Intro+to+Machine+Learning+-+Circuit+Component+Identification.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Machine Learning Project- Circuit Component Identification\n",
    "\n",
    "Programmers: Tanzila Rahman (tr1106) and Nnamdi Nwaokorie (npn225)                                               \n",
    "Class: EE-UY 4563 - Intro to Machine Learning                                                                     \n",
    "Professor: Pr. Sundeep Rangan                                                            \n",
    "Date: 22 December 2017                                                               "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Problem Formulation\n",
    "\n",
    "Problem: Is it possible to take a picture of different circuit components and have a neural network \n",
    "         identify the difference between them?\n",
    "\n",
    "Goal:  Our goal is to test a normal convolutional neural network on 8000 handwritten circuit component samples, 1600 for each of \n",
    "        the five circuit components used!  The five circuit components are (1) a resistor, (2) an independent volt source, (3) a \n",
    "        dependent volt source, (4) an independent current sources, and (5) a dependent current source!"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Review of Past Techniques\n",
    "\n",
    "One team has attempted this same problem but instead of using a convolutional neural network, they implemented an artificial neural networks.  In addition, instead of identifying specific circuit components, they sought to indentify entire circuit composed of various components.  \n",
    "The team's work can be seen here: http://www.sciencedirect.com/science/article/pii/S1877050916300783"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Our Solution\n",
    "\n",
    "Below, our solution to this unique problem is presented!  \n",
    "\n",
    "The steps to our solution are as follows:\n",
    "\n",
    "(1) Create and test the images NOTE: Images can be found in the \"Electronics\" folder of this repository!                         \n",
    "(2) Resize the images and Threshold them to black and white                                       \n",
    "(3) Capture 8000 random images and keep track of the 8000 labels                                            \n",
    "(4) Rescale the general data and create the training and testing datasets                                        \n",
    "(5) Create the neural network and its model                                                                \n",
    "(6) Test the neural network on the data                                                                                \n",
    "(7) Create a Support Vector Machine (SVM) and test it on the data                                               \n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Step 1: Load and Test the Images"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "import keras\n",
    "\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "from sklearn import linear_model, preprocessing\n",
    "\n",
    "import librosa\n",
    "import librosa.display\n",
    "import librosa.feature"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Create a function plt_circuit that plots the circuit components.\n",
    "\n",
    "def plt_circuit(img):\n",
    "    nrow = 28\n",
    "    ncol = 28\n",
    "    img_sq = img.reshape((nrow,ncol))\n",
    "    plt.imshow(img_sq,  cmap='Greys_r')\n",
    "    plt.xticks([])\n",
    "    plt.yticks([])\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Define our exception as follows which has an optional string argument.\n",
    "\n",
    "class ImgException(Exception):\n",
    "    def __init__(self, msg='No msg'):\n",
    "        self.msg = msg"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import matplotlib.image as mpimg\n",
    "import skimage.io\n",
    "from skimage.filters import threshold_otsu\n",
    "from skimage.segmentation import clear_border\n",
    "from skimage.measure import label, regionprops\n",
    "from skimage.morphology import closing, square\n",
    "from skimage.color import label2rgb\n",
    "from skimage.transform import resize\n",
    "import matplotlib.patches as mpatches\n",
    "from skimage import data\n",
    "import skimage"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import os.path\n",
    "\n",
    "# Create a function which can find and load the required image\n",
    "\n",
    "def load_img(circ_ind, samp_ind):\n",
    "    \"\"\"\n",
    "    Returns the image from the dataset given a circuit component and sample index.\n",
    "    \n",
    "        \n",
    "    If the file doesn't exist, it raises an Exception with the filename.   \n",
    "    \"\"\" \n",
    "    \n",
    "    # TODO:  Set the file name based on circ_ind and samp_ind\n",
    "    # fname = ...\n",
    "    fname = \"C:\\\\Users\\\\Nnamdi\\\\Documents\\\\Electronics\\\\Img\\\\Sample{0:03d}\\\\img{0:03d}- ({1:d}).JPEG\".format(circ_ind, samp_ind)\n",
    "    \n",
    "    # TODO:  Use the os.path.isfile command to check if the file exists.  \n",
    "    # If not raise an ImgException with the message \"[fname] not found\"\n",
    "    if os.path.isfile(fname) == False:\n",
    "        raise ImgException(\"%s not found\" % fname)\n",
    "\n",
    "    # TODO:  Use the skimage.io.imread() command to read the png file and return the image.\n",
    "    # img = ...\n",
    "    img = skimage.io.imread(fname)\n",
    "    return img"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Circ = 5 samp = 47\n",
      "C:\\Users\\Nnamdi\\Documents\\Electronics\\Img\\Sample005\\img005- (1670).JPEG not found\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXoAAADRCAYAAADCDHkJAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x10ddef28>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Test the load_img function and make sure it works properly\n",
    "\n",
    "circ_ind = 5\n",
    "samp_inds = [47,1670]\n",
    "for samp_ind in samp_inds:\n",
    "    try:\n",
    "        img = load_img(circ_ind=circ_ind, samp_ind=samp_ind)\n",
    "        print(\"Circ = %d samp = %d\" % (circ_ind, samp_ind))\n",
    "        plt.imshow(img)\n",
    "    except ImgException as e:\n",
    "        print(e.msg)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Step 2: Resize and Threshold the Images"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Complete the following method to find the image and place it in a 28 x 28 box.\n",
    "\n",
    "def mnist_resize(img):\n",
    "    \"\"\"\n",
    "    Extracts a circuit component from the image, and places in a 28x28 image to match the MNIST format.\n",
    "    \n",
    "    Returns:\n",
    "    img1:  MNIST formatted 28 x 28 size image with the circuit component from img\n",
    "    box:   A bounding box indicating the locations where the circuit component was found in img.    \n",
    "    \"\"\"\n",
    "    # Image sizes (fixed for now).  To match the MNIST data, the image \n",
    "    # will be first resized to 20 x 20.  Then, the image will be placed in center of 28 x 28 box\n",
    "    # offet by 4 on each side.\n",
    "    nx_img = 20   \n",
    "    ny_img = 20\n",
    "    nx_box = 28   \n",
    "    ny_box = 28\n",
    "    offx = 4\n",
    "    offy = 4\n",
    "    \n",
    "    # TODO:  Convert the image to gray scale using the skimage.color.rgb2gray method.\n",
    "    # bw = ...\n",
    "    bw = skimage.color.rgb2gray(img)\n",
    "    \n",
    "    # Threshold the image using OTSU threshold\n",
    "    thresh = threshold_otsu(bw)\n",
    "    bw = closing(bw < thresh, square(3)).astype(int)\n",
    "    \n",
    "    # Get the regions in the image.\n",
    "    # This creates a list of regions in the image where the circuit component possibly is.\n",
    "    regions = regionprops(bw)\n",
    "\n",
    "    # TODO:  Find region with the largest area.  You can get the region area from region.area.\n",
    "    # region_max = ...\n",
    "    region_max = regions[0];\n",
    "    area_max = 0\n",
    "    for region in regions:\n",
    "        if region.area > area_max:\n",
    "            region_max = region\n",
    "    area_max = region_max.area\n",
    "    \n",
    "    # Raise an ImgException if no region with area >= 100 was found\n",
    "    if (area_max < 100):\n",
    "        raise ImgException(\"No image found\")    \n",
    "                \n",
    "    # Get the bounding box of the circuit component from region_max.bbox\n",
    "    minr, minc, maxr, maxc = region_max.bbox\n",
    "    box = [minr, minc, maxr, maxc]\n",
    "    \n",
    "    # TODO:  Crop the image in bw to the bounding box\n",
    "    # bw_crop = bw[...]\n",
    "    bw_crop = bw[minr:maxr, minc:maxc]\n",
    "    \n",
    "    # TODO:  Resize the cropped image to a 20x20 using the resize command.\n",
    "    # You will need to use the mode = 'constant' option\n",
    "    # bw_resize = ...\n",
    "    bw_resize = skimage.transform.resize(bw_crop, (nx_img, ny_img), mode = 'constant')\n",
    "    \n",
    "    # TODO:  Threshold back to a 0-1 image by comparing the pixels to their mean value\n",
    "    mean_value = skimage.filters.threshold_mean(bw_resize)\n",
    "    bw_resize2 = bw_resize; \n",
    "    \n",
    "    # Loop through the rows and columns of the bw_resize matrix\n",
    "    for i in range(np.shape(bw_resize)[0]):\n",
    "        for j in range(np.shape(bw_resize)[1]):\n",
    "            # Set all values in the matrix less than the mean to zero\n",
    "            if bw_resize[i, j] < mean_value:\n",
    "                bw_resize2[i, j] = 0.0\n",
    "            # Else, set those values to one\n",
    "            else:\n",
    "                bw_resize2[i, j] = 1.0\n",
    "\n",
    "    # TODO:  Place extracted 20 x 20 image in larger image 28 x 28\n",
    "    # img1 = ...\n",
    "    img1 = np.zeros((nx_box, ny_box))\n",
    "    img1[offx:-offx, offy:-offy] = bw_resize2\n",
    "    \n",
    "    return img1, box"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXoAAADHCAYAAAAXg5iPAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzsvX3MtVtaH/S71rrvvZ/3vGeGYfiYTAcq04SalCZiJPhH\n/cCQKiYqNhoCJgWBcNqkxRhJLFAjjVhTFSQmVeJpIEBSoRjENg1apyRKTESZNk2kRRQoE2ecMrRg\npnPe99n7vte6/ONa11rXWve69/O85z1nZp+z13Xmmb33/bG+7vf+rWv9ro9FzIwhQ4YMGfLuFfe5\nbsCQIUOGDHl7ZQD9kCFDhrzLZQD9kCFDhrzLZQD9kCFDhrzLZQD9kCFDhrzLZQD9kCFDhrzL5W0D\neiL6OiL6VSL6NSL67rerniFDhgwZclno7fCjJyIP4P8C8IcBfBzALwH4Jmb+O295ZUOGDBky5KK8\nXRr9VwP4NWb+DWY+A/gpAF//NtU1ZMiQIUMuyPQ2lfshAP+P+f1xAP/k3sXvefqEv+B9r+bfRAQw\nAEq/9QtY/1euBcDp/9vjAOUyeisXtl+oPkdE6T65gIhQF8H1tbldUhiZJhPVhffbwunaqmNNXXrQ\nXlS3XYsmMu2q6uPqXjtqOs6Ux8xWy80wUTnOsEPdSNWRVD7l/jI41avXuN2ytK0EKn3KjdX/67fi\nY//vp/4+M39R9+SQIe9yebuA/kEhotcAvAYAX/C+V/F9f+JfA7MAKsHn19U5l0A2veSRK+AiInCc\nwBwRDKA4l85RAboIe69DYPnNDDDZex2IymJH25XbkL7HGEFE8N4D7BBCyPd7J/d7kJk0GESc740G\nk2KM4HCGcy7/EREQg5xLbZA+pPoh15DXtpY2eu/hycm9a8ht1XK891X92q+2vXJC2hxCkPqoXBNj\nzP323pe+MG+endQhx/U6bZeUF0E05+9WNs88PTvn5J8wR5jn4/N1Kt/2p3/oYxgy5Ebl7QL6TwD4\nUvP7S9KxLMz8OoDXAeDDX/LF3AJCZIZrjgFAJICyMicAER0hRjSKo2sPGJDy2ogEuoyo+rgCP8cK\nKBRYWsCvQZzy8fxpmkBUt0evseXVbWWQc7nuGGMCdUIIAWuM8N6XVQAKkMc0ICEEIMS2WqmzM76M\nGlT1O6OAsu375v48eZZJNU/gnf4r6Eu59Xi3Zdox2gA/6vPtWA4ZcsvydgH9LwH4ciL6MATgvxHA\nv/HQTarNq1jQsi9rTKYFSv/JvQ7MoSovKj2QtEtQAZSiFaYrOVbAtCc9QGdWumYLKFY77YGUFQW9\nAmzSftWamRmTd03bsVtmOyFZYM4TXCyTgIx1v20xRjBiVaZtrz4nnZD0njImEUTN6mwDyj5NslzX\nu9O3PTAfID9kSC1vC9Az80pEfxLAXwPgAfwoM//th28sYMyJE47y5pfz+3WiaIQJKMmhpQCgdAcR\nPBMCKV1Ug+AlgGjPM4uVoKdN2vo3WiqUYmbh1E0bYlxlHIKA/LquhnqRScFhO3kU7bmsJlpQznXw\nFugtNWbLLuWLdt8bHwX5Qqvo8yogb+WhMW7FrqAuXdN+H2A/5NblbePomfnnAPzc429w2YBotU/5\nEyBW2LhEF2RxaitMhlqSg/ka1slkq8VbQOlRMy0Qiuba0hkxgeJWg2XEzD8xwwC4lh+yZsuJx1Zp\n6RNmXbUAMYYK8F22RNcgnMuieoJwzgGhXFOAurZTMANMsWqDHYtybUA2+nbooD36x46VFcv5b1df\nNEB+yJAd+ZwZYyvhWhO1APCt/94Pf65bN+QdJj/2H/07n+smZKGeYWLIkLdQWLTZi3IVKRDsm1Br\nYFfRvCHvUBna/JAhIleh0SvtAcC423k4V6aAH/kP/2St7UdkSoMa7p5Z3ADhxEXHLuNj4GzY9H7O\ndMAK7tIbpZHF6yS7PqJQHhwTBUHJE4aRPV7UU1PLdw7ZvdG6JvrkJcMIuY64htxegs/uleu6Ygmp\nzcl/XMdDOX3E5OIYuUv5BONumvsdoriHeg/vPZip1J/ul7YVbxzvPeZ5xuEw5fqdczgcDjI2XFw6\nay8b+29AV3FUla/jZq+zdgB1NSU4fMf3/RfVdQCwrmtFfQ0ZcotyFUCvKj0nDzvhzQt/DhjuHi4H\nxyiIoPVmIS8GQwAEZ4ycTWAOLOCoO2ZtnDQ/MkhlEDbgKXx/MF5AlPzha59z54pNIXKpL4Ti1WI9\nV8jw60SMyFRdE0LILQ5B/OWXZcGyLEBkeO8xUXHRrCY92nLnlCcjJy6rxm2R4AHS2hxiXCtPnBhX\nxLhmf/vzecY0TSAiTNME5kP+beWS5t3aSFqDrNoASlBdua8yTA8ZcsNyFUDPBMDZV9VVAASoT3oJ\nmZSX1yESw0UHRnGtVM+UTT2NZpi/w0MDmXqAn69xxQCpQJ41YeIyYRmPGoI17BbNO4QFnCYBBWzi\nkCevvOpIywHV/BkuXc9Y1oh1XRGMVq5Av65rniR0SdEaQduAqXRRLgvGu4ag2riMtdalqxj1Egph\nKa6g04R5nvOnGqmJ/GZlZKtvn19rWNUxsho9OjRlf2IYMuT25CqAXqX2tGn8pOEQnQb0FDdKQEGJ\nENmCPUAUqxe9AMWUPl126dTzpU6PjXa/AxzMUn9pC1CvHpDOhQTiSwLJFKEaVyzLGZTAkZmxLAti\nDAnohaqRCUQnAsZpEYommrYrlTTPs6xliKAM2KbdTfsAgFyJPZC+MTgSyJXnA3ZgXo33CyPGkPuo\nQG+fqZYlx1Y4N1VRwMX9s3aV7Wn/9b+NQuVt+jKAfsgQAFcC9NYHvY5oNe59jgCrwcFo6Okl1wnA\navcK9oUb9nBO3Q4fMvZu/fDtJFSBofkeQgBI3EUZQmnY9lqAZC6cOmeAZKyraMYhrQDWNVRAz0xY\nVtHeucN/e+83QC+cO1f1b54Fy8qp0rYbwNxzb5WgqlXsIiiukNL+wpWLDYAbW4BScVpPH5xbN9rS\nxv4THEA/ZMiVAP2e9Hyvq9/sJJdN4o3zFSSBPyIOGvUqhr7aH95OMOW62u+91EdghOZ8SaNgy+QU\nJGT5dim75K7hKJPCsiRQX85ZM1a+HZEz0ANAzEDoEFNwEyXAVBqoAKpPn0Vr3gN67asnAKYOAKLR\nE8H7WdoQV0QGQqD0ndMKivIniBOf7/Pk1wK9rl5qcdUz6E2mbZtbaW0nQ4bculwJ0NsIzHSko9VV\n7qLkM/VgA3tEPNiFXKbLqwATCcpl5VAH7TgANpWCaPXZ4MgxGxtF60Wup/TBaPwxYl0XAALeISwJ\nnBkRRZuPMSIkI2oFTgnodSKJXOp1zmGaJvh5xuFwqLh95xw8CbjPzueJwGrf1DF2T05oEPV84qhG\nZ59XDcweIUwIfhGwZwfmRCnFJngtgb2Ozfl8TnTVNvDMPotLQG8ToVkKp/w72U4KQ4bcslwJ0Iso\n0Pc0NcvbgqjysnA9zc54lFB0VaZEKTDdu5kk2jrNxUClMVceNx1AIiJw0tgB4Hw+Y13PiDHCT5SB\nPkaJlI0h5N+Zt84pCeo+qmHUOYdpnnE8HqtMks45TE7AfaJao8/G355G78QtVTV6zhOaz1SLukC6\nKFlCIzsAUtcaSm4eO076dz6f0zGfx1L7W+isLVDblUh9jpuJejvZDtAfcutyFUDPLBzuNE0gJ/z0\nutaarXfn7OLHTIgxIHIU7ROJ56VCkWRPlPQfHFIqYkJAhEvJwSIz1mg5ffmLVBJ0qYYsWvwk6BaA\nsCr/7+EjsrdLjCvWdcUaluRPLsN8Op2wLIv0Mzjc399jWZZMuSi4EknaY4IH+ZobnxL/Pk1Txblj\nPcMTwTtU8QEydlPmwm2un7Cy+Nsncc4h8poBXetU6oacTmYey7IiRIkR8PCIUTxuAOSMmsuyIvIK\nYAWRA3jCcT4ggsDRgeOE09nhvDDmI+Hu7oDJO5xDxDll3JxmYPaTeOpoTqDGPhJaIKeAEENe/Xj/\nkC1myJB3t1wH0OunuFy8uTK4+N3HGBGTpgdqqZkXi5hsPXayVtlQFGRohdPpJABKnKgZ0WLv7+8l\n0CmBvQ3AUtC2rpvWGyWPD5Cvt2Csx/XTucl4qNh+MDTBmPMMzyVbqPcekVFNEjFGwJUySjs5TQzB\nrJY4LzxiJABRwFapuSjWaedczlykfVa3UKV5rGE5j3HzPC7ZcPTYCJYaMuRKgB4QT5nIDJe4854L\nI5Hv0hhZ86aSHjcD/SOVuQ1QkPLvoTovqwnkZGP5vuQVsywL7u/vhY7wlI6dSruSB0oIAfM8Z+58\nnueK6lAw7wG9RqIq0Os91s9cqZ1SpqWpxB2UiOCncr/3wr9bkG/LnCa5YZrke4hrpmokFfQ53Rvg\nPYExAcnovCbMVaDXyTnGCASdMGQCOxwO1eOwk+Lm2cWtZ9Tucx0y5AbleoA+868KTFv3v6LxpwyK\nKQ0CJS2x5xee73ug7vY6/copXLfkWE88+loALoSAcDpnoFeDqkupCs7newDA4XCA9z4bc+d5zvW2\n7pFWa+/5kit1o79bg6aeC4GzEZsqjNTUyvJLIngZlF1PlVeXzJzaFik35gmBYm0cRd5BixLQB8RV\nUygUGwxBct+HKBNEXPX5+2xY1rGJLPsFPM6esj02tPohty5XAfTq/S7ujeLlstXoE/CnySBGZPdC\nReU6wAqJg69z1+wF0Txk/FMjp1IMnHLQKLCv96eKwiAiLMuaXSf13mmaMkirdtxqq6o9K7AqlWKD\nhVpaxxoeW+2+AN02N74GjyHRLhrhKnmeGeQARx7eC3ADsRoXW/80TWAE45XkASKEFCvgPRBzgJN6\nE2l7koeUk/w0ajtwXqckBieKp13lPOR2Ofzoh9y6DCvVkCFDhrzL5So0ehXRzPZ3L+opZszcDYps\ng516ZWqAz7YNZfcl1TpFw4xZi9cNt5dlwel0QjwvuR7vU/6aWHzagZIhUvO/WMNrL7JVtf5Wowfq\nxGp70q5eWpdDOV5SNjCXzJPOCZ3DZkMYLcNmj9RjuT1c3CSzcTbZFBwRYiAwOQQz7ix8Tm5bMC6p\n5Di5iQLxkasxmzGz/bcwZMgtypsGeiL6UgA/AeADEDL9dWb+z4nozwD4DgC/nS79Xpbdpi4UpqAb\noF4XknNlS6coJaNukM45CdtvrosKaju8/UOiRtd1LTx8lRkyCLWghlUCBMycy26GmszLUibzPOPu\n7i4FDclxTee7x9HvAb1KMbiqXUNz7yTenSxXrWPVGrV1AgimDcX4LVkqa790zXCp6aS990CsgZ6c\neN0458CTRwyEJURwjInKKeVJX+X5KXVDjkET4L2rKC47Ye0BvfZjAP2QW5eX0ehXAN/FzH+TiN4D\n4G8Q0UfSuR9i5h94kcLkZTSuc815q9HLy665bxiahtFqmBnge7FUnZff8t8StRlrY6sB+vP5DIpc\npRyevMfxeAQR4f5eEn4dDocK6Jkl7P/JkycgIjx//jxr79Z3HajTFligrwyfpi9WW+/1twVERgCR\ny8ZXotp7xa4GNIo1Ipj0CGVCYgPAml0z55SnCNI+YBb/+fOCsK6yakgafDEgl3EHABeA6Ep+ona/\ngL3+tuMzZMgty5sGemb+JIBPpu//kIh+BcCHXqK8pEXbXC1Fa5U0t0AMmilRsjrGGOFpqiYHZvHm\nEB/w1v9brsngmNLvhhDw/Pnz5AEzIxIQg/wtZ3WdXHE+rVjDmoGeