reconstruct_img.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Train a  Bernoulli-Bernoulli RBM (BBRBM) to reconstruct img"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import timeit\n",
    "from tfrbm.bbrbm import BBRBM\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "bm = BBRBM(n_visible=36,n_hidden= 56)\n",
    "bm.image_height = 6"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### dataset[i][0] xor dataset[i][1] = dataset[i][2]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "f = open(\"6x6.txt\")\n",
    "dataset = np.array([[int(data) for data in f.read() if data in \"01\"]])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "plt.figure()\n",
    "plt.imshow(dataset[0].reshape(bm.image_height, -1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Stopped training early because the model can reconstruct the inputs\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "err = bm.fit(dataset, n_epoches=50, learning_rate = 1000, decay = 0.01, epochs_to_test = 1, early_stop = True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Measure calculation time on CPU"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "import time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Time to calculate hidden nodes: 0.0109770298004\n",
      "Time to reconstruct visible nodes: 0.0105681419373\n"
     ]
    }
   ],
   "source": [
    "start_time = time.time()\n",
    "bm.sess.run(bm.hidden_p, feed_dict={bm.x: dataset})\n",
    "hidden_elapsed_time = time.time() - start_time\n",
    "print \"Time to calculate hidden nodes:\",hidden_elapsed_time\n",
    "\n",
    "start_time = time.time()\n",
    "bm.sess.run(bm.visible_recon_p, feed_dict={bm.x: dataset})\n",
    "visible_elapsed_time = time.time() - start_time - hidden_elapsed_time\n",
    "print \"Time to reconstruct visible nodes:\",visible_elapsed_time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAPgAAAEPCAYAAABvMQkeAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvIxREBQAACDxJREFUeJzt3H+o3XUdx/HXS66w2qSMClmuUkfm+qMNK6UftBgWIhv+NTJFhaKs8L8aDGNN0ZB+ISESUe02caUFk4xEGWkxFY1+2A8ZNnV5dc0frJW6exXu3v1xvoPvOSfaPXed85WXzwcMzjnf+/1832fseT/nnO3OVSUAmU7oegAA40PgQDACB4IROBCMwIFgBA4EI/AgttfafqrrOfDqQeBjZnuf7VnbL9o+YHva9rKu5zpetsv2yglcZ9r2teO+TioCn4z1VbVM0mpJayRt7ngevEYQ+ARV1QFJd6kXuiTJ9gW2/2D737ZnbG9tHXtns1NeZvtJ28/bvqp1/HXNDvdP249Ien/7erbPsn2v7UO2/2p7Q+vYtO2bbN/ZvLq4z/Yptm9o1ttje81CnpftrbZvs73d9gvNtd7XOr7P9mbbjzRrb7O9pDl2ue3dA+uV7ZW2PyvpYkmbmhnvWNjvNI4i8Amyfaqk8yXtbT38kqRLJb1R0gWSPm/7woFTPyzpTEnrJG2xfVbz+FclndH8+oSky1rXOlHSHZLulvRWSVdKusX2ma11N0r6iqQ3S3pZ0gOSft/c/5mkb4/w9DZI+knzPH4u6caB4xc3M54h6V3Ndf+nqvqepFskfb2qllXV+hHmgQh8Um63/YKkGUnPqhemJKmq7q2qP1fVkar6k6QfS/rowPlXV9VsVT0s6WFJ720e3yjpuqo6WFUzkr7TOudcScskXV9Vr1TVryT9QtJFra/ZWVW/q6o5STslzVXV9qqal3Srem8nFmp3Vf2yOffm1oxH3VhVM1V1UNJ1A3NgTAh8Mi6sqpMkrZX0bvV2SEmS7XNs32P7Odv/knRF+3jjQOv2YfXClaTl6n3TOOrvrdvLJc1U1ZGB429r3X+mdXv2v9wf5cPAwRmX2J5qPTY45/IR1sYiEfgEVdWvJU1L+mbr4R3qvaRdUVVvkPRdSV7gkv+QtKJ1/+2t2/slrbB9wsDxp0cc+/9lcM79ze2XJL3+6AHbpwycx487HgcCn7wbJJ1n++hL2JMkHayqOdsfkPSpEda6TdJm2yc37++vbB17UL2ddJPtE22vlbRevffJXfii7VNtv0nSVeq9BZB6bzneY3t188Hb1oHznpF0+uTGzELgE1ZVz0naLmlL89AXJF3TvEffol60C3W1ei93n1Dvw7SbW9d5Rb2gz5f0vKSbJF1aVXuO9zks0g71Znxc0mOSrm3mfFTSNZJ2SfqbpN0D5/1A0qrmbwJun9y4Gcx/+IBxs71P0meqalfXs7zWsIMDwQgcCMZLdCAYOzgQjMCBYAQOBCNwIBiBA8EIHAhG4EAwAgeCTR37S0Z37kXfelX965mlT891PcKQxz+30J8InYzH1m3reoQ+q+6/pOsRhpy26cWuR+hz595vHPMPETs4EIzAgWAEDgQjcCAYgQPBCBwIRuBAMAIHghE4EIzAgWAEDgQjcCAYgQPBCBwIRuBAMAIHghE4EIzAgWAEDgQjcCAYgQPBCBwIRuBAMAIHghE4EIzAgWAEDgQjcCAYgQPBCBwIRuBAMAIHghE4EIzAgWAEDgSbGseiz254eRzLLtr87Fie5nF5Yt33ux6hz5rffrLrEfq845K9XY8wZM/1q7seYWTs4EAwAgeCETgQjMCBYAQOBCNwIBiBA8EIHAhG4EAwAgeCETgQjMCBYAQOBCNwIBiBA8EIHAhG4EAwAgeCETgQjMCBYAQOBCNwIBiBA8EIHAhG4EAwAgeCETgQjMCBYAQOBCNwIBiBA8EIHAhG4EAwAgeCETgQjMCBYFPjWPRLa+4ex7KLtnPVW7oeYcjG3eu6HqHP4b+c3PUIfT700B+7HmHIngfnux5hZOzgQDACB4IROBCMwIFgBA4EI3AgGIEDwQgcCEbgQDACB4IROBCMwIFgBA4EI3AgGIEDwQgcCEbgQDACB4IROBCMwIFgBA4EI3AgGIEDwQgcCEbgQDACB4IROBCMwIFgBA4EI3AgGIEDwQgcCEbgQDACB4IROBBsahyL/nT/2eNYdtF+9OSOrkcY8rEdX+56hD7LH5jveoQ+93/w9K5HGHLarUe6HqHfp4/9JezgQDACB4IROBCMwIFgBA4EI3AgGIEDwQgcCEbgQDACB4IROBCMwIFgBA4EI3AgGIEDwQgcCEbgQDACB4IROBCMwIFgBA4EI3AgGIEDwQgcCEbgQDACB4IROBCMwIFgBA4EI3AgGIEDwQgcCEbgQDACB4IROBBsahyLHppdMo5lF+3jD13R9QhD5pce6XqEPud97Tddj9Dnh/es7XqEISsPz3Y9wsjYwYFgBA4EI3AgGIEDwQgcCEbgQDACB4IROBCMwIFgBA4EI3AgGIEDwQgcCEbgQDACB4IROBCMwIFgBA4EI3AgGIEDwQgcCEbgQDACB4IROBCMwIFgBA4EI3AgGIEDwQgcCEbgQDACB4IROBCMwIFgBA4EI3AgGIEDwabGseiu1dPjWHbR5upI1yMM2Xbo7K5H6HPO0r1dj9Bn110f6XqEIVOPPtX1CCNjBweCETgQjMCBYAQOBCNwIBiBA8EIHAhG4EAwAgeCETgQjMCBYAQOBCNwIBiBA8EIHAhG4EAwAgeCETgQjMCBYAQOBCNwIBiBA8EIHAhG4EAwAgeCETgQjMCBYAQOBCNwIBiBA8EIHAhG4EAwAgeCETgQjMCBYK6qrmcAMCbs