'] ='\\\\usepackage{libertine}\\n\\\\usepackage[utf8]{inputenc}'\n",
    "\n",
    "import seaborn\n",
    "seaborn.set(style='whitegrid'); seaborn.set_context('talk')\n",
    "\n",
    "%matplotlib inline\n",
    "%config InlineBackend.figure_format = 'retina'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# What about `theano`?\n",
    "\n",
    "* Mathematical symbolic expression compiler\n",
    "  * Easy to define expressions\n",
    "  * Expressions are similar numpy's syntax and semantics\n",
    "* Possible to manipulate those expressions\n",
    "  * Substitutions\n",
    "  * **Gradient**, R operator\n",
    "* Stability optimizations\n",
    "Fast to compute values for those expressions\n",
    "* Speed optimizations\n",
    "* Use fast back-ends (GPUs (CUDA), BLAS, custom C code)\n",
    "* Tools to inspect and check for correctness"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Theano defines a *language*, a *compiler*, and a *library*.\n",
    "  * Define a symbolic expression\n",
    "  * Compile a function that can compute values\n",
    "  * Execute that function on numeric values"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The graph that represents mathematical operations is bipartite, and has two sorts of nodes:\n",
    "* **Variable nodes**, that represent data\n",
    "* **Apply nodes**, that represent the application of mathematical operations \n",
    "\n",
    "In practice:\n",
    "* Variables values are used for the graph inputs and outputs, and intermediate values\n",
    "* Variables will hold data during the function execution phase\n",
    "* An Apply node has inputs and outputs, which are Variables\n",
    "* An Apply node represents the specific application of an Op on these input variables\n",
    "* The same Variable can be used as inputs by several Apply nodes"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Shared variables\n",
    "* Symbolic variables, with a value associated to them\n",
    "*  The value is persistent across function calls\n",
    "*   The value is shared among all functions\n",
    "*   The variable has to be an input variable\n",
    "*   The variable is an implicit input to all functions using it"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "An optimization replaces a part of the graph with different nodes.\n",
    "* The types of the replaced nodes have to match\n",
    "* Merge equivalent computations\n",
    "* Simplify expressions: $\\frac{x}{x}$ becomes $1$\n",
    "* Numerical stability: For example, gives the right answer for  $\\log(1 + x)$  even if $x$ is really small.\n",
    "* Insert in-place an destructive versions of operations\n",
    "* Use specialized, high-performance versions (Elemwise loop fusion, GEMV,\n",
    "GEMM)\n",
    "* Shape inference\n",
    "* Constant folding\n",
    "* Transfer to GPU"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Using GPUs\n",
    "\n",
    "Theano tries to make the use of GPUs as transparent as possible,\n",
    "* Currently limited to `float32` `dtype`\n",
    "* Not easy to interact in Python with `CudaNdarrays`\n",
    "\n",
    "Select GPU by setting the device  flag to ’gpu’ or 'gpu{0,1,2,...}'.\n",
    "* All `float32` shared variables will be created in GPU memory\n",
    "* Enables optimizations moving supported operations to GPU\n",
    "\n",
    "You want to make sure you use `float32`\n",
    "* `floatX` is the default type of all tensors and sparse matrices.\n",
    "* By default, aliased to ’float64’ for double precision on CPU\n",
    "* Can be set to `float32` by a configuration  flag\n",
    "* You can always explicitly use `T.fmatrix()` or `T.matrix(dtype=’float32’)`"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Configuration  flags can be set in a couple of ways:\n",
    "* In the shell:\n",
    "```bash\n",
    "$ THEANO_FLAGS=device=gpu0,floatX=float32\n",
    "```\n",
    "* In Python:\n",
    "```python\n",
    "theano.config.device = 'gpu0'\n",
    "theano.config.floatX = 'float32'\n",
    "```\n",
    "* In the `.theanorc` configuration file:\n",
    "```\n",
    "[global]\n",
    "    device = gpu0\n",
    "    floatX = float32\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# A first contact with `theano`"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Defining theano variables and expressions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "x = T.dscalar('x')\n",
    "y = x ** 2"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "You might need to install `pydot` by running `pip install pydot`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "        <iframe\n",
       "            width=\"600\"\n",
       "            height=\"400\"\n",
       "            src=\"figs/y.html\"\n",
       "            frameborder=\"0\"\n",
       "            allowfullscreen\n",
       "        ></iframe>\n",
       "        "
      ],
      "text/plain": [
       "<IPython.lib.display.IFrame at 0x112835fd0>"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "d3viz.d3viz(y, 'figs'+os.sep+'y.html')\n",
    "IFrame('figs'+os.sep+'y.html', width=600, height=400)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Compiling the expression as a function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "f = theano.function(inputs=[x], outputs=y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Using the function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(16.0)"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "f(4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "dy = T.grad(y, x)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "        <iframe\n",
       "            width=\"1000\"\n",
       "            height=\"600\"\n",
       "            src=\"figs/dy.html\"\n",
       "            frameborder=\"0\"\n",
       "            allowfullscreen\n",
       "        ></iframe>\n",
       "        "
      ],
      "text/plain": [
       "<IPython.lib.display.IFrame at 0x112d6b978>"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "d3viz.d3viz(dy, 'figs'+os.sep+'dy.html')\n",
    "IFrame('figs'+os.sep+'dy.html', width=1000, height=600)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "df = theano.function([x], dy)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(8.0)"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df(4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "False"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "numpy.allclose(f(94.2), 188.4)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Programming Multilayer Perceptrons with `theano`\n",
    "\n",
    "I have based this notebook on the code of the [theano deep learning tutorial](http://deeplearning.net/tutorial/) (&copy; 2008–2013, Theano Development Team All rights reserved)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Loading data\n",
    "\n",
    "* We are going to use the [MNIST database](https://en.wikipedia.org/wiki/MNIST_database)\n",
    "* Each MNIST image has size 28*28\n",
    "* Contains 60,000 training images and 10,000 testing images.\n",
    "* That is too much data to process at one time, we need to make batches!\n",
    "\n",
    "<hr/>\n",
    "<div class=\"container-fluid\">\n",
    "  <div class=\"row\">\n",
    "      <div class=\"col-md-4\" align='center'> \n",
    "      </div>\n",
    "      <div class='col-md-4' align='center'>\n",
    "      <img src='http://myselph.de/mnistExamples.png'/>\n",
    "      </div>\n",
    "      <div class=\"col-md-4\" align='center'>\n",
    "      </div>\n",
    "  </div>\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def load_data(dataset):\n",
    "    ''' Loads the dataset\n",
    "\n",
    "    :type dataset: string\n",
    "    :param dataset: the path to the dataset (here MNIST)\n",
    "    '''\n",
    "\n",
    "    # Download the MNIST dataset if it is not present\n",
    "    data_dir, data_file = os.path.split(dataset)\n",
    "    if data_dir == \"\" and not os.path.isfile(dataset):\n",
    "        # Check if dataset is in the data directory.\n",
    "        new_path = os.path.join(\n",
    "            os.getcwd(),\n",
    "            \"data\",\n",
    "            dataset\n",
    "        )\n",
    "        if os.path.isfile(new_path) or data_file == 'mnist.pkl.gz':\n",
    "            dataset = new_path\n",
    "\n",
    "    if (not os.path.isfile(dataset)) and data_file == 'mnist.pkl.gz':\n",
    "        from six.moves import urllib\n",
    "        origin = (\n",
    "            'http://www.iro.umontreal.ca/~lisa/deep/data/mnist/mnist.pkl.gz'\n",
    "        )\n",
    "        print('Downloading data from %s' % origin)\n",
    "        urllib.request.urlretrieve(origin, dataset)\n",
    "\n",
    "    # Load the dataset\n",
    "    with gzip.open(dataset, 'rb') as f:\n",
    "        try:\n",
    "            train_set, valid_set, test_set = pickle.load(f, encoding='latin1')\n",
    "        except:\n",
    "            train_set, valid_set, test_set = pickle.load(f)\n",
    "    # train_set, valid_set, test_set format: tuple(input, target)\n",
    "    # input is a numpy.ndarray of 2 dimensions (a matrix)\n",
    "    # where each row corresponds to an example. target is a\n",
    "    # numpy.ndarray of 1 dimension (vector) that has the same length as\n",
    "    # the number of rows in the input. It should give the target\n",
