LocalizedRBM.ipynb

{
 "metadata": {
  "name": ""
 },
 "nbformat": 3,
 "nbformat_minor": 0,
 "worksheets": [
  {
   "cells": [
    {
     "cell_type": "raw",
     "metadata": {},
     "source": [
      "In this file, we will try to formalize the localized RBM."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "import RBM as RBM\n",
      "import theanoRBM\n",
      "import numpy as np\n",
      "import amitgroup.io.mnist as mn\n",
      "import amitgroup.plot as gr\n",
      "import amitgroup.features as features\n",
      "import amitgroup as ag\n",
      "from sklearn import svm\n",
      "from sklearn import neighbors\n",
      "from experimentRBMSep import subsample\n",
      "from experimentRBMSep import spreading\n",
      "import matplotlib.pylab as plt\n",
      "import math\n",
      "import scipy\n",
      "import scikitRBM"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stderr",
       "text": [
        "/Users/jiajunshen/Library/Enthought/Canopy_64bit/User/lib/python2.7/site-packages/scipy/lib/_util.py:35: DeprecationWarning: Module scipy.linalg.blas.fblas is deprecated, use scipy.linalg.blas instead\n",
        "  DeprecationWarning)\n",
        "/Users/jiajunshen/Library/Enthought/Canopy_64bit/User/lib/python2.7/site-packages/scipy/lib/_util.py:35: DeprecationWarning: Module scipy.linalg.blas.fblas is deprecated, use scipy.linalg.blas instead\n",
        "  DeprecationWarning)\n",
        "/Users/jiajunshen/Library/Enthought/Canopy_64bit/User/lib/python2.7/site-packages/scipy/lib/_util.py:35: DeprecationWarning: Module scipy.linalg.blas.fblas is deprecated, use scipy.linalg.blas instead\n",
        "  DeprecationWarning)\n",
        "/Users/jiajunshen/Library/Enthought/Canopy_64bit/User/lib/python2.7/site-packages/scipy/lib/_util.py:35: DeprecationWarning: Module scipy.linalg.blas.fblas is deprecated, use scipy.linalg.blas instead\n",
        "  DeprecationWarning)\n",
        "/Users/jiajunshen/Library/Enthought/Canopy_64bit/User/lib/python2.7/site-packages/scipy/lib/_util.py:35: DeprecationWarning: Module scipy.linalg.blas.fblas is deprecated, use scipy.linalg.blas instead\n",
        "  DeprecationWarning)\n",
        "/Users/jiajunshen/Library/Enthought/Canopy_64bit/User/lib/python2.7/site-packages/scipy/lib/_util.py:35: DeprecationWarning: Module scipy.linalg.blas.fblas is deprecated, use scipy.linalg.blas instead\n",
        "  DeprecationWarning)\n",
        "/Users/jiajunshen/Library/Enthought/Canopy_64bit/User/lib/python2.7/site-packages/scipy/lib/_util.py:35: DeprecationWarning: Module scipy.linalg.blas.fblas is deprecated, use scipy.linalg.blas instead\n",
        "  DeprecationWarning)\n",
        "/Users/jiajunshen/Library/Enthought/Canopy_64bit/User/lib/python2.7/site-packages/scipy/lib/_util.py:35: DeprecationWarning: Module scipy.linalg.blas.fblas is deprecated, use scipy.linalg.blas instead\n",
        "  DeprecationWarning)\n"
       ]
      }
     ],
     "prompt_number": 1
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "def get_patches(image):\n",
      "        '''\n",
      "        Toy Function: setting not as parameter\n",
      "        Settings Here Need:\n",
      "1.sample_per_image\n",
      "2.patch_Size\n",
      "3.patch_frame\n",
      "4.bedges\n",
      "5.threshold\n",
      "\n",
      "        '''\n",
      "        patchesSize = 7\n",
      "        bedges_settings=dict(radius = 2, spread = 'orthogonal',contrast_insensitive = False)\n",
      "        settings = {}\n",
      "        settings['samples_per_image'] = 10\n",
      "        settings['patch_Size'] = [patchesSize,patchesSize]\n",
      "        settings['patch_frame'] = 1\n",
      "        settings['bedges'] = dict(k=5,radius=1,minimum_contrast=0.05,contrast_insensitive=True)\n",
      "        settings['bedges'].update(bedges_settings)\n",
      "        settings['threshold'] =  10\n",
      "\n",
      "        samples_per_image = settings['samples_per_image']\n",
      "        patchSize = settings['patch_Size']\n",
      "        fr = settings['patch_frame']\n",
      "        the_patches = []\n",
      "        img = image\n",
      "        w,h = [img.shape[i]-patchSize[i] + 1 for i in xrange(2)]\n",
      "        edges = ag.features.bedges(img,**settings['bedges'])\n",
      "        #To get Patches From Images#\n",
      "        for samples in xrange(samples_per_image):\n",
      "            temp_patches_num = 0;\n",
      "            temp_patches = numpy.zeros((int(patchesSize),int(patchesSize)));\n",
      "            for tries in xrange(20):\n",
      "                x,y = random.randint(0,w - 1), random.randint(0,h-1)\n",
      "                selection = [slice(x,x+patchSize[0]),slice(y,y + patchSize[1])]\n",
      "                imagePatch = img[selection]\n",
      "                edgepatch = edges[selection]\n",
      "                edgeNum = edgepatch[fr:-fr,fr:-fr].sum()\n",
      "                #print edgeNum\n",
      "                if edgeNum < settings['threshold']:\n",
      "                    continue;\n",
      "                else:\n",
      "                    if edgeNum >= temp_patches_num:\n",
      "                        temp_patches_num = edgeNum;\n",
      "                        temp_patches = imagePatch      \n",
      "            the_patches.append(temp_patches)\n",
      "        return the_patches"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 2
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "    digits = [0,1,2,3,4,5,6,7,8,9]\n",
      "    images,labels = mn.load_mnist('training',digits,'../data/',False,slice(0,6000,5),True,False)\n",
      "    c = map(get_patches,images)\n",
      "    raw_patches = []\n",
      "    for patches in c:\n",
      "        raw_patches.extend(patches)\n",
      "    raw_patches = np.array(raw_patches)\n",
      "    print raw_patches.shape"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "(12000, 7, 7)\n"
       ]
      }
     ],
     "prompt_number": 3
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "gr.images(raw_patches[0:20].reshape(20,5,5))"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAaYAAAD8CAYAAAA8EwAmAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nJgBS+Q9h8s1mXEAnvwAAAABJRU5ErkJggg==\n",
       "text": [
        "<matplotlib.figure.Figure at 0x10295a150>"
       ]
      }
     ],
     "prompt_number": 98
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "\"\"\"This tutorial introduces restricted boltzmann machines (RBM) using Theano.\n",
      "\n",
      "Boltzmann Machines (BMs) are a particular form of energy-based model which\n",
      "contain hidden variables. Restricted Boltzmann Machines further restrict BMs\n",
      "to those without visible-visible and hidden-hidden connections.\n",
      "\"\"\"\n",
      "import cPickle\n",
      "import gzip\n",
      "import time\n",
      "import PIL.Image\n",
      "\n",
      "import numpy\n",
      "\n",
      "import theano\n",
      "import theano.tensor as T\n",
      "import os\n",
      "\n",
      "from theano.tensor.shared_randomstreams import RandomStreams\n",
      "\n",
      "from utils import tile_raster_images\n",
      "from logistic_sgd import load_data\n",
      "\n",
      "\n",
      "class RBM(object):\n",
      "    \"\"\"Restricted Boltzmann Machine (RBM)  \"\"\"\n",
      "    def __init__(self, input=None, n_visible=784, n_hidden=500, \\\n",