4EAwAgeCETgQjMCBYAQOBCNwIBiBA8EIHAhG4EAwAgeCETgQjMCBYAQOBCNwIBiBA8EIHAhG4EAwAgeCETgQjMCBYAQOBCNwIBiBA8H+A/PSUmbShqMwAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAPgAAAEPCAYAAABvMQkeAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvIxREBQAAClZJREFUeJzt2n+snfMBx/HPpy5F2ig2tioaP4Z2wZb9iKwJs9LWGFvGmIjJhklWW0ZILaiF1ULC/LHZzHSUCBOiSTc/ltWMNBJmwZYlpbR+tJhf9WsJvvvj+Z55zv3RW9W7p/ns/Upueu75Pud7v+c8532e5z63LqUIQKZxXS8AwNghcCAYgQPBCBwIRuBAMAIHghE4NjrbB9l+uut1gMD/y/aTtt+y/brt1bYX2p7Q9bqGY3u+7UVjOP9C2xeO4fzF9h5jNT/eR+D9jiilTJC0v6RPSZrX8Xo2iBvsWxD4cEopqyXdoSZ0SZLt8bYvtb3S9hrbV9reqjV+pO2Hbb9m+3Hbs+v9k23fbvsl28ttn9x6zHzbN9m+1vZa24/Z/kxr/Gzbz9Sxf9r+Up33HEnfqGcbf6vbLrV9ke37JL0pabd6VjJz0M9b1Pp+hu37bb9ie5Xtb9k+RdLxks6q8y9uPY9bbL9ge4Xt01vzbFWP+i/b/rukz67va13XdLPtRfV5PmL7E7bn2X6+ruvQ1vYn2f5H3fYJ26cOmu8s28/Zftb2d9pnC6Ptw0ilFL6a/677pKSZ9fYUSY9I+llr/DJJt0vaTtJESYslLahjn5P0qqRD1Hxo7iRp7zr2Z0k/l7Slmg+MFyQdXMfmS3pb0mGSNpO0QNKyOraXpFWSJtfvp0ravfW4RYPWv1TSSknTJQ1I2rz9nAY/TtKuktZKOq5uu72k/evYQkkXth43TtKDks6TtIWk3SQ9IWlWHb9Y0r31tdlZ0qOSnl7Ha10k7THoNZhV132tpBWSflTXdbKkFa3HflnS7pIs6UA1H2afrmOzJa2ur8HWkhYN+lkj7sPUr84XsKl81Rher2/6IumPkibVMUt6oxdYve+A3htP0i8lXTbMnDtLelfSxNZ9CyQtrLfnS7q7NTZN0lv19h6Snpc0U9Lmg+YdKfAfD/OcRgp8nqRbR3gtBgf+eUkrB20zT9I19fYTkma3xk75gIHf1Ro7ou6Hzer3E+v2k0aY6zZJ36+3f9MOtr6Gpf67zn2Y+sUper+jSikTJR0kaW9JH6n3f1TNEeHBejr7iqQ/1PulJuTHh5lvsqSXSilrW/c9peYI37O6dftNSVvaHiilLJf0AzUBPG/7RtuTR1n/qlHG20Za83B2lTS599zr8z9H0o51fPKgn/3UB1iHJK1p3X5L0oullHdb30vSBEmyPcf2svorzytqzn56+2nwOtq3R9uHkQh8GKWUe9QcxS6td72o5o02vZQyqX5tU5oLclLzRtp9mKmelbSd7Ymt+3aR9Mx6ruOGUsoMNYEVST/tDY30kEHfv6HmTd3zsdbtkdY83Dyr1BzpJrW+JpZSDqvjz6n5wOjZZYR5PxTb4yXdoma/7FhKmSRpiZqjc28dU1oPaa9ptH0YicBHdrmkQ2zvV0p5T9JVki6zvYMk2d7J9qy67dWSTqoXwcbVsb1LKask3S9pge0tbe8r6dtqfjdcJ9t72T64vqnfVvPmfK8Or5E0dT2ulD8s6Vjbm9eLd19vjV0vaabtY2wP2N7edu+i4ho1v2f3PCBpbb3ot5XtzWx/0nbvYtpNkubZ3tb2FElzR3t+G2gLSePVXMd4x/YcSYe2xm9Ssx/2sb21pHN7A+uxDyMR+AhKKS+oueBzXr3rbEnLJS2z/Zqku9VcCFMp5QFJJ6m5iPOqpHvUHHWl5iLWVDVH81slnV9KuXs9ljBezcWrF9Wcxu+g9/9sd3P991+2H1rHHOeqOUq/LOkCSTe0nt9KNae3Z0h6Sc2HwX51+GpJ0+qp7G31dPlwNRcJV9Q1/VrSNnX7C9Sclq+QdKek69bj+X1g9Ved09WE/LKkb6q5aNYb/72kKyT9SXVf1aF/139H3IepXC82AHFs76Pmiv74Uso7Xa+nCxzBEcX2V+vfu7dVc81i8f9r3BKBI8+pav68+LiaP1Ge1u1yusUpOhCMIzgQjMCBYAQOBCNwIBiBA8EIHAhG4EAwAgeCDYzFpN998IRN6n/PPHrRvl0vYYgZ5y8bfaP/ob+eOK3rJfQ58sZ7u17CEIvOPbzrJfS5/+YzPdo2HMGBYAQOBCNwIBiBA8EIHAhG4EAwAgeCETgQjMCBYAQOBCNwIBiBA8EIHAhG4EAwAgeCETgQjMCBYAQOBCNwIBiBA8EIHAhG4EAwAgeCETgQjMCBYAQOBCNwIBiBA8EIHAhG4EAwAgeCETgQjMCBYAQOBCNwINjAWEz62AX7jsW0G2zJlVd0vYQhDj9lbtdL6PPF6+7regl9bj9wWtdLGOK4e5Z0vYRBzhx1C47gQDACB4IROBCMwIFgBA4EI3AgGIEDwQgcCEbgQDACB4IROBCMwIFgBA4EI3AgGIEDwQgcCEbgQDACB4IROBCMwIFgBA4EI3AgGIEDwQgcCEbgQDACB4IROBCMwIFgBA4EI3AgGIEDwQgcCEbgQDACB4IROBCMwIFgA2Mx6cfnLR+LaTfYMbNO7HoJQyy966qul9Bnzp5f6HoJfU57+L6ulzDEL44+qusl9Jn70OjbcAQHghE4EIzAgWAEDgQjcCAYgQPBCBwIRuBAMAIHghE4EIzAgWAEDgQjcCAYgQPBCBwIRuBAMAIHghE4EIzAgWAEDgQjcCAYgQPBCBwIRuBAMAIHghE4EIzAgWAEDgQjcCAYgQPBCBwIRuBAMAIHghE4EIzAgWAupWz0SQ8Zd/TGn/RDePqW6V0vYYidL960PlvfvPD1rpfQZ8LcTev1kaTjFy/tegl9TthzmUfbZtN7FQFsNAQOBCNwIBiBA8EIHAhG4EAwAgeCETgQjMCBYAQOBCNwIBiBA8EIHAhG4EAwAgeCETgQjMCBYAQOBCNwIBiBA8EIHAhG4EAwAgeCETgQjMCBYAQOBCNwIBiBA8EIHAhG4EAwAgeCETgQjMCBYAQOBCNwINjAWEz6kxUPjMW0G+y8A3bseglDzP3L0q6X0OfyY4/uegl95t9xTddLGOKMM77X9RL6nPC70bfhCA4EI3AgGIEDwQgcCEbgQDACB4IROBCMwIFgBA4EI3AgGIEDwQgcCEbgQDACB4IROBCMwIFgBA4EI3AgGIEDwQgcCEbgQDACB4IROBCMwIFgBA4EI3AgGIEDwQgcCEbgQDACB4IROBCMwIFgBA4EI3AgGIEDwQgcCOZSykafdMbXLtn4k34IP7zkhq6XMMSvvjKn6yX0uf7O33a9hD7HT5/d9RKGuPLRJV0voc/UKc95tG04ggPBCBwIRuBAMAIHghE4EIzAgWAEDgQjcCAYgQPBCBwIRuBAMAIHghE4EIzAgWAEDgQjcCAYgQPBCBwIRuBAMAIHghE4EIzAgWAEDgQjcCAYgQPBCBwIRuBAMAIHghE4EIzAgWAEDgQjcCAYgQPBCBwIRuBAMJdSul4DgDHCERwIRuBAMAIHghE4EIzAgWAEDgQjcCAYgQPBCBwIRuBAMAIHghE4EIzAgWAEDgQjcCAYgQPBCBwIRuBAMAIHghE4EIzAgWAEDgQjcCDYfwAAr7dC1hQAIAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "for i in range(3):\n",
    "    random_data = np.random.rand(dataset.shape[0],dataset.shape[1])\n",
    "\n",
    "    plt.figure()\n",
    "    plt.axis('off')\n",
    "    plt.title(\"Random Input\", y=1.03)\n",
    "    plt.imshow(random_data.reshape(bm.image_height, -1))\n",
    "    samples = bm.reconstruct(random_data)\n",
    "    sample = samples[0]\n",
    "    plt.figure()\n",
    "    plt.axis('off')\n",
    "    plt.title(\"Reconstructed Image\", y=1.03)\n",
    "    plt.imshow(sample.reshape(bm.image_height, -1))\n",
    "    sample = sample > 0.5\n",
    "    plt.figure()\n",
    "    plt.axis('off')\n",
    "    plt.title(\"Binary Reconstructed Image\", y=1.03)\n",
    "    plt.imshow(sample.reshape(bm.image_height, -1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.15"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}