    "    # to the example with the same index in the input.\n",
    "\n",
    "    def shared_dataset(data_xy, borrow=True):\n",
    "        \"\"\" Function that loads the dataset into shared variables\n",
    "\n",
    "        The reason we store our dataset in shared variables is to allow\n",
    "        Theano to copy it into the GPU memory (when code is run on GPU).\n",
    "        Since copying data into the GPU is slow, copying a minibatch everytime\n",
    "        is needed (the default behaviour if the data is not in a shared\n",
    "        variable) would lead to a large decrease in performance.\n",
    "        \"\"\"\n",
    "        data_x, data_y = data_xy\n",
    "        shared_x = theano.shared(numpy.asarray(data_x,\n",
    "                                               dtype=theano.config.floatX),\n",
    "                                 borrow=borrow)\n",
    "        shared_y = theano.shared(numpy.asarray(data_y,\n",
    "                                               dtype=theano.config.floatX),\n",
    "                                 borrow=borrow)\n",
    "        # When storing data on the GPU it has to be stored as floats\n",
    "        # therefore we will store the labels as ``floatX`` as well\n",
    "        # (``shared_y`` does exactly that). But during our computations\n",
    "        # we need them as ints (we use labels as index, and if they are\n",
    "        # floats it doesn't make sense) therefore instead of returning\n",
    "        # ``shared_y`` we will have to cast it to int. This little hack\n",
    "        # lets ous get around this issue\n",
    "        return shared_x, T.cast(shared_y, 'int32')\n",
    "\n",
    "    test_set_x, test_set_y = shared_dataset(test_set)\n",
    "    valid_set_x, valid_set_y = shared_dataset(valid_set)\n",
    "    train_set_x, train_set_y = shared_dataset(train_set)\n",
    "\n",
    "    rval = [(train_set_x, train_set_y), (valid_set_x, valid_set_y),\n",
    "            (test_set_x, test_set_y)]\n",
    "    return rval"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Programming a multi-class logistic regression\n",
    "\n",
    "* The logistic regression is fully described by a weight matrix $\\vec{W}$ and bias vector $\\vec{b}$. \n",
    "* Classification is done by projecting data points onto a set of hyperplanes, the distance to which is used to determine a class membership probability."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "class LogisticRegression(object):\n",
    "\n",
    "    def __init__(self, input, n_in, n_out):\n",
    "        # initialize with 0 the weights W as a matrix of shape (n_in, n_out)\n",
    "        self.W = theano.shared(value=numpy.zeros((n_in, n_out), dtype=theano.config.floatX),\n",
    "                               name='W', borrow=True)\n",
    "        # initialize the biases b as a vector of n_out 0s\n",
    "        self.b = theano.shared(value=numpy.zeros((n_out,), dtype=theano.config.floatX), \n",
    "                               name='b', borrow=True)\n",
    "\n",
    "        # symbolic expression for computing the matrix of class-membership\n",
    "        # probabilities\n",
    "        # Where:\n",
    "        # W is a matrix where column-k represent the separation hyperplane for\n",
    "        # class-k\n",
    "        # x is a matrix where row-j  represents input training sample-j\n",
    "        # b is a vector where element-k represent the free parameter of\n",
    "        # hyperplane-k\n",
    "        self.p_y_given_x = T.nnet.softmax(T.dot(input, self.W) + self.b)\n",
    "\n",
    "        # symbolic description of how to compute prediction as class whose\n",
    "        # probability is maximal\n",
    "        self.y_pred = T.argmax(self.p_y_given_x, axis=1)\n",
    "\n",
    "        # parameters of the model\n",
    "        self.params = [self.W, self.b]\n",
    "\n",
    "        # keep track of model input\n",
    "        self.input = input\n",
    "\n",
    "    def negative_log_likelihood(self, y):\n",
    "        \"\"\"Return the mean of the negative log-likelihood of the prediction\n",
    "        of this model under a given target distribution.\n",
    "\n",
    "        Note: we use the mean instead of the sum so that\n",
    "              the learning rate is less dependent on the batch size\n",
    "        \"\"\"\n",
    "        return -T.mean(T.log(self.p_y_given_x)[T.arange(y.shape[0]), y])\n",
    "\n",
    "    def errors(self, y):\n",
    "        \"\"\"Return a float representing the number of errors in the minibatch\n",
    "        over the total number of examples of the minibatch ; zero one\n",
    "        loss over the size of the minibatch\n",
    "\n",
    "        :type y: theano.tensor.TensorType\n",
    "        :param y: corresponds to a vector that gives for each example the\n",
    "                  correct label\n",
    "        \"\"\"\n",
    "\n",
    "        # check if y has same dimension of y_pred\n",
    "        if y.ndim != self.y_pred.ndim:\n",
    "            raise TypeError(\n",
    "                'y should have the same shape as self.y_pred',\n",
    "                ('y', y.type, 'y_pred', self.y_pred.type)\n",
    "            )\n",
    "        # check if y is of the correct datatype\n",
    "        if y.dtype.startswith('int'):\n",
    "            # the T.neq operator returns a vector of 0s and 1s, where 1\n",
    "            # represents a mistake in prediction\n",
    "            return T.mean(T.neq(self.y_pred, y))\n",
    "        else:\n",
    "            raise NotImplementedError()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "learning_rate=0.13\n",
    "n_epochs=100\n",
    "dataset='mnist.pkl.gz'\n",
    "batch_size=30"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Loading data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "datasets = load_data(dataset)\n",
    "\n",
    "train_set_x, train_set_y = datasets[0]\n",
    "valid_set_x, valid_set_y = datasets[1]\n",
    "test_set_x, test_set_y = datasets[2]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Compute number of minibatches for training, validation and testing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "n_train_batches = train_set_x.get_value(borrow=True).shape[0] // batch_size\n",
    "n_valid_batches = valid_set_x.get_value(borrow=True).shape[0] // batch_size\n",
    "n_test_batches = test_set_x.get_value(borrow=True).shape[0] // batch_size"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# allocate symbolic variables for the data\n",
    "index = T.lscalar()  # index to a [mini]batch\n",
    "\n",
    "# generate symbolic variables for input (x and y represent a\n",
    "# minibatch)\n",
    "x = T.matrix('x')  # data, presented as rasterized images\n",
    "y = T.ivector('y')  # labels, presented as 1D vector of [int] labels"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Construct the logistic regression class instance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "classifier = LogisticRegression(input=x, n_in=28 * 28, n_out=10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The cost function (error or loss function) we minimize during training is the negative log likelihood of the model in symbolic format."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "cost = classifier.negative_log_likelihood(y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Compiling a Theano function that computes the mistakes that are made by the model on a validation minibatch."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "validate_model = theano.function(inputs=[index], outputs=classifier.errors(y),\n",
    "                                 givens={x: valid_set_x[index * batch_size: (index + 1) * batch_size],\n",
    "                                         y: valid_set_y[index * batch_size: (index + 1) * batch_size]})"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "A similar function for testing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "test_model = theano.function(inputs=[index], outputs=classifier.errors(y),\n",
    "                             givens={x: test_set_x[index * batch_size: (index + 1) * batch_size],\n",
    "                                     y: test_set_y[index * batch_size: (index + 1) * batch_size]})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train_set_loss = theano.function(inputs=[index], outputs=classifier.errors(y),\n",
    "                                 givens={x: train_set_x[index * batch_size: (index + 1) * batch_size],\n",
    "                                         y: train_set_y[index * batch_size: (index + 1) * batch_size]})"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Compute the gradient of cost with respect to $\\vec{W}$ and $\\vec{b}$."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "g_W = T.grad(cost=cost, wrt=classifier.W)\n",
    "g_b = T.grad(cost=cost, wrt=classifier.b)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Specify how to update the parameters of the model as a list of (variable, update expression) pairs."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "updates = [(classifier.W, classifier.W - learning_rate * g_W),\n",
    "           (classifier.b, classifier.b - learning_rate * g_b)]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Compiling a Theano function `train_model` that returns the cost, but in the same time updates the parameter of the model based on the rules\n",
    "defined in `updates`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING (theano.gof.compilelock): Overriding existing lock by dead process '22753' (I am process '23028')\n"
     ]
    }
   ],
   "source": [
    "train_model = theano.function(inputs=[index], outputs=cost, updates=updates,\n",
    "                              givens={x: train_set_x[index * batch_size: (index + 1) * batch_size],\n",
    "                                      y: train_set_y[index * batch_size: (index + 1) * batch_size]})"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "All set! we are ready for training."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "scrolled": false
   },
   "outputs": [],
   "source": [
    "def train(classifier, train_model, validate_model, test_model, n_epochs, \n",
    "          n_train_batches, n_valid_batches, n_test_batches, \n",
    "          validation_frequency, patience=5000,\n",
    "          patience_increase=2, improvement_threshold=0.995,\n",
    "          verbose=False, log_results=False):\n",
    "    train_errors = []\n",
    "    validation_errors = []\n",
    "    test_errors = []\n",
    "    \n",
    "    epoch = 0\n",
    "    done_looping = False\n",
    "    \n",
    "    best_validation_loss = numpy.inf\n",
    "    test_score = 0.\n",
    "    while epoch < n_epochs and not done_looping:\n",
    "        epoch = epoch + 1\n",
    "        for minibatch_index in range(n_train_batches):\n",
    "            minibatch_avg_cost = train_model(minibatch_index)\n",
    "            # iteration number\n",
    "            iter_num = (epoch - 1) * n_train_batches + minibatch_index\n",
    "\n",
    "            # is it time to compute the validation set error?\n",
    "            if (iter_num + 1) % validation_frequency == 0:\n",
    "                validation_losses = [validate_model(i) for i in range(n_valid_batches)]\n",
    "                current_validation_loss = numpy.mean(validation_losses)\n",
    "\n",
    "                if log_results:\n",
    "                    # test it on the test set\n",
    "                    test_losses = [test_model(i) for i in range(n_test_batches)]\n",
    "                    current_test_loss = numpy.mean(test_losses)\n",
    "\n",
    "                    # test it on the trainig set\n",
    "                    train_losses = [train_set_loss(i) for i in range(n_train_batches)]\n",
    "                    current_train_loss = numpy.mean(train_losses)\n",
    "\n",
    "                    train_errors += [current_train_loss]\n",
    "                    validation_errors += [current_validation_loss]\n",
    "                    test_errors += [current_test_loss]\n",
    "\n",
    "                if verbose:\n",
    "                    print('epoch %i, minibatch %i/%i, validation error %f%%' %\n",
    "                          (epoch, minibatch_index + 1, n_train_batches, current_validation_loss * 100.))\n",
    "\n",
    "                # if we got the best validation score until now\n",
    "                if current_validation_loss < best_validation_loss:\n",
    "                    #improve patience if loss improvement is good enough\n",
    "                    if current_validation_loss < best_validation_loss * improvement_threshold:\n",
    "                        patience = max(patience, iter_num * patience_increase)\n",
    "\n",
    "                    best_validation_loss = current_validation_loss\n",
    "                    ## save the best model\n",
    "                    # with open('best_model.pkl', 'wb') as f:\n",
    "                    #     pickle.dump(classifier, f)\n",
    "\n",
    "            if patience <= iter_num:\n",
    "                print('Ran out of patience!')\n",
    "                done_looping = True\n",
    "                break\n",
    "    return train_errors, validation_errors, test_errors"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "patience = 5000  # look as this many examples regardless\n",
    "patience_increase = 2  # wait this much longer when a new best is found\n",
    "improvement_threshold = 0.995  # a relative improvement of this much is\n",
    "                               # considered significant\n",
    "validation_frequency = min(n_train_batches, patience // 2)\n",
    "                              # go through this many\n",
    "                              # minibatches before checking the network\n",
    "                              # on the validation set; in this case we\n",
    "                              # check every epoch\n",
    "n_epochs = 100"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 1, minibatch 1666/1666, validation error 8.548549%\n",
      "epoch 2, minibatch 1666/1666, validation error 8.118118%\n",
      "epoch 3, minibatch 1666/1666, validation error 7.877878%\n",
      "epoch 4, minibatch 1666/1666, validation error 7.727728%\n",
      "epoch 5, minibatch 1666/1666, validation error 7.507508%\n",
      "epoch 6, minibatch 1666/1666, validation error 7.367367%\n",
      "epoch 7, minibatch 1666/1666, validation error 7.387387%\n",
      "epoch 8, minibatch 1666/1666, validation error 7.307307%\n",
      "epoch 9, minibatch 1666/1666, validation error 7.257257%\n",
      "epoch 10, minibatch 1666/1666, validation error 7.247247%\n",
      "epoch 11, minibatch 1666/1666, validation error 7.267267%\n",
      "epoch 12, minibatch 1666/1666, validation error 7.247247%\n",
      "epoch 13, minibatch 1666/1666, validation error 7.227227%\n",
      "epoch 14, minibatch 1666/1666, validation error 7.237237%\n",
      "epoch 15, minibatch 1666/1666, validation error 7.277277%\n",
      "epoch 16, minibatch 1666/1666, validation error 7.247247%\n",
      "epoch 17, minibatch 1666/1666, validation error 7.257257%\n",
      "Ran out of patience!\n",
      "CPU times: user 20.6 s, sys: 498 ms, total: 21.1 s\n",
      "Wall time: 11.2 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "train_errors, validation_errors, test_errors = train(classifier, train_model, validate_model, test_model, \n",
    "                                                     n_epochs, \n",
    "                                                     n_train_batches, n_valid_batches, n_test_batches, \n",
    "                                                     validation_frequency, verbose=True,\n",
    "                                                     log_results=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.text.Text at 0x113276518>"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAykAAAIcCAYAAADyquHAAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x11323cc50>"
      ]
     },
     "metadata": {
      "image/png": {
       "height": 270,
       "width": 404
      }
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(train_errors, label='Training set')\n",
    "plt.plot(validation_errors, label='Validation set')\n",
    "plt.plot(test_errors, label='Test set')\n",
    "plt.legend(loc='best', frameon=True);\n",
    "plt.xlabel('Epochs');plt.ylabel('Error');"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "## Programming Multilayer Perceptron"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We need to start by defining a hidden layer.\n",
    "\n",
    "Typical hidden layer of a MLP: units are fully-connected and have sigmoidal activation function. Weight matrix W is of shape (n_in,n_out) and the bias vector b is of shape (n_out,).\n",
    "\n",
    "**Note**: \n",
    "* The nonlinearity used here is $\\tanh()$\n",
    "* Hidden unit output is given by: `tanh(dot(input,W) + b)`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "class HiddenLayer():\n",
    "    def __init__(self, rng, input, n_in, n_out, W=None, b=None,\n",
    "                 activation=T.tanh):\n",
    "        \"\"\"\n",
    "        :type rng: numpy.random.RandomState\n",
    "        :param rng: a random number generator used to initialize weights\n",
    "\n",
    "        :type input: theano.tensor.dmatrix\n",
    "        :param input: a symbolic tensor of shape (n_examples, n_in)\n",
    "\n",
    "        :type n_in: int\n",
    "        :param n_in: dimensionality of input\n",
    "\n",
    "        :type n_out: int\n",
    "        :param n_out: number of hidden units\n",
    "\n",
    "        :type activation: theano.Op or function\n",
    "        :param activation: Non linearity to be applied in the hidden\n",
    "                           layer\n",
    "        \"\"\"\n",
    "        self.input = input\n",
    "\n",
    "        # `W` is initialized with `W_values` which is uniformely sampled\n",
    "        # from sqrt(-6./(n_in+n_hidden)) and sqrt(6./(n_in+n_hidden))\n",
    "        # for tanh activation function\n",
    "        # the output of uniform if converted using asarray to dtype\n",
    "        # theano.config.floatX so that the code is runable on GPU\n",
    "        # Note : optimal initialization of weights is dependent on the\n",
    "        #        activation function used (among other things).\n",
    "        #        For example, results presented in [Xavier10] suggest that you\n",
    "        #        should use 4 times larger initial weights for sigmoid\n",
    "        #        compared to tanh\n",
    "        #        We have no info for other function, so we use the same as\n",
    "        #        tanh.\n",
    "        if W is None:\n",
    "            W_values = numpy.asarray(\n",
    "                rng.uniform(\n",
    "                    low=-numpy.sqrt(6. / (n_in + n_out)),\n",
    "                    high=numpy.sqrt(6. / (n_in + n_out)),\n",
    "                    size=(n_in, n_out)\n",
    "                ),\n",
    "                dtype=theano.config.floatX\n",
    "            )\n",
    "            if activation == theano.tensor.nnet.sigmoid:\n",
    "                W_values *= 4\n",
    "\n",
    "            W = theano.shared(value=W_values, name='W', borrow=True)\n",
    "\n",
    "        if b is None:\n",
    "            b_values = numpy.zeros((n_out,), dtype=theano.config.floatX)\n",
    "            b = theano.shared(value=b_values, name='b', borrow=True)\n",
    "\n",
    "        self.W = W\n",
    "        self.b = b\n",
    "\n",
    "        lin_output = T.dot(input, self.W) + self.b\n",
    "        self.output = (\n",
    "            lin_output if activation is None\n",
    "            else activation(lin_output)\n",
    "        )\n",
    "        # parameters of the model\n",
    "        self.params = [self.W, self.b]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Multi-Layer Perceptron Class\n",
    "\n",
    "A multilayer perceptron is a feedforward artificial neural network model that has one layer or more of hidden units and nonlinear activations. Intermediate layers usually have as activation function tanh or the sigmoid function (defined here by a ``HiddenLayer`` class)  while the top layer is a softmax layer (defined here by a ``LogisticRegression`` class)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The softmax function, or normalized exponential, is a generalization of the logistic function that \"squashes\" a $K$-dimensional vector $\\vec{z}$ of arbitrary real values to a $K$-dimensional vector $\\sigma(\\vec{z})$ of real values in the range (0, 1) that add up to 1. The function is given by\n",
    "\n",
    "$$\n",
    "\\sigma(\\mathbf{z})_j = \\frac{e^{z_j}}{\\sum_{k=1}^K e^{z_k}}\\quad    \\text{for}\\ j = 1,\\ldots, K.\n",
    "$$"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "class MLP:\n",
    "    def __init__(self, rng, input, n_in, n_hidden, n_out, activation=T.tanh):\n",
    "        \"\"\"Initialize the parameters for the multilayer perceptron\n",
    "\n",
    "        :type rng: numpy.random.RandomState\n",
    "        :param rng: a random number generator used to initialize weights\n",
    "\n",
    "        :type input: theano.tensor.TensorType\n",
    "        :param input: symbolic variable that describes the input of the\n",
    "        architecture (one minibatch)\n",
    "\n",
    "        :type n_in: int\n",
    "        :param n_in: number of input units, the dimension of the space in\n",
    "        which the datapoints lie\n",
    "\n",
    "        :type n_hidden: int\n",
    "        :param n_hidden: number of hidden units\n",
    "\n",
    "        :type n_out: int\n",
    "        :param n_out: number of output units, the dimension of the space in\n",
    "        which the labels lie\n",
    "        \"\"\"\n",
    "\n",
    "        # Since we are dealing with a one hidden layer MLP, this will translate\n",
    "        # into a HiddenLayer with a tanh activation function connected to the\n",
    "        # LogisticRegression layer; the activation function can be replaced by\n",
    "        # sigmoid or any other nonlinear function\n",
    "        self.hidden_layer = HiddenLayer(rng=rng, input=input, n_in=n_in, \n",
    "                                        n_out=n_hidden, activation=activation)\n",
    "\n",
    "        # The logistic regression layer gets as input the hidden units\n",
    "        # of the hidden layer\n",
    "        self.log_regression_layer = LogisticRegression(input=self.hidden_layer.output,\n",
    "                                                       n_in=n_hidden, n_out=n_out)\n",
    "\n",
    "        # L1 norm ; one regularization option is to enforce L1 norm to be small\n",
    "        self.L1 = abs(self.hidden_layer.W).sum() + abs(self.log_regression_layer.W).sum()\n",
    "\n",
    "        # square of L2 norm ; one regularization option is to enforce\n",
    "        # square of L2 norm to be small\n",
    "        self.L2_sqr = (self.hidden_layer.W ** 2).sum() + (self.log_regression_layer.W ** 2).sum()\n",
    "\n",
    "        # negative log likelihood of the MLP is given by the negative\n",
    "        # log likelihood of the output of the model, computed in the\n",
    "        # logistic regression layer\n",
    "        self.negative_log_likelihood = self.log_regression_layer.negative_log_likelihood\n",
    "        \n",
    "        # same holds for the function computing the number of errors\n",
    "        self.errors = self.log_regression_layer.errors\n",
    "\n",
    "        # the parameters of the model are the parameters of the two layer it is\n",
    "        # made out of\n",
    "        self.params = self.hidden_layer.params + self.log_regression_layer.params\n",
    "\n",
    "        # keep track of model input\n",
    "        self.input = input"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "dataset='mnist.pkl.gz'\n",
    "batch_size=30"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "datasets = load_data(dataset)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train_set_x, train_set_y = datasets[0]\n",
    "valid_set_x, valid_set_y = datasets[1]\n",
    "test_set_x, test_set_y = datasets[2]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Compute number of minibatches for training, validation and testing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "n_train_batches = train_set_x.get_value(borrow=True).shape[0] // batch_size\n",
    "n_valid_batches = valid_set_x.get_value(borrow=True).shape[0] // batch_size\n",
    "n_test_batches = test_set_x.get_value(borrow=True).shape[0] // batch_size"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "patience = 5000  # look as this many examples regardless\n",
    "validation_frequency = min(n_train_batches, patience // 2)\n",
    "                              # go through this many\n",
    "                              # minibatches before checking the network\n",
    "                              # on the validation set; in this case we\n",
    "                              # check every epoch"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Allocate symbolic variables for the data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "index = T.lscalar()  # index to a [mini]batch\n",
    "x = T.matrix('x')  # the data is presented as rasterized images\n",
    "y = T.ivector('y')  # the labels are presented as 1D vector of [int] labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "rng = numpy.random.RandomState(42)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Creating the MLP class instance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "learning_rate=0.05\n",
    "L1_reg=0.0001\n",
    "L2_reg=0.0001 \n",
    "n_epochs=300\n",
    "n_hidden=500"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "mlp = MLP(rng=rng, input=x, n_in=28 * 28, n_hidden=n_hidden, n_out=10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The cost we minimize during training is the negative log likelihood of the model plus the regularization terms (L1 and L2); cost is expressed here symbolically."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "cost = mlp.negative_log_likelihood(y) + L1_reg * mlp.L1 + L2_reg * mlp.L2_sqr"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Compiling a Theano function that computes the mistakes that are made by the model on a minibatch"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "test_model = theano.function(inputs=[index], outputs=mlp.errors(y),\n",
    "                             givens={x: test_set_x[index * batch_size:(index + 1) * batch_size],\n",
    "                                     y: test_set_y[index * batch_size:(index + 1) * batch_size]})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "validate_model = theano.function(inputs=[index], outputs=mlp.errors(y),\n",
    "                                 givens={x: valid_set_x[index * batch_size:(index + 1) * batch_size],\n",
    "                                         y: valid_set_y[index * batch_size:(index + 1) * batch_size]})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train_set_loss = theano.function(inputs=[index], outputs=mlp.errors(y),\n",
    "                                 givens={x: train_set_x[index * batch_size: (index + 1) * batch_size],\n",
    "                                         y: train_set_y[index * batch_size: (index + 1) * batch_size]})"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Compute the gradient of cost with respect to the parameters (sorted in `classifier.params`) the resulting gradients will be stored in a list `gparams`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "gparams = [T.grad(cost, param) for param in mlp.params]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Specify how to update the parameters of the model as a list of `(variable, update expression)` pairs."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "updates = [(param, param - learning_rate * gparam) \n",
    "           for param, gparam in zip(mlp.params, gparams)]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Compiling a Theano function `train_model` that returns the cost, and at the same time, updates the parameter of the model based on the rules\n",
    "defined in `updates`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train_model = theano.function(inputs=[index], outputs=cost, updates=updates,\n",
    "                              givens={x: train_set_x[index * batch_size: (index + 1) * batch_size],\n",
    "                                      y: train_set_y[index * batch_size: (index + 1) * batch_size]})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 1, minibatch 1666/1666, validation error 8.518519%\n",
      "epoch 2, minibatch 1666/1666, validation error 7.907908%\n",
      "epoch 3, minibatch 1666/1666, validation error 7.487487%\n",
      "epoch 4, minibatch 1666/1666, validation error 6.996997%\n",
      "epoch 5, minibatch 1666/1666, validation error 6.466466%\n",
      "epoch 6, minibatch 1666/1666, validation error 5.875876%\n",
      "epoch 7, minibatch 1666/1666, validation error 5.335335%\n",
      "epoch 8, minibatch 1666/1666, validation error 4.984985%\n",
      "epoch 9, minibatch 1666/1666, validation error 4.634635%\n",
      "epoch 10, minibatch 1666/1666, validation error 4.284284%\n",
      "epoch 11, minibatch 1666/1666, validation error 4.104104%\n",
      "epoch 12, minibatch 1666/1666, validation error 3.883884%\n",
      "epoch 13, minibatch 1666/1666, validation error 3.603604%\n",
      "epoch 14, minibatch 1666/1666, validation error 3.493493%\n",
      "epoch 15, minibatch 1666/1666, validation error 3.383383%\n",
      "epoch 16, minibatch 1666/1666, validation error 3.283283%\n",
      "epoch 17, minibatch 1666/1666, validation error 3.163163%\n",
      "epoch 18, minibatch 1666/1666, validation error 3.103103%\n",
      "epoch 19, minibatch 1666/1666, validation error 3.053053%\n",
      "epoch 20, minibatch 1666/1666, validation error 3.013013%\n",
      "epoch 21, minibatch 1666/1666, validation error 2.992993%\n",
      "epoch 22, minibatch 1666/1666, validation error 2.972973%\n",
      "epoch 23, minibatch 1666/1666, validation error 2.952953%\n",
      "epoch 24, minibatch 1666/1666, validation error 2.932933%\n",
      "epoch 25, minibatch 1666/1666, validation error 2.912913%\n",
      "epoch 26, minibatch 1666/1666, validation error 2.812813%\n",
      "epoch 27, minibatch 1666/1666, validation error 2.792793%\n",
      "epoch 28, minibatch 1666/1666, validation error 2.812813%\n",
      "epoch 29, minibatch 1666/1666, validation error 2.742743%\n",
      "epoch 30, minibatch 1666/1666, validation error 2.682683%\n",
      "epoch 31, minibatch 1666/1666, validation error 2.652653%\n",
      "epoch 32, minibatch 1666/1666, validation error 2.642643%\n",
      "epoch 33, minibatch 1666/1666, validation error 2.592593%\n",
      "epoch 34, minibatch 1666/1666, validation error 2.602603%\n",
      "epoch 35, minibatch 1666/1666, validation error 2.562563%\n",
      "epoch 36, minibatch 1666/1666, validation error 2.542543%\n",
      "epoch 37, minibatch 1666/1666, validation error 2.532533%\n",
      "epoch 38, minibatch 1666/1666, validation error 2.522523%\n",
      "epoch 39, minibatch 1666/1666, validation error 2.512513%\n",
      "epoch 40, minibatch 1666/1666, validation error 2.502503%\n",
      "epoch 41, minibatch 1666/1666, validation error 2.492492%\n",
      "epoch 42, minibatch 1666/1666, validation error 2.502503%\n",
      "epoch 43, minibatch 1666/1666, validation error 2.492492%\n",
      "epoch 44, minibatch 1666/1666, validation error 2.462462%\n",
      "epoch 45, minibatch 1666/1666, validation error 2.432432%\n",
      "epoch 46, minibatch 1666/1666, validation error 2.432432%\n",
      "epoch 47, minibatch 1666/1666, validation error 2.442442%\n",
      "epoch 48, minibatch 1666/1666, validation error 2.432432%\n",
      "epoch 49, minibatch 1666/1666, validation error 2.432432%\n",
      "epoch 50, minibatch 1666/1666, validation error 2.432432%\n",
      "epoch 51, minibatch 1666/1666, validation error 2.442442%\n",
      "epoch 52, minibatch 1666/1666, validation error 2.462462%\n",
      "epoch 53, minibatch 1666/1666, validation error 2.472472%\n",
      "epoch 54, minibatch 1666/1666, validation error 2.462462%\n",
      "epoch 55, minibatch 1666/1666, validation error 2.462462%\n",
      "epoch 56, minibatch 1666/1666, validation error 2.462462%\n",
      "epoch 57, minibatch 1666/1666, validation error 2.462462%\n",
      "epoch 58, minibatch 1666/1666, validation error 2.502503%\n",
      "epoch 59, minibatch 1666/1666, validation error 2.502503%\n",
      "epoch 60, minibatch 1666/1666, validation error 2.502503%\n",
      "epoch 61, minibatch 1666/1666, validation error 2.492492%\n",
      "epoch 62, minibatch 1666/1666, validation error 2.502503%\n",
      "epoch 63, minibatch 1666/1666, validation error 2.482482%\n",
      "epoch 64, minibatch 1666/1666, validation error 2.482482%\n",
      "epoch 65, minibatch 1666/1666, validation error 2.482482%\n",
      "epoch 66, minibatch 1666/1666, validation error 2.482482%\n",
      "epoch 67, minibatch 1666/1666, validation error 2.492492%\n",
      "epoch 68, minibatch 1666/1666, validation error 2.452452%\n",
      "epoch 69, minibatch 1666/1666, validation error 2.452452%\n",
      "epoch 70, minibatch 1666/1666, validation error 2.442442%\n",
      "epoch 71, minibatch 1666/1666, validation error 2.432432%\n",
      "epoch 72, minibatch 1666/1666, validation error 2.432432%\n",
      "epoch 73, minibatch 1666/1666, validation error 2.422422%\n",
      "epoch 74, minibatch 1666/1666, validation error 2.402402%\n",
      "epoch 75, minibatch 1666/1666, validation error 2.402402%\n",
      "epoch 76, minibatch 1666/1666, validation error 2.402402%\n",
      "epoch 77, minibatch 1666/1666, validation error 2.412412%\n",
      "epoch 78, minibatch 1666/1666, validation error 2.402402%\n",
      "epoch 79, minibatch 1666/1666, validation error 2.402402%\n",
      "epoch 80, minibatch 1666/1666, validation error 2.402402%\n",
      "epoch 81, minibatch 1666/1666, validation error 2.402402%\n",
      "epoch 82, minibatch 1666/1666, validation error 2.402402%\n",
      "epoch 83, minibatch 1666/1666, validation error 2.402402%\n",
      "epoch 84, minibatch 1666/1666, validation error 2.392392%\n",
      "epoch 85, minibatch 1666/1666, validation error 2.392392%\n",
      "epoch 86, minibatch 1666/1666, validation error 2.382382%\n",
      "epoch 87, minibatch 1666/1666, validation error 2.382382%\n",
      "epoch 88, minibatch 1666/1666, validation error 2.382382%\n",
      "epoch 89, minibatch 1666/1666, validation error 2.392392%\n",
      "epoch 90, minibatch 1666/1666, validation error 2.392392%\n",
      "epoch 91, minibatch 1666/1666, validation error 2.402402%\n",
      "epoch 92, minibatch 1666/1666, validation error 2.392392%\n",
      "epoch 93, minibatch 1666/1666, validation error 2.392392%\n",
      "epoch 94, minibatch 1666/1666, validation error 2.382382%\n",
      "epoch 95, minibatch 1666/1666, validation error 2.392392%\n",
      "epoch 96, minibatch 1666/1666, validation error 2.402402%\n",
      "epoch 97, minibatch 1666/1666, validation error 2.382382%\n",
      "epoch 98, minibatch 1666/1666, validation error 2.382382%\n",
      "epoch 99, minibatch 1666/1666, validation error 2.362362%\n",
      "epoch 100, minibatch 1666/1666, validation error 2.352352%\n",
      "epoch 101, minibatch 1666/1666, validation error 2.342342%\n",
      "epoch 102, minibatch 1666/1666, validation error 2.342342%\n",
      "epoch 103, minibatch 1666/1666, validation error 2.342342%\n",
      "epoch 104, minibatch 1666/1666, validation error 2.342342%\n",
      "epoch 105, minibatch 1666/1666, validation error 2.312312%\n",
      "epoch 106, minibatch 1666/1666, validation error 2.292292%\n",
      "epoch 107, minibatch 1666/1666, validation error 2.292292%\n",
      "epoch 108, minibatch 1666/1666, validation error 2.282282%\n",
      "epoch 109, minibatch 1666/1666, validation error 2.292292%\n",
      "epoch 110, minibatch 1666/1666, validation error 2.282282%\n",
      "epoch 111, minibatch 1666/1666, validation error 2.272272%\n",
      "epoch 112, minibatch 1666/1666, validation error 2.262262%\n",
      "epoch 113, minibatch 1666/1666, validation error 2.282282%\n",
      "epoch 114, minibatch 1666/1666, validation error 2.282282%\n",
      "epoch 115, minibatch 1666/1666, validation error 2.292292%\n",
      "epoch 116, minibatch 1666/1666, validation error 2.292292%\n",
      "epoch 117, minibatch 1666/1666, validation error 2.282282%\n",
      "epoch 118, minibatch 1666/1666, validation error 2.272272%\n",
      "epoch 119, minibatch 1666/1666, validation error 2.262262%\n",
      "epoch 120, minibatch 1666/1666, validation error 2.262262%\n",
      "epoch 121, minibatch 1666/1666, validation error 2.262262%\n",
      "epoch 122, minibatch 1666/1666, validation error 2.272272%\n",
      "epoch 123, minibatch 1666/1666, validation error 2.272272%\n",
      "epoch 124, minibatch 1666/1666, validation error 2.272272%\n",
      "epoch 125, minibatch 1666/1666, validation error 2.272272%\n",
      "epoch 126, minibatch 1666/1666, validation error 2.262262%\n",
      "epoch 127, minibatch 1666/1666, validation error 2.272272%\n",
      "epoch 128, minibatch 1666/1666, validation error 2.282282%\n",
      "epoch 129, minibatch 1666/1666, validation error 2.282282%\n",
      "epoch 130, minibatch 1666/1666, validation error 2.282282%\n",
      "epoch 131, minibatch 1666/1666, validation error 2.282282%\n",
      "epoch 132, minibatch 1666/1666, validation error 2.282282%\n",
      "epoch 133, minibatch 1666/1666, validation error 2.282282%\n",
      "epoch 134, minibatch 1666/1666, validation error 2.282282%\n",
      "epoch 135, minibatch 1666/1666, validation error 2.282282%\n",
      "epoch 136, minibatch 1666/1666, validation error 2.282282%\n",
      "epoch 137, minibatch 1666/1666, validation error 2.282282%\n",
      "epoch 138, minibatch 1666/1666, validation error 2.282282%\n",
      "epoch 139, minibatch 1666/1666, validation error 2.292292%\n",
      "epoch 140, minibatch 1666/1666, validation error 2.282282%\n",
      "epoch 141, minibatch 1666/1666, validation error 2.272272%\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 142, minibatch 1666/1666, validation error 2.272272%\n",
      "epoch 143, minibatch 1666/1666, validation error 2.272272%\n",
      "epoch 144, minibatch 1666/1666, validation error 2.272272%\n",
      "epoch 145, minibatch 1666/1666, validation error 2.292292%\n",
      "epoch 146, minibatch 1666/1666, validation error 2.302302%\n",
      "epoch 147, minibatch 1666/1666, validation error 2.302302%\n",
      "epoch 148, minibatch 1666/1666, validation error 2.312312%\n",
      "epoch 149, minibatch 1666/1666, validation error 2.292292%\n",
      "epoch 150, minibatch 1666/1666, validation error 2.292292%\n",
      "epoch 151, minibatch 1666/1666, validation error 2.282282%\n",
      "epoch 152, minibatch 1666/1666, validation error 2.282282%\n",
      "epoch 153, minibatch 1666/1666, validation error 2.282282%\n",
      "epoch 154, minibatch 1666/1666, validation error 2.292292%\n",
      "epoch 155, minibatch 1666/1666, validation error 2.292292%\n",
      "epoch 156, minibatch 1666/1666, validation error 2.302302%\n",
      "epoch 157, minibatch 1666/1666, validation error 2.302302%\n",
      "epoch 158, minibatch 1666/1666, validation error 2.302302%\n",
      "epoch 159, minibatch 1666/1666, validation error 2.302302%\n",
      "epoch 160, minibatch 1666/1666, validation error 2.302302%\n",
      "epoch 161, minibatch 1666/1666, validation error 2.312312%\n",
      "epoch 162, minibatch 1666/1666, validation error 2.312312%\n",
      "epoch 163, minibatch 1666/1666, validation error 2.302302%\n",
      "epoch 164, minibatch 1666/1666, validation error 2.292292%\n",
      "epoch 165, minibatch 1666/1666, validation error 2.292292%\n",
      "epoch 166, minibatch 1666/1666, validation error 2.292292%\n",
      "epoch 167, minibatch 1666/1666, validation error 2.292292%\n",
      "epoch 168, minibatch 1666/1666, validation error 2.292292%\n",
      "epoch 169, minibatch 1666/1666, validation error 2.292292%\n",
      "epoch 170, minibatch 1666/1666, validation error 2.292292%\n",
      "epoch 171, minibatch 1666/1666, validation error 2.282282%\n",
      "epoch 172, minibatch 1666/1666, validation error 2.282282%\n",
      "epoch 173, minibatch 1666/1666, validation error 2.272272%\n",
      "epoch 174, minibatch 1666/1666, validation error 2.282282%\n",
      "epoch 175, minibatch 1666/1666, validation error 2.282282%\n",
      "epoch 176, minibatch 1666/1666, validation error 2.292292%\n",
      "epoch 177, minibatch 1666/1666, validation error 2.292292%\n",
      "epoch 178, minibatch 1666/1666, validation error 2.282282%\n",
      "epoch 179, minibatch 1666/1666, validation error 2.272272%\n",
      "epoch 180, minibatch 1666/1666, validation error 2.272272%\n",
      "epoch 181, minibatch 1666/1666, validation error 2.272272%\n",
      "epoch 182, minibatch 1666/1666, validation error 2.272272%\n",
      "epoch 183, minibatch 1666/1666, validation error 2.272272%\n",
      "epoch 184, minibatch 1666/1666, validation error 2.272272%\n",
      "epoch 185, minibatch 1666/1666, validation error 2.272272%\n",
      "epoch 186, minibatch 1666/1666, validation error 2.272272%\n",
      "epoch 187, minibatch 1666/1666, validation error 2.272272%\n",
      "epoch 188, minibatch 1666/1666, validation error 2.272272%\n",
      "epoch 189, minibatch 1666/1666, validation error 2.272272%\n",
      "epoch 190, minibatch 1666/1666, validation error 2.272272%\n",
      "epoch 191, minibatch 1666/1666, validation error 2.272272%\n",
      "epoch 192, minibatch 1666/1666, validation error 2.272272%\n",
      "epoch 193, minibatch 1666/1666, validation error 2.272272%\n",
      "epoch 194, minibatch 1666/1666, validation error 2.272272%\n",
      "epoch 195, minibatch 1666/1666, validation error 2.282282%\n",
      "epoch 196, minibatch 1666/1666, validation error 2.282282%\n",
      "epoch 197, minibatch 1666/1666, validation error 2.272272%\n",
      "epoch 198, minibatch 1666/1666, validation error 2.272272%\n",
      "epoch 199, minibatch 1666/1666, validation error 2.272272%\n",
      "epoch 200, minibatch 1666/1666, validation error 2.262262%\n",
      "epoch 201, minibatch 1666/1666, validation error 2.262262%\n",
      "epoch 202, minibatch 1666/1666, validation error 2.262262%\n",
      "epoch 203, minibatch 1666/1666, validation error 2.272272%\n",
      "epoch 204, minibatch 1666/1666, validation error 2.272272%\n",
      "epoch 205, minibatch 1666/1666, validation error 2.272272%\n",
      "epoch 206, minibatch 1666/1666, validation error 2.252252%\n",
      "epoch 207, minibatch 1666/1666, validation error 2.252252%\n",
      "epoch 208, minibatch 1666/1666, validation error 2.262262%\n",
      "epoch 209, minibatch 1666/1666, validation error 2.262262%\n",
      "epoch 210, minibatch 1666/1666, validation error 2.262262%\n",
      "epoch 211, minibatch 1666/1666, validation error 2.262262%\n",
      "Ran out of patience!\n",
      "CPU times: user 1h 43min 54s, sys: 1min 44s, total: 1h 45min 38s\n",
      "Wall time: 59min 12s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "train_errors, validation_errors, test_errors = train(mlp, train_model, validate_model, test_model,\n",
    "                                                     n_epochs, \n",
    "                                                     n_train_batches, n_valid_batches, n_test_batches, \n",
    "                                                     validation_frequency, verbose=True, log_results=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAxwAAAIcCAYAAAB8cKi8AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nP/RMqK1WjwjsAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x112fdbf28>"
      ]
     },
     "metadata": {
      "image/png": {
       "height": 270,
       "width": 398
      }
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(train_errors, label='Training')\n",
    "plt.plot(validation_errors, label='Validation')\n",
    "plt.plot(test_errors, label='Test')\n",
    "plt.legend(loc='best', frameon=True);\n",
    "plt.xlabel('Epochs');plt.ylabel('Error');"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Improvement 1: Rectified Linear Unit activation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def relu(x):\n",
    "    return x * (x > 0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [],
   "source": [
    "rng = numpy.random.RandomState(1234)\n",
    "\n",
    "mlp_relu = MLP(rng=rng, input=x, n_in=28*28,\n",
    "               n_hidden=n_hidden, n_out=10, activation=relu)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "cost = mlp_relu.negative_log_likelihood(y) + L1_reg * mlp_relu.L1 + L2_reg * mlp_relu.L2_sqr"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Compiling a Theano function that computes the mistakes that are made by the model on a minibatch"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "test_model = theano.function(inputs=[index], outputs=mlp_relu.errors(y),\n",
    "                             givens={x: test_set_x[index * batch_size:(index + 1) * batch_size],\n",
    "                                     y: test_set_y[index * batch_size:(index + 1) * batch_size]})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "validate_model = theano.function(inputs=[index], outputs=mlp_relu.errors(y),\n",
    "                                 givens={x: valid_set_x[index * batch_size:(index + 1) * batch_size],\n",
    "                                         y: valid_set_y[index * batch_size:(index + 1) * batch_size]})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train_set_loss = theano.function(inputs=[index], outputs=mlp_relu.errors(y),\n",
    "                                 givens={x: train_set_x[index * batch_size: (index + 1) * batch_size],\n",
    "                                         y: train_set_y[index * batch_size: (index + 1) * batch_size]})"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Compute the gradient of cost with respect to the parameters (sorted in `classifier.params`) the resulting gradients will be stored in a list `gparams`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "gparams = [T.grad(cost, param) for param in mlp_relu.params]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Specify how to update the parameters of the model as a list of `(variable, update expression)` pairs.\n",
    "\n",
    "* given two lists of the same length, `A = [a1, a2, a3, a4]` and `B = [b1, b2, b3, b4]`, `zip()` generates a list `C` of same size, where eachnelement is a pair formed from the two lists:\n",
    "```python\n",
    "C = [(a1, b1), (a2, b2), (a3, b3), (a4, b4)]\n",
    "```"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "updates = [(param, param - learning_rate * gparam) \n",
    "           for param, gparam in zip(mlp_relu.params, gparams)]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Compiling a Theano function `train_model` that returns the cost, and at the same time, updates the parameter of the model based on the rules\n",
    "defined in `updates`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train_model = theano.function(inputs=[index], outputs=cost, updates=updates,\n",
    "                              givens={x: train_set_x[index * batch_size: (index + 1) * batch_size],\n",
    "                                      y: train_set_y[index * batch_size: (index + 1) * batch_size]})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 1, minibatch 1666/1666, validation error 7.607608%\n",
      "epoch 2, minibatch 1666/1666, validation error 5.735736%\n",
      "epoch 3, minibatch 1666/1666, validation error 4.604605%\n",
      "epoch 4, minibatch 1666/1666, validation error 4.134134%\n",
      "epoch 5, minibatch 1666/1666, validation error 3.763764%\n",
      "epoch 6, minibatch 1666/1666, validation error 3.563564%\n",
      "epoch 7, minibatch 1666/1666, validation error 3.403403%\n",
      "epoch 8, minibatch 1666/1666, validation error 3.243243%\n",
      "epoch 9, minibatch 1666/1666, validation error 3.133133%\n",
      "epoch 10, minibatch 1666/1666, validation error 3.023023%\n",
      "epoch 11, minibatch 1666/1666, validation error 2.882883%\n",
      "epoch 12, minibatch 1666/1666, validation error 2.792793%\n",
      "epoch 13, minibatch 1666/1666, validation error 2.702703%\n",
      "epoch 14, minibatch 1666/1666, validation error 2.702703%\n",
      "epoch 15, minibatch 1666/1666, validation error 2.682683%\n",
      "epoch 16, minibatch 1666/1666, validation error 2.602603%\n",
      "epoch 17, minibatch 1666/1666, validation error 2.552553%\n",
      "epoch 18, minibatch 1666/1666, validation error 2.542543%\n",
      "epoch 19, minibatch 1666/1666, validation error 2.472472%\n",
      "epoch 20, minibatch 1666/1666, validation error 2.482482%\n",
      "epoch 21, minibatch 1666/1666, validation error 2.482482%\n",
      "epoch 22, minibatch 1666/1666, validation error 2.502503%\n",
      "epoch 23, minibatch 1666/1666, validation error 2.462462%\n",
      "epoch 24, minibatch 1666/1666, validation error 2.462462%\n",
      "epoch 25, minibatch 1666/1666, validation error 2.452452%\n",
      "epoch 26, minibatch 1666/1666, validation error 2.492492%\n",
      "epoch 27, minibatch 1666/1666, validation error 2.502503%\n",
      "epoch 28, minibatch 1666/1666, validation error 2.472472%\n",
      "epoch 29, minibatch 1666/1666, validation error 2.502503%\n",
      "epoch 30, minibatch 1666/1666, validation error 2.472472%\n",
      "epoch 31, minibatch 1666/1666, validation error 2.422422%\n",
      "epoch 32, minibatch 1666/1666, validation error 2.452452%\n",
      "epoch 33, minibatch 1666/1666, validation error 2.472472%\n",
      "epoch 34, minibatch 1666/1666, validation error 2.462462%\n",
      "epoch 35, minibatch 1666/1666, validation error 2.442442%\n",
      "epoch 36, minibatch 1666/1666, validation error 2.412412%\n",
      "epoch 37, minibatch 1666/1666, validation error 2.412412%\n",
      "epoch 38, minibatch 1666/1666, validation error 2.422422%\n",
      "epoch 39, minibatch 1666/1666, validation error 2.402402%\n",
      "epoch 40, minibatch 1666/1666, validation error 2.382382%\n",
      "epoch 41, minibatch 1666/1666, validation error 2.372372%\n",
      "epoch 42, minibatch 1666/1666, validation error 2.392392%\n",
      "epoch 43, minibatch 1666/1666, validation error 2.352352%\n",
      "epoch 44, minibatch 1666/1666, validation error 2.372372%\n",
      "epoch 45, minibatch 1666/1666, validation error 2.372372%\n",
      "epoch 46, minibatch 1666/1666, validation error 2.352352%\n",
      "epoch 47, minibatch 1666/1666, validation error 2.362362%\n",
      "epoch 48, minibatch 1666/1666, validation error 2.352352%\n",
      "epoch 49, minibatch 1666/1666, validation error 2.372372%\n",
      "epoch 50, minibatch 1666/1666, validation error 2.312312%\n",
      "epoch 51, minibatch 1666/1666, validation error 2.342342%\n",
      "epoch 52, minibatch 1666/1666, validation error 2.352352%\n",
      "epoch 53, minibatch 1666/1666, validation error 2.312312%\n",
      "epoch 54, minibatch 1666/1666, validation error 2.322322%\n",
      "epoch 55, minibatch 1666/1666, validation error 2.332332%\n",
      "epoch 56, minibatch 1666/1666, validation error 2.342342%\n",
      "epoch 57, minibatch 1666/1666, validation error 2.302302%\n",
      "epoch 58, minibatch 1666/1666, validation error 2.312312%\n",
      "epoch 59, minibatch 1666/1666, validation error 2.292292%\n",
      "epoch 60, minibatch 1666/1666, validation error 2.282282%\n",
      "epoch 61, minibatch 1666/1666, validation error 2.252252%\n",
      "epoch 62, minibatch 1666/1666, validation error 2.282282%\n",
      "epoch 63, minibatch 1666/1666, validation error 2.252252%\n",
      "epoch 64, minibatch 1666/1666, validation error 2.282282%\n",
      "epoch 65, minibatch 1666/1666, validation error 2.282282%\n",
      "epoch 66, minibatch 1666/1666, validation error 2.282282%\n",
      "epoch 67, minibatch 1666/1666, validation error 2.322322%\n",
      "epoch 68, minibatch 1666/1666, validation error 2.282282%\n",
      "epoch 69, minibatch 1666/1666, validation error 2.292292%\n",
      "epoch 70, minibatch 1666/1666, validation error 2.262262%\n",
      "epoch 71, minibatch 1666/1666, validation error 2.242242%\n",
      "epoch 72, minibatch 1666/1666, validation error 2.252252%\n",
      "epoch 73, minibatch 1666/1666, validation error 2.252252%\n",
      "epoch 74, minibatch 1666/1666, validation error 2.272272%\n",
      "epoch 75, minibatch 1666/1666, validation error 2.232232%\n",
      "epoch 76, minibatch 1666/1666, validation error 2.252252%\n",
      "epoch 77, minibatch 1666/1666, validation error 2.282282%\n",
      "epoch 78, minibatch 1666/1666, validation error 2.252252%\n",
      "epoch 79, minibatch 1666/1666, validation error 2.272272%\n",
      "epoch 80, minibatch 1666/1666, validation error 2.262262%\n",
      "epoch 81, minibatch 1666/1666, validation error 2.262262%\n",
      "epoch 82, minibatch 1666/1666, validation error 2.252252%\n",
      "epoch 83, minibatch 1666/1666, validation error 2.262262%\n",
      "epoch 84, minibatch 1666/1666, validation error 2.242242%\n",
      "epoch 85, minibatch 1666/1666, validation error 2.292292%\n",
      "epoch 86, minibatch 1666/1666, validation error 2.252252%\n",
      "epoch 87, minibatch 1666/1666, validation error 2.232232%\n",
      "epoch 88, minibatch 1666/1666, validation error 2.232232%\n",
      "epoch 89, minibatch 1666/1666, validation error 2.282282%\n",
      "epoch 90, minibatch 1666/1666, validation error 2.262262%\n",
      "epoch 91, minibatch 1666/1666, validation error 2.242242%\n",
      "epoch 92, minibatch 1666/1666, validation error 2.292292%\n",
      "epoch 93, minibatch 1666/1666, validation error 2.262262%\n",
      "epoch 94, minibatch 1666/1666, validation error 2.282282%\n",
      "epoch 95, minibatch 1666/1666, validation error 2.272272%\n",
      "epoch 96, minibatch 1666/1666, validation error 2.272272%\n",
      "epoch 97, minibatch 1666/1666, validation error 2.252252%\n",
      "epoch 98, minibatch 1666/1666, validation error 2.252252%\n",
      "epoch 99, minibatch 1666/1666, validation error 2.252252%\n",
      "epoch 100, minibatch 1666/1666, validation error 2.252252%\n",
      "epoch 101, minibatch 1666/1666, validation error 2.322322%\n",
      "epoch 102, minibatch 1666/1666, validation error 2.292292%\n",
      "epoch 103, minibatch 1666/1666, validation error 2.292292%\n",
      "epoch 104, minibatch 1666/1666, validation error 2.282282%\n",
      "epoch 105, minibatch 1666/1666, validation error 2.282282%\n",
      "epoch 106, minibatch 1666/1666, validation error 2.282282%\n",
      "epoch 107, minibatch 1666/1666, validation error 2.312312%\n",
      "epoch 108, minibatch 1666/1666, validation error 2.322322%\n",
      "epoch 109, minibatch 1666/1666, validation error 2.302302%\n",
      "epoch 110, minibatch 1666/1666, validation error 2.302302%\n",
      "epoch 111, minibatch 1666/1666, validation error 2.322322%\n",
      "epoch 112, minibatch 1666/1666, validation error 2.292292%\n",
      "epoch 113, minibatch 1666/1666, validation error 2.302302%\n",
      "epoch 114, minibatch 1666/1666, validation error 2.332332%\n",
      "epoch 115, minibatch 1666/1666, validation error 2.322322%\n",
      "epoch 116, minibatch 1666/1666, validation error 2.322322%\n",
      "epoch 117, minibatch 1666/1666, validation error 2.332332%\n",
      "epoch 118, minibatch 1666/1666, validation error 2.342342%\n",
      "epoch 119, minibatch 1666/1666, validation error 2.342342%\n",
      "epoch 120, minibatch 1666/1666, validation error 2.372372%\n",
      "epoch 121, minibatch 1666/1666, validation error 2.372372%\n",
      "Ran out of patience!\n",
      "CPU times: user 48min 14s, sys: 48.8 s, total: 49min 3s\n",
      "Wall time: 38min 55s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "errors_relu = train(mlp_relu, train_model, validate_model, test_model, n_epochs, \n",
    "                    n_train_batches, n_valid_batches, n_test_batches, \n",
    "                    validation_frequency, verbose=True, log_results=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train_errors_relu, validation_errors_relu, test_errors_relu = errors_relu"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAABrEAAAIxCAYAAADqs5N3AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x11ae32358>"
      ]
     },
     "metadata": {
      "image/png": {
       "height": 280,
       "width": 856
      }
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(12,4))\n",
    "ax = plt.subplot(131)\n",
    "plt.plot(train_errors, label='Tanh')\n",
    "plt.plot(train_errors_relu, label='ReLU')\n",
    "plt.legend(loc='best', frameon=True)\n",
    "plt.xlabel('Epochs');plt.ylabel('Error')\n",
    "plt.title('Training set errors')\n",
    "plt.subplot(132, sharey=ax)\n",
    "plt.plot(validation_errors, label='Tanh')\n",
    "plt.plot(validation_errors_relu, label='ReLU')\n",
    "plt.legend(loc='best', frameon=True);\n",
    "plt.xlabel('Epochs')\n",
    "plt.title('Validation set errors')\n",
    "plt.subplot(133, sharey=ax)\n",
    "plt.plot(test_errors, label='Tanh')\n",
    "plt.plot(test_errors_relu, label='ReLU')\n",
    "plt.legend(loc='best', frameon=True);\n",
    "plt.xlabel('Epochs')\n",
    "plt.title('Test set errors')\n",
    "plt.tight_layout()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Your task:\n",
    "\n",
    "* Implement a MLP that can have different numbers of hidden layers.\n",
    "* Add the momentum term to gradients."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/json": {
       "Software versions": [
        {
         "module": "Python",
         "version": "3.6.1 64bit [GCC 4.2.1 Compatible Apple LLVM 6.0 (clang-600.0.57)]"
        },
        {
         "module": "IPython",
         "version": "5.3.0"
        },
        {
         "module": "OS",
         "version": "Darwin 16.5.0 x86_64 i386 64bit"
        },
        {
         "module": "scipy",
         "version": "0.19.0"
        },
        {
         "module": "numpy",
         "version": "1.12.1"
        },
        {
         "module": "matplotlib",
         "version": "2.0.0"
        },
        {
         "module": "sklearn",
         "version": "0.18.1"
        },
        {
         "module": "theano",
         "version": "0.9.0.dev-c697eeab84e5b8a74908da654b66ec9eca4f1291"
        }
       ]
      },
      "text/html": [
       "<table><tr><th>Software</th><th>Version</th></tr><tr><td>Python</td><td>3.6.1 64bit [GCC 4.2.1 Compatible Apple LLVM 6.0 (clang-600.0.57)]</td></tr><tr><td>IPython</td><td>5.3.0</td></tr><tr><td>OS</td><td>Darwin 16.5.0 x86_64 i386 64bit</td></tr><tr><td>scipy</td><td>0.19.0</td></tr><tr><td>numpy</td><td>1.12.1</td></tr><tr><td>matplotlib</td><td>2.0.0</td></tr><tr><td>sklearn</td><td>0.18.1</td></tr><tr><td>theano</td><td>0.9.0.dev-c697eeab84e5b8a74908da654b66ec9eca4f1291</td></tr><tr><td colspan='2'>Sat Apr 08 18:09:37 2017 -03</td></tr></table>"
      ],
      "text/latex": [
       "\\begin{tabular}{|l|l|}\\hline\n",
       "{\\bf Software} & {\\bf Version} \\\\ \\hline\\hline\n",
       "Python & 3.6.1 64bit [GCC 4.2.1 Compatible Apple LLVM 6.0 (clang-600.0.57)] \\\\ \\hline\n",
       "IPython & 5.3.0 \\\\ \\hline\n",
       "OS & Darwin 16.5.0 x86\\_64 i386 64bit \\\\ \\hline\n",
       "scipy & 0.19.0 \\\\ \\hline\n",
       "numpy & 1.12.1 \\\\ \\hline\n",
       "matplotlib & 2.0.0 \\\\ \\hline\n",
       "sklearn & 0.18.1 \\\\ \\hline\n",
       "theano & 0.9.0.dev-c697eeab84e5b8a74908da654b66ec9eca4f1291 \\\\ \\hline\n",
       "\\hline \\multicolumn{2}{|l|}{Sat Apr 08 18:09:37 2017 -03} \\\\ \\hline\n",
       "\\end{tabular}\n"
      ],
      "text/plain": [
       "Software versions\n",
       "Python 3.6.1 64bit [GCC 4.2.1 Compatible Apple LLVM 6.0 (clang-600.0.57)]\n",
       "IPython 5.3.0\n",
       "OS Darwin 16.5.0 x86_64 i386 64bit\n",
       "scipy 0.19.0\n",
       "numpy 1.12.1\n",
       "matplotlib 2.0.0\n",
       "sklearn 0.18.1\n",
       "theano 0.9.0.dev-c697eeab84e5b8a74908da654b66ec9eca4f1291\n",
       "Sat Apr 08 18:09:37 2017 -03"
      ]
     },
     "execution_count": 62,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%load_ext version_information\n",
    "%version_information scipy, numpy, matplotlib, sklearn, theano"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<link href=\"https://fonts.googleapis.com/css?family=Avenir|Fira+Sans|Inconsolata\" rel=\"stylesheet\">\n",
       "<link href=\"https://fonts.googleapis.com/css?family=Titillium+Web:400,400i,700,700i\" rel=\"stylesheet\">\n",
       "<style>\n",
       ".text_cell_render {\n",
       "font-style: regular;\n",
       "font-family: 'Fira Sans', sans-serif;\n",
       "display: block;\n",
       "}\n",
       "/*font-weight: 200;*/\n",
       "/*text-align: left;\n",
       "line-height: 100%;\n",
       "display: block;\n",
       "}*/\n",
       ".text_cell_render h1 {\n",
       "/*font-size: 24pt;*/\n",
       "font-family: 'Titillium Web', sans-serif;\n",
       "font-weight: bold;\n",
       "margin-bottom: 0.1em;\n",
       "margin-top: 0.5em;\n",
       "color:#4a4a4a;\n",
       "}\n",
       "\n",
       ".reveal h1 {\n",
       "font-family: 'Titillium Web', sans-serif;\n",
       "/*font-size: 24pt;*/\n",
       "font-weight: bold;\n",
       "margin-bottom: 0.1em;\n",
       "margin-top: 0.5em;\n",
       "color:#4a4a4a;\n",
       "}\n",
       ".text_cell_render h2 {\n",
       "/*font-size: 21pt;*/\n",
       "    font-family: 'Titillium Web', sans-serif;\n",
       "margin-bottom: 0.1em;\n",
       "margin-top: 0.3em;\n",
       "color:#595959;\n",
       "}\n",
       ".text_cell_render h3 {\n",
       "    font-family: 'Titillium Web', sans-serif;\n",
       "/*font-size: 19pt;*/\n",
       "margin-bottom: 0.1em;\n",
       "margin-top: 0.3em;\n",
       "color:#595959;\n",
       "}\n",
       ".text_cell_render h4 {\n",
       "    font-family: 'Titillium Web', sans-serif;\n",
       "/*font-size: 17pt;*/\n",
       "margin-bottom: 0.1em;\n",
       "margin-top: 0.3em;\n",
       "color:#595959;\n",
       "}\n",
       ".text_cell_render h5 {\n",
       "    font-family: 'Titillium Web', sans-serif;\n",
       "/*font-size: 15pt;*/\n",
       "margin-bottom: 0.1em;\n",
       "margin-top: 0.3em;\n",
       "color:#595959;\n",
       "}\n",
       "div.text_cell_render{\n",
       "line-height: 120%;\n",
       "font-size: 100%;\n",
       "font-weight: 400;\n",
       "text-align: justify;\n",
       "margin-left:0em;\n",
       "margin-right:0em;\n",
       "}\n",
       ".reveal div.text_cell_render{\n",
       "line-height: 120%;\n",
       "font-size: 74%;\n",
       "font-weight: 400;\n",
       "text-align: justify;\n",
       "margin-left:0em;\n",
       "margin-right:0em;\n",
       "}\n",
       "\n",
       ".reveal h2 {\n",
       "font-family: 'Titillium Web', sans-serif;\n",
       "/*font-size: 24pt;*/\n",
       "font-weight: bold;\n",
       "margin-bottom: 0.1em;\n",
       "margin-top: 0.5em;\n",
       "color:#595959;\n",
       "}\n",
       ".reveal h3 {\n",
       "font-family: 'Titillium Web', sans-serif;\n",
       "/*font-size: 24pt;*/\n",
       "font-weight: bold;\n",
       "margin-bottom: 0.1em;\n",
       "margin-top: 0.5em;\n",
       "color:#595959;\n",
       "}\n",
       ".reveal h4 {\n",
       "font-family: 'Titillium Web', sans-serif;\n",
       "font-weight: bold;\n",
       "margin-bottom: 0.1em;\n",
       "margin-top: 0.5em;\n",
       "color:#595959;\n",
       "}\n",
       ".reveal .code_cell {\n",
       "    font-size: 83%;\n",
       "}\n",
       ".reveal code {\n",
       "font-family: 'Inconsolata', monospace;\n",
       "}\n",
       ".reveal pre {\n",
       "font-family: 'Inconsolata', monospace;\n",
       "}\n",
       "code {\n",
       "font-family: 'Inconsolata', monospace;\n",
       "}\n",
       "pre {\n",
       "font-family: 'Inconsolata', monospace;\n",
       "}\n",
       ".CodeMirror{\n",
       "font-family: \"Inconsolata\", monospace;\n",
       "}\n",
       "</style>\n"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "execution_count": 63,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# this code is here for cosmetic reasons\n",
    "from IPython.core.display import HTML\n",
    "from urllib.request import urlopen\n",
    "HTML(urlopen('https://raw.githubusercontent.com/lmarti/jupyter_custom/master/custom.include').read().decode('utf-8'))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "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": "3.6.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}