      "        W=None, hbias=None, vbias=None, numpy_rng=None,\n",
      "        theano_rng=None):\n",
      "        \"\"\"\n",
      "        RBM constructor. Defines the parameters of the model along with\n",
      "        basic operations for inferring hidden from visible (and vice-versa),\n",
      "        as well as for performing CD updates.\n",
      "\n",
      "        :param input: None for standalone RBMs or symbolic variable if RBM is\n",
      "        part of a larger graph.\n",
      "\n",
      "        :param n_visible: number of visible units\n",
      "\n",
      "        :param n_hidden: number of hidden units\n",
      "\n",
      "        :param W: None for standalone RBMs or symbolic variable pointing to a\n",
      "        shared weight matrix in case RBM is part of a DBN network; in a DBN,\n",
      "        the weights are shared between RBMs and layers of a MLP\n",
      "\n",
      "        :param hbias: None for standalone RBMs or symbolic variable pointing\n",
      "        to a shared hidden units bias vector in case RBM is part of a\n",
      "        different network\n",
      "\n",
      "        :param vbias: None for standalone RBMs or a symbolic variable\n",
      "        pointing to a shared visible units bias\n",
      "        \"\"\"\n",
      "\n",
      "        self.n_visible = n_visible\n",
      "        self.n_hidden = n_hidden\n",
      "\n",
      "        if numpy_rng is None:\n",
      "            # create a number generator\n",
      "            numpy_rng = numpy.random.RandomState(1234)\n",
      "\n",
      "        if theano_rng is None:\n",
      "            theano_rng = RandomStreams(numpy_rng.randint(2 ** 30))\n",
      "\n",
      "        if W is None:\n",
      "            # W is initialized with `initial_W` which is uniformely\n",
      "            # sampled from -4*sqrt(6./(n_visible+n_hidden)) and\n",
      "            # 4*sqrt(6./(n_hidden+n_visible)) the output of uniform if\n",
      "            # converted using asarray to dtype theano.config.floatX so\n",
      "            # that the code is runable on GPU\n",
      "            initial_W = numpy.asarray(numpy_rng.uniform(\n",
      "                      low=-4 * numpy.sqrt(6. / (n_hidden + n_visible)),\n",
      "                      high=4 * numpy.sqrt(6. / (n_hidden + n_visible)),\n",
      "                      size=(n_visible, n_hidden)),\n",
      "                      dtype=theano.config.floatX)\n",
      "            # theano shared variables for weights and biases\n",
      "            W = theano.shared(value=initial_W, name='W', borrow=True)\n",
      "\n",
      "        if hbias is None:\n",
      "            # create shared variable for hidden units bias\n",
      "            hbias = theano.shared(value=numpy.zeros(n_hidden,\n",
      "                                                    dtype=theano.config.floatX),\n",
      "                                  name='hbias', borrow=True)\n",
      "\n",
      "        if vbias is None:\n",
      "            # create shared variable for visible units bias\n",
      "            vbias = theano.shared(value=numpy.zeros(n_visible,\n",
      "                                                    dtype=theano.config.floatX),\n",
      "                                  name='vbias', borrow=True)\n",
      "\n",
      "        # initialize input layer for standalone RBM or layer0 of DBN\n",
      "        self.input = input\n",
      "        if not input:\n",
      "            self.input = T.matrix('input')\n",
      "\n",
      "        self.W = W\n",
      "        self.hbias = hbias\n",
      "        self.vbias = vbias\n",
      "        self.theano_rng = theano_rng\n",
      "        # **** WARNING: It is not a good idea to put things in this list\n",
      "        # other than shared variables created in this function.\n",
      "        self.params = [self.W, self.hbias, self.vbias]\n",
      "\n",
      "    def free_energy(self, v_sample):\n",
      "        ''' Function to compute the free energy '''\n",
      "        wx_b = T.dot(v_sample, self.W) + self.hbias\n",
      "        vbias_term = T.dot(v_sample, self.vbias)\n",
      "        hidden_term = T.sum(T.log(1 + T.exp(wx_b)), axis=1)\n",
      "        return -hidden_term - vbias_term\n",
      "\n",
      "    def propup(self, vis):\n",
      "        '''This function propagates the visible units activation upwards to\n",
      "        the hidden units\n",
      "\n",
      "        Note that we return also the pre-sigmoid activation of the\n",
      "        layer. As it will turn out later, due to how Theano deals with\n",
      "        optimizations, this symbolic variable will be needed to write\n",
      "        down a more stable computational graph (see details in the\n",
      "        reconstruction cost function)\n",
      "\n",
      "        '''\n",
      "        pre_sigmoid_activation = T.dot(vis, self.W) + self.hbias\n",
      "        return [pre_sigmoid_activation, T.nnet.sigmoid(pre_sigmoid_activation)]\n",
      "\n",
      "    def sample_h_given_v(self, v0_sample):\n",
      "        ''' This function infers state of hidden units given visible units '''\n",
      "        # compute the activation of the hidden units given a sample of\n",
      "        # the visibles\n",
      "        pre_sigmoid_h1, h1_mean = self.propup(v0_sample)\n",
      "        # get a sample of the hiddens given their activation\n",
      "        # Note that theano_rng.binomial returns a symbolic sample of dtype\n",
      "        # int64 by default. If we want to keep our computations in floatX\n",
      "        # for the GPU we need to specify to return the dtype floatX\n",
      "        h1_sample = self.theano_rng.binomial(size=h1_mean.shape,\n",
      "                                             n=1, p=h1_mean,\n",
      "                                             dtype=theano.config.floatX)\n",
      "        return [pre_sigmoid_h1, h1_mean, h1_sample]\n",
      "\n",
      "    def propdown(self, hid):\n",
      "        '''This function propagates the hidden units activation downwards to\n",
      "        the visible units\n",
      "\n",
      "        Note that we return also the pre_sigmoid_activation of the\n",
      "        layer. As it will turn out later, due to how Theano deals with\n",
      "        optimizations, this symbolic variable will be needed to write\n",
      "        down a more stable computational graph (see details in the\n",
      "        reconstruction cost function)\n",
      "\n",
      "        '''\n",
      "        pre_sigmoid_activation = T.dot(hid, self.W.T) + self.vbias\n",
      "        return [pre_sigmoid_activation, T.nnet.sigmoid(pre_sigmoid_activation)]\n",
      "\n",
      "    def sample_v_given_h(self, h0_sample):\n",
      "        ''' This function infers state of visible units given hidden units '''\n",
      "        # compute the activation of the visible given the hidden sample\n",
      "        pre_sigmoid_v1, v1_mean = self.propdown(h0_sample)\n",
      "        # get a sample of the visible given their activation\n",
      "        # Note that theano_rng.binomial returns a symbolic sample of dtype\n",
      "        # int64 by default. If we want to keep our computations in floatX\n",
      "        # for the GPU we need to specify to return the dtype floatX\n",
      "        v1_sample = self.theano_rng.binomial(size=v1_mean.shape,\n",
      "                                             n=1, p=v1_mean,\n",
      "                                             dtype=theano.config.floatX)\n",
      "        return [pre_sigmoid_v1, v1_mean, v1_sample]\n",
      "\n",
      "    def gibbs_hvh(self, h0_sample):\n",
      "        ''' This function implements one step of Gibbs sampling,\n",
      "            starting from the hidden state'''\n",
      "        pre_sigmoid_v1, v1_mean, v1_sample = self.sample_v_given_h(h0_sample)\n",
      "        pre_sigmoid_h1, h1_mean, h1_sample = self.sample_h_given_v(v1_sample)\n",
      "        return [pre_sigmoid_v1, v1_mean, v1_sample,\n",
      "                pre_sigmoid_h1, h1_mean, h1_sample]\n",
      "\n",
      "    def gibbs_vhv(self, v0_sample):\n",
      "        ''' This function implements one step of Gibbs sampling,\n",
      "            starting from the visible state'''\n",
      "        pre_sigmoid_h1, h1_mean, h1_sample = self.sample_h_given_v(v0_sample)\n",
      "        pre_sigmoid_v1, v1_mean, v1_sample = self.sample_v_given_h(h1_sample)\n",
      "        return [pre_sigmoid_h1, h1_mean, h1_sample,\n",
      "                pre_sigmoid_v1, v1_mean, v1_sample]\n",
      "\n",
      "    def get_cost_updates(self, lr=0.1, persistent=None, k=1):\n",
      "        \"\"\"This functions implements one step of CD-k or PCD-k\n",
      "\n",
      "        :param lr: learning rate used to train the RBM\n",
      "\n",
      "        :param persistent: None for CD. For PCD, shared variable\n",
      "            containing old state of Gibbs chain. This must be a shared\n",
      "            variable of size (batch size, number of hidden units).\n",
      "\n",
      "        :param k: number of Gibbs steps to do in CD-k/PCD-k\n",
      "\n",
      "        Returns a proxy for the cost and the updates dictionary. The\n",
      "        dictionary contains the update rules for weights and biases but\n",
      "        also an update of the shared variable used to store the persistent\n",
      "        chain, if one is used.\n",
      "\n",
      "        \"\"\"\n",
      "\n",
      "        # compute positive phase\n",
      "        pre_sigmoid_ph, ph_mean, ph_sample = self.sample_h_given_v(self.input)\n",
      "\n",
      "        # decide how to initialize persistent chain:\n",
      "        # for CD, we use the newly generate hidden sample\n",
      "        # for PCD, we initialize from the old state of the chain\n",
      "        if persistent is None:\n",
      "            chain_start = ph_sample\n",
      "        else:\n",
      "            chain_start = persistent\n",
      "\n",
      "        # perform actual negative phase\n",
      "        # in order to implement CD-k/PCD-k we need to scan over the\n",
      "        # function that implements one gibbs step k times.\n",
      "        # Read Theano tutorial on scan for more information :\n",
      "        # http://deeplearning.net/software/theano/library/scan.html\n",
      "        # the scan will return the entire Gibbs chain\n",
      "        [pre_sigmoid_nvs, nv_means, nv_samples,\n",
      "         pre_sigmoid_nhs, nh_means, nh_samples], updates = \\\n",
      "            theano.scan(self.gibbs_hvh,\n",
      "                    # the None are place holders, saying that\n",
      "                    # chain_start is the initial state corresponding to the\n",
      "                    # 6th output\n",
      "                    outputs_info=[None,  None,  None, None, None, chain_start],\n",
      "                    n_steps=k)\n",
      "\n",
      "        # determine gradients on RBM parameters\n",
      "        # not that we only need the sample at the end of the chain\n",
      "        chain_end = nv_samples[-1]\n",
      "\n",
      "        cost = T.mean(self.free_energy(self.input)) - T.mean(\n",
      "            self.free_energy(chain_end))\n",
      "        # We must not compute the gradient through the gibbs sampling\n",
      "        gparams = T.grad(cost, self.params, consider_constant=[chain_end])\n",
      "\n",
      "        # constructs the update dictionary\n",
      "        for gparam, param in zip(gparams, self.params):\n",
      "            # make sure that the learning rate is of the right dtype\n",
      "            updates[param] = param - gparam * T.cast(lr,\n",
      "                                                    dtype=theano.config.floatX)\n",
      "        if persistent:\n",
      "            # Note that this works only if persistent is a shared variable\n",
      "            updates[persistent] = nh_samples[-1]\n",
      "            # pseudo-likelihood is a better proxy for PCD\n",
      "            monitoring_cost = self.get_pseudo_likelihood_cost(updates)\n",
      "        else:\n",
      "            # reconstruction cross-entropy is a better proxy for CD\n",
      "            monitoring_cost = self.get_reconstruction_cost(updates,\n",
      "                                                           pre_sigmoid_nvs[-1])\n",
      "\n",
      "        return monitoring_cost, updates\n",
      "\n",
      "    def get_pseudo_likelihood_cost(self, updates):\n",
      "        \"\"\"Stochastic approximation to the pseudo-likelihood\"\"\"\n",
      "\n",
      "        # index of bit i in expression p(x_i | x_{\\i})\n",
      "        bit_i_idx = theano.shared(value=0, name='bit_i_idx')\n",
      "\n",
      "        # binarize the input image by rounding to nearest integer\n",
      "        xi = T.round(self.input)\n",
      "\n",
      "        # calculate free energy for the given bit configuration\n",
      "        fe_xi = self.free_energy(xi)\n",
      "\n",
      "        # flip bit x_i of matrix xi and preserve all other bits x_{\\i}\n",
      "        # Equivalent to xi[:,bit_i_idx] = 1-xi[:, bit_i_idx], but assigns\n",
      "        # the result to xi_flip, instead of working in place on xi.\n",
      "        xi_flip = T.set_subtensor(xi[:, bit_i_idx], 1 - xi[:, bit_i_idx])\n",
      "\n",
      "        # calculate free energy with bit flipped\n",
      "        fe_xi_flip = self.free_energy(xi_flip)\n",
      "\n",
      "        # equivalent to e^(-FE(x_i)) / (e^(-FE(x_i)) + e^(-FE(x_{\\i})))\n",
      "        cost = T.mean(self.n_visible * T.log(T.nnet.sigmoid(fe_xi_flip -\n",
      "                                                            fe_xi)))\n",
      "\n",
      "        # increment bit_i_idx % number as part of updates\n",
      "        updates[bit_i_idx] = (bit_i_idx + 1) % self.n_visible\n",
      "\n",
      "        return cost\n",
      "\n",
      "    def get_reconstruction_cost(self, updates, pre_sigmoid_nv):\n",
      "        \"\"\"Approximation to the reconstruction error\n",
      "\n",
      "        Note that this function requires the pre-sigmoid activation as\n",
      "        input.  To understand why this is so you need to understand a\n",
      "        bit about how Theano works. Whenever you compile a Theano\n",
      "        function, the computational graph that you pass as input gets\n",
      "        optimized for speed and stability.  This is done by changing\n",
      "        several parts of the subgraphs with others.  One such\n",
      "        optimization expresses terms of the form log(sigmoid(x)) in\n",
      "        terms of softplus.  We need this optimization for the\n",
      "        cross-entropy since sigmoid of numbers larger than 30. (or\n",
      "        even less then that) turn to 1. and numbers smaller thans\\\n",
      "        -30. turn to 0 which in terms will force theano to compute\n",
      "        log(0) and therefore we will get either -inf or NaN as\n",
      "        cost. If the value is expressed in terms of softplus we do not\n",
      "        get this undesirable behaviour. This optimization usually\n",
      "        works fine, but here we have a special case. The sigmoid is\n",
      "        applied inside the scan op, while the log is\n",
      "        outside. Therefore Theano will only see log(scan(..)) instead\n",
      "        of log(sigmoid(..)) and will not apply the wanted\n",
      "        optimization. We can not go and replace the sigmoid in scan\n",
      "        with something else also, because this only needs to be done\n",
      "        on the last step. Therefore the easiest and more efficient way\n",
      "        is to get also the pre-sigmoid activation as an output of\n",
      "        scan, and apply both the log and sigmoid outside scan such\n",
      "        that Theano can catch and optimize the expression.\n",
      "\n",
      "        \"\"\"\n",
      "\n",
      "        cross_entropy = T.mean(\n",
      "                T.sum(self.input * T.log(T.nnet.sigmoid(pre_sigmoid_nv)) +\n",
      "                (1 - self.input) * T.log(1 - T.nnet.sigmoid(pre_sigmoid_nv)),\n",
      "                      axis=1))\n",
      "\n",
      "        return cross_entropy"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 4
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "raw_patches.shape"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 5,
       "text": [
        "(12000, 7, 7)"
       ]
      }
     ],
     "prompt_number": 5
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "    %paste\n",
      "    learning_rate=0.1\n",
      "    training_epochs=100\n",
      "    dataset='/Users/jiajunshen/Documents/Jiajun/Research/Research In Summer and Autumn 2013/DeepLearningTutorials/data/mnist.pkl.gz'\n",
      "    batch_size=20\n",
      "    n_chains=100\n",
      "    n_samples=100\n",
      "    output_folder='rbm_plots'\n",
      "    n_hidden=50\n",
      "    patchSize = 12\n",
      "    \"\"\"\n",
      "    Demonstrate how to train and afterwards sample from it using Theano.\n",
      "\n",
      "    This is demonstrated on MNIST.\n",
      "\n",
      "    :param learning_rate: learning rate used for training the RBM\n",
      "\n",
      "    :param training_epochs: number of epochs used for training\n",
      "\n",
      "    :param dataset: path the the pickled dataset\n",
      "\n",
      "    :param batch_size: size of a batch used to train the RBM\n",
      "\n",
      "    :param n_chains: number of parallel Gibbs chains to be used for sampling\n",
      "\n",
      "    :param n_samples: number of samples to plot for each chain\n",
      "\n",
      "    \"\"\"\n",
      "    datasets = theanoRBM.load_data(dataset)\n",
      "#    theanoRBM.load_data()\n",
      "#    numpy.asarray(trainingData,dtype = theano.config.floatX)\n",
      "    train_set_x = theano.shared(numpy.asarray(raw_patches.reshape(12000,-1),dtype=theano.config.floatX), borrow=True)\n",
      "    #train_set_x, train_set_y = datasets[0]\n",
      "    #test_set_x, test_set_y = datasets[2]\n",
      "\n",
      "    # compute number of minibatches for training, validation and testing\n",
      "    n_train_batches = train_set_x.get_value(borrow=True).shape[0] / batch_size\n",
      "\n",
      "    # allocate symbolic variables for the data\n",
      "    index = T.lscalar()    # index to a [mini]batch\n",
      "    x = T.matrix('x')  # the data is presented as rasterized images\n",
      "\n",
      "    rng = numpy.random.RandomState(123)\n",
      "    theano_rng = RandomStreams(rng.randint(2 ** 30))\n",
      "\n",
      "    # initialize storage for the persistent chain (state = hidden\n",
      "    # layer of chain)\n",
      "    persistent_chain = theano.shared(numpy.zeros((batch_size, n_hidden),\n",
      "                                                 dtype=theano.config.floatX),\n",
      "                                     borrow=True)\n",
      "\n",
      "    # construct the RBM class\n",
      "    rbm = theanoRBM.RBM(input=x, n_visible=patchSize * patchSize,\n",
      "              n_hidden=n_hidden, numpy_rng=rng, theano_rng=theano_rng)\n",
      "\n",
      "    # get the cost and the gradient corresponding to one step of CD-15\n",
      "    cost, updates = rbm.get_cost_updates(lr=learning_rate,\n",
      "                                         persistent=persistent_chain, k=1)\n",
      "\n",
      "    #################################\n",
      "    #     Training the RBM          #\n",
      "    #################################\n",
      "    if not os.path.isdir(output_folder):\n",
      "        os.makedirs(output_folder)\n",
      "    os.chdir(output_folder)\n",
      "\n",
      "    # it is ok for a theano function to have no output\n",
      "    # the purpose of train_rbm is solely to update the RBM parameters\n",
      "    train_rbm = theano.function([index], cost,\n",
      "           updates=updates,\n",
      "           givens={x: train_set_x[index * batch_size:\n",
      "                                  (index + 1) * batch_size]},\n",
      "           name='train_rbm')\n",
      "\n",
      "    plotting_time = 0.\n",
      "    start_time = time.clock()\n",
      "\n",
      "    # go through training epochs\n",
      "    for epoch in xrange(training_epochs):\n",
      "\n",
      "        # go through the training set\n",
      "        mean_cost = []\n",
      "        for batch_index in xrange(n_train_batches):\n",
      "            mean_cost += [train_rbm(batch_index)]\n",
      "\n",
      "        print 'Training epoch %d, cost is ' % epoch, numpy.mean(mean_cost)\n",
      "\n",
      "        # Plot filters after each training epoch\n",
      "        plotting_start = time.clock()\n",
      "        # Construct image from the weight matrix\n",
      "        image = PIL.Image.fromarray(tile_raster_images(\n",
      "                 X=rbm.W.get_value(borrow=True).T,\n",
      "                 img_shape=(patchSize, patchSize), tile_shape=(10, 10),\n",
      "                 tile_spacing=(1, 1)))\n",
      "        image.save('filters_at_epoch_%i.png' % epoch)\n",
      "        plotting_stop = time.clock()\n",
      "        plotting_time += (plotting_stop - plotting_start)\n",
      "\n",
      "    end_time = time.clock()\n",
      "\n",
      "    pretraining_time = (end_time - start_time) - plotting_time\n",
      "\n",
      "    print ('Training took %f minutes' % (pretraining_time / 60.))\n",
      "\n",
      "    #################################\n",
      "    #     Sampling from the RBM     #\n",
      "    #################################\n",
      "    # find out the number of test samples"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "... loading data\n",
        "(50000,)"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Training epoch 0, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -69.3184244792\n",
        "Training epoch 1, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -58.905546875\n",
        "Training epoch 2, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -55.7598307292\n",
        "Training epoch 3, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -54.1412727865\n",
        "Training epoch 4, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -52.7653938802\n",
        "Training epoch 5, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -50.8998860677\n",
        "Training epoch 6, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -51.2156705729\n",
        "Training epoch 7, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -50.0221126302\n",
        "Training epoch 8, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -49.5744270833\n",
        "Training epoch 9, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -48.6675716146\n",
        "Training epoch 10, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -49.067985026\n",
        "Training epoch 11, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -47.3938216146\n",
        "Training epoch 12, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -48.1953808594\n",
        "Training epoch 13, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -47.7952929687\n",
        "Training epoch 14, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -46.52421875\n",
        "Training epoch 15, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -47.1293782552\n",
        "Training epoch 16, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -46.3890364583\n",
        "Training epoch 17, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -46.585563151\n",
        "Training epoch 18, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -46.3145572917\n",
        "Training epoch 19, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -46.1986914062\n",
        "Training epoch 20, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -46.5490071615\n",
        "Training epoch 21, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -46.3996809896\n",
        "Training epoch 22, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -46.2733072917\n",
        "Training epoch 23, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -45.5831282552\n",
        "Training epoch 24, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -46.3627864583\n",
        "Training epoch 25, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -46.0975358073\n",
        "Training epoch 26, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -46.0956445313\n",
        "Training epoch 27, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -45.9636295573\n",
        "Training epoch 28, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -45.2847884115\n",
        "Training epoch 29, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -45.0264550781\n",
        "Training epoch 30, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -45.0716503906\n",
        "Training epoch 31, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -45.2592675781\n",
        "Training epoch 32, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -45.3884375\n",
        "Training epoch 33, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -45.5009505208\n",
        "Training epoch 34, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -45.689140625\n",
        "Training epoch 35, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -45.0691080729\n",
        "Training epoch 36, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -45.4315039063\n",
        "Training epoch 37, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -44.7954329427\n",
        "Training epoch 38, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -44.626171875\n",
        "Training epoch 39, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -45.0794563802\n",
        "Training epoch 40, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -44.4477148437\n",
        "Training epoch 41, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -45.1343391927\n",
        "Training epoch 42, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -44.8649316406\n",
        "Training epoch 43, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -44.8760970052\n",
        "Training epoch 44, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -43.8630891927\n",
        "Training epoch 45, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -45.1799121094\n",
        "Training epoch 46, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -45.1880110677\n",
        "Training epoch 47, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -44.0994563802\n",
        "Training epoch 48, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -44.5924641927\n",
        "Training epoch 49, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -44.70765625\n",
        "Training epoch 50, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -44.3988509115\n",
        "Training epoch 51, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -45.4239355469\n",
        "Training epoch 52, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -45.2680371094\n",
        "Training epoch 53, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -44.3957356771\n",
        "Training epoch 54, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -44.8255924479\n",
        "Training epoch 55, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -43.9373404948\n",
        "Training epoch 56, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -44.2569401042\n",
        "Training epoch 57, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -44.5143424479\n",
        "Training epoch 58, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -44.896656901\n",
        "Training epoch 59, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -44.314703776\n",
        "Training epoch 60, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -45.2401464844\n",
        "Training epoch 61, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -44.2864550781\n",
        "Training epoch 62, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -43.2404166667\n",
        "Training epoch 63, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -44.9660416667\n",
        "Training epoch 64, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -44.4776009115\n",
        "Training epoch 65, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -44.144469401\n",
        "Training epoch 66, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -44.541813151\n",
        "Training epoch 67, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -45.2258365885\n",
        "Training epoch 68, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -43.7660904948\n",
        "Training epoch 69, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -44.0267708333\n",
        "Training epoch 70, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -44.4229101562\n",
        "Training epoch 71, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -43.9368652344\n",
        "Training epoch 72, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -44.5844563802\n",
        "Training epoch 73, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -44.6180924479\n",
        "Training epoch 74, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -43.6848990885\n",
        "Training epoch 75, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -43.9047102865\n",
        "Training epoch 76, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -44.4993359375\n",
        "Training epoch 77, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -43.2605013021\n",
        "Training epoch 78, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -44.7778841146\n",
        "Training epoch 79, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -43.6359049479\n",
        "Training epoch 80, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -43.6869986979\n",
        "Training epoch 81, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -44.606936849\n",
        "Training epoch 82, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -44.396344401\n",
        "Training epoch 83, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -43.1224967448\n",
        "Training epoch 84, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -44.1198925781\n",
        "Training epoch 85, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -43.7551236979\n",
        "Training epoch 86, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -43.9957063802\n",
        "Training epoch 87, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -44.616077474\n",
        "Training epoch 88, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -43.8160221354\n",
        "Training epoch 89, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -43.8225455729\n",
        "Training epoch 90, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -44.5859277344\n",
        "Training epoch 91, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -43.6843977865\n",
        "Training epoch 92, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -43.5577669271\n",
        "Training epoch 93, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -43.9218489583\n",
        "Training epoch 94, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -44.7672363281\n",
        "Training epoch 95, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -43.8571614583\n",
        "Training epoch 96, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -44.4791210938\n",
        "Training epoch 97, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -43.8224283854\n",
        "Training epoch 98, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -43.2727734375\n",
        "Training epoch 99, cost is "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " -43.7019075521\n",
        "Training took 1.099159 minutes\n"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stderr",
       "text": [
        "ERROR: Line magic function `%paste` not found.\n"
       ]
      }
     ],
     "prompt_number": 12
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "weights = np.array(rbm.W.get_value(borrow=True))\n",
      "newWeights = np.rollaxis(weights,1,0)\n",
      "import amitgroup.plot as gr\n",
      "gr.images(newWeights[0:50].reshape(50,patchSize,patchSize),zero_to_one = False, vmin = -2, vmax = 2, show = True)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "ename": "NameError",
       "evalue": "name 'rbm' is not defined",
       "output_type": "pyerr",
       "traceback": [
        "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m\n\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
        "\u001b[0;32m<ipython-input-4-421a306c28b7>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mweights\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0marray\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mrbm\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mW\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mget_value\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mborrow\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mTrue\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      2\u001b[0m \u001b[0mnewWeights\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mrollaxis\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mweights\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      3\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mamitgroup\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mplot\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0mgr\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      4\u001b[0m \u001b[0mgr\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mimages\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mnewWeights\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;36m50\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mreshape\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m50\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mpatchSize\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mpatchSize\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mzero_to_one\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mFalse\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mvmin\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m-\u001b[0m\u001b[0;36m2\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mvmax\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;36m2\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mshow\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mTrue\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
        "\u001b[0;31mNameError\u001b[0m: name 'rbm' is not defined"
       ]
      }
     ],
     "prompt_number": 4
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "from sklearn.neural_network import BernoulliRBM\n",
      "fitData = raw_patches.reshape(12000,-1)\n",
      "model = BernoulliRBM(n_components=50,learning_rate = 0.1, n_iter = 15,verbose=True)\n",
      "model.fit(fitData)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "Iteration 0, pseudo-likelihood = -22.75, time = 0.67s\n",
        "Iteration 1, pseudo-likelihood = -18.26, time = 0.63s"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Iteration 2, pseudo-likelihood = -16.90, time = 0.64s"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Iteration 3, pseudo-likelihood = -16.36, time = 0.64s"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Iteration 4, pseudo-likelihood = -15.98, time = 0.64s"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Iteration 5, pseudo-likelihood = -15.89, time = 0.64s"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Iteration 6, pseudo-likelihood = -15.86, time = 0.67s"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Iteration 7, pseudo-likelihood = -15.77, time = 0.67s"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Iteration 8, pseudo-likelihood = -15.50, time = 0.64s"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Iteration 9, pseudo-likelihood = -15.59, time = 0.64s"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Iteration 10, pseudo-likelihood = -15.39, time = 0.64s"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Iteration 11, pseudo-likelihood = -15.27, time = 0.63s"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Iteration 12, pseudo-likelihood = -15.49, time = 0.66s"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Iteration 13, pseudo-likelihood = -15.34, time = 0.65s"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Iteration 14, pseudo-likelihood = -15.61, time = 0.68s"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n"
       ]
      },
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 6,
       "text": [
        "BernoulliRBM(batch_size=10, learning_rate=0.1, n_components=50, n_iter=15,\n",
        "       random_state=None, verbose=True)"
       ]
      }
     ],
     "prompt_number": 6
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "gr.images(model.components_.reshape(50,7,7),zero_to_one = False, vmin = -2, vmax = 2)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAbAAAAEVCAYAAACBn7B3AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nABSJCQwAUCQmMABAkZjAAABF+j9ap9UsBE+OSwAAAABJRU5ErkJggg==\n",
       "text": [
        "<matplotlib.figure.Figure at 0x117863250>"
       ]
      }
     ],
     "prompt_number": 8
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "rbmLocal = scikitRBM.BernoulliRBM(n_components = 50, learning_rate = 0.1, batch_size = 10, n_iter = 15, verbose=True, getNum = 1)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 99
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "posterList = np.zeros(rbmLocal.n_iter)\n",
      "fitData = raw_patches. reshape(raw_patches.shape[0], -1)\n",
      "#fitData = images.reshape(1200,-1)\n",
      "rbmLocal.fit(fitData, winnerTakeAll=False,plList = posterList)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "Iteration 0, pseudo-likelihood = -13.71, time = 0.70s\n",
        "Iteration 1, pseudo-likelihood = -10.17, time = 0.68s"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Iteration 2, pseudo-likelihood = -9.41, time = 0.69s"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Iteration 3, pseudo-likelihood = -9.23, time = 0.69s"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Iteration 4, pseudo-likelihood = -9.18, time = 0.68s"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Iteration 5, pseudo-likelihood = -9.06, time = 0.68s"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Iteration 6, pseudo-likelihood = -8.94, time = 0.69s"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Iteration 7, pseudo-likelihood = -9.06, time = 0.69s"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Iteration 8, pseudo-likelihood = -9.01, time = 0.68s"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Iteration 9, pseudo-likelihood = -9.05, time = 0.68s"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Iteration 10, pseudo-likelihood = -8.95, time = 0.69s"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Iteration 11, pseudo-likelihood = -8.87, time = 0.69s"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Iteration 12, pseudo-likelihood = -8.90, time = 0.69s"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Iteration 13, pseudo-likelihood = -8.92, time = 0.69s"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Iteration 14, pseudo-likelihood = -8.89, time = 0.71s"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n"
       ]
      },
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 100,
       "text": [
        "BernoulliRBM(batch_size=10, getNum=1, learning_rate=0.1, n_components=50,\n",
        "       n_iter=15, random_state=None, verbose=True)"
       ]
      }
     ],
     "prompt_number": 100
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "patchSize = 5\n",
      "rbmLocal.components_ = rbmLocal.components_.mean() - rbmLocal.components_\n",
      "gr.images(rbmLocal.components_.reshape(50,patchSize,patchSize),zero_to_one = False, vmin = -2, vmax = 2)\n"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAbAAAAEUCAYAAABKw2PSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAQBc+h+JVdC1/deJ5AAAAABJRU5ErkJggg==\n",
       "text": [
        "<matplotlib.figure.Figure at 0x119fb0ad0>"
       ]
      }
     ],
     "prompt_number": 105
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "from sklearn.utils import check_random_state\n",
      "rng = check_random_state(rbmLocal.random_state)\n",
      "rbmLocal._sample_hiddens(raw_patches[0:5].reshape(5,-1),rng).shape"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 22,
       "text": [
        "(5, 50)"
       ]
      }
     ],
     "prompt_number": 22
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "test.shape"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "ename": "AttributeError",
       "evalue": "'function' object has no attribute 'shape'",
       "output_type": "pyerr",
       "traceback": [
        "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m\n\u001b[0;31mAttributeError\u001b[0m                            Traceback (most recent call last)",
        "\u001b[0;32m<ipython-input-8-3de001cc2de6>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mtest\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mshape\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
        "\u001b[0;31mAttributeError\u001b[0m: 'function' object has no attribute 'shape'"
       ]
      }
     ],
     "prompt_number": 8
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "testImages,tlbs = mn.load_mnist('testing',digits,'../data/',False,slice(0,1000),True,False)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 10
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "testImages.shape"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 11,
       "text": [
        "(1000, 28, 28)"
       ]
      }
     ],
     "prompt_number": 11
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "patchSize = 5\n",
      "def codeHiddenVector(image,hiddenNodeNum):\n",
      "    totalRange = image.shape[1] - patchSize + 1\n",
      "    hiddenVector = np.zeros((totalRange * totalRange,hiddenNodeNum),dtype=uint8)\n",
      "    localPatch = np.zeros((totalRange,totalRange,patchSize * patchSize))\n",
      "    for i in range(totalRange):\n",
      "        for j in range(totalRange):\n",
      "            localPatch[i,j,:] = image[i:i+patchSize,j:j+patchSize].ravel()\n",
      "    localPatch = localPatch.reshape((totalRange * totalRange, patchSize * patchSize))\n",
      "    hiddenVector = rbmLocal._sample_hiddens(localPatch,rng)\n",
      "    #rbmLocal.getNum = 1\n",
      "    #hiddenVector = rbmLocal._sample_hiddens_part(localPatch,rng)\n",
      "    return hiddenVector"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 74
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "testData.shape"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "ename": "NameError",
       "evalue": "name 'testData' is not defined",
       "output_type": "pyerr",
       "traceback": [
        "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m\n\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
        "\u001b[0;32m<ipython-input-24-0e16aa5422b2>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mtestData\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mshape\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
        "\u001b[0;31mNameError\u001b[0m: name 'testData' is not defined"
       ]
      }
     ],
     "prompt_number": 24
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "testData[0,9,:]"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "ename": "NameError",
       "evalue": "name 'testData' is not defined",
       "output_type": "pyerr",
       "traceback": [
        "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m\n\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
        "\u001b[0;32m<ipython-input-25-8890198ce4a2>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mtestData\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;36m9\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
        "\u001b[0;31mNameError\u001b[0m: name 'testData' is not defined"
       ]
      }
     ],
     "prompt_number": 25
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "testImages = 1 - testImages\n",
      "trainingImages = 1 - images"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 43
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "gr.images(trainingImages[0:20].reshape(20,28,28))"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAa0AAAEACAYAAAAA1CkHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAASUVORK5CYII=\n",
       "text": [
        "<matplotlib.figure.Figure at 0x115fad510>"
       ]
      }
     ],
     "prompt_number": 75
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "testData = np.array([codeHiddenVector(x,50) for x in testImages],dtype=uint8)\n",
      "testData = np.array([spreading(x,50,1,originalSize = 24) for x in testData])\n",
      "#testData = features.bspread(testData,spread = 'box',radius = 2)\n",
      "testData = np.array([subsample(x,50,12,originalSize = 24) for x in testData])"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 106
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "trainingData = np.array([codeHiddenVector(x,50) for x in trainingImages],dtype=uint8)\n",
      "trainingData = np.array([spreading(x,50,1,originalSize = 24) for x in trainingData])\n",
      "#trainingData = features.bspread(trainingData,spread = 'box', radius = 2)\n",
      "trainingData = np.array([subsample(x,50,12,originalSize = 24) for x in trainingData])"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 107
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "trainingData.shape"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 108,
       "text": [
        "(1200, 12, 12, 50)"
       ]
      }
     ],
     "prompt_number": 108
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "clf = svm.SVC(kernel='linear')\n",
      "clf.fit(trainingData.reshape((1200,-1)),labels)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 109,
       "text": [
        "SVC(C=1.0, cache_size=200, class_weight=None, coef0=0.0, degree=3, gamma=0.0,\n",
        "  kernel='linear', max_iter=-1, probability=False, random_state=None,\n",
        "  shrinking=True, tol=0.001, verbose=False)"
       ]
      }
     ],
     "prompt_number": 109
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "ty = clf.predict(testData.reshape((1000,-1)))"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 110
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "(ty==tlbs).mean()"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 111,
       "text": [
        "0.79300000000000004"
       ]
      }
     ],
     "prompt_number": 111
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "for k in range(10):\n",
      "    gr.images(1-images[k])\n",
      "    plt.imshow(trainingData[k].sum(axis = 2),interpolation = 'nearest')\n",
      "    plt.colorbar()"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAPwAAAD8CAYAAABTq8lnAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
       "text": [
        "<matplotlib.figure.Figure at 0x111adc710>"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAASgAAAEACAYAAAAA+OtCAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRK5CYII=\n",
       "text": [
        "<matplotlib.figure.Figure at 0x111e6d1d0>"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAPwAAAD8CAYAAABTq8lnAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nwUMQwUOQfwAbOzP9S3drHAAAAABJRU5ErkJggg==\n",
       "text": [
        "<matplotlib.figure.Figure at 0x110e68e10>"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAASgAAAEACAYAAAAA+OtCAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nmZMAAAAASUVORK5CYII=\n",
       "text": [
        "<matplotlib.figure.Figure at 0x111e5f290>"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAPwAAAD8CAYAAABTq8lnAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAAAElFTkSuQmCC\n",
       "text": [
        "<matplotlib.figure.Figure at 0x111a36090>"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAATEAAAD7CAYAAAACYaMOAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nEekakxgR6RqTGBHp2v8B7VzE8M5b4ZoAAAAASUVORK5CYII=\n",
       "text": [
        "<matplotlib.figure.Figure at 0x118d39510>"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAPwAAAD8CAYAAABTq8lnAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nCCJ4CPInzi7pfJIAMIQAAAAASUVORK5CYII=\n",
       "text": [
        "<matplotlib.figure.Figure at 0x110eb7910>"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAASgAAAEACAYAAAAA+OtCAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
       "text": [
        "<matplotlib.figure.Figure at 0x110e61590>"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAPwAAAD8CAYAAABTq8lnAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQmCC\n",
       "text": [
        "<matplotlib.figure.Figure at 0x111e08250>"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAASgAAAEACAYAAAAA+OtCAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRNJigSIiaf0/6MjgZI+4LroAAAAASUVORK5CYII=\n",
       "text": [
        "<matplotlib.figure.Figure at 0x119fb03d0>"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAPwAAAD8CAYAAABTq8lnAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nfgOnTvMClQAAAABJRU5ErkJggg==\n",
       "text": [
        "<matplotlib.figure.Figure at 0x111ef6ed0>"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAASgAAAEACAYAAAAA+OtCAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
       "text": [
        "<matplotlib.figure.Figure at 0x10c8394d0>"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAPwAAAD8CAYAAABTq8lnAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
       "text": [
        "<matplotlib.figure.Figure at 0x1103c5c50>"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAASgAAAEACAYAAAAA+OtCAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
       "text": [
        "<matplotlib.figure.Figure at 0x118d3f510>"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAPwAAAD8CAYAAABTq8lnAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
       "text": [
        "<matplotlib.figure.Figure at 0x11a3e2a90>"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAASgAAAEACAYAAAAA+OtCAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n/wOWw3YmwPFQUAAAAABJRU5ErkJggg==\n",
       "text": [
        "<matplotlib.figure.Figure at 0x1103cb850>"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAPwAAAD8CAYAAABTq8lnAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nCBE8hAgeQgQPIYKHkD93ZEhN1e2POwAAAABJRU5ErkJggg==\n",
       "text": [
        "<matplotlib.figure.Figure at 0x11a3fe850>"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAASgAAAEACAYAAAAA+OtCAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAASUVORK5CYII=\n",
       "text": [
        "<matplotlib.figure.Figure at 0x11a3e3f50>"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAPwAAAD8CAYAAABTq8lnAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
       "text": [
        "<matplotlib.figure.Figure at 0x118d1f6d0>"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAASgAAAEACAYAAAAA+OtCAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
       "text": [
        "<matplotlib.figure.Figure at 0x118d1b4d0>"
       ]
      }
     ],
     "prompt_number": 112
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [],
     "language": "python",
     "metadata": {},
     "outputs": []
    }
   ],
   "metadata": {}
  }
 ]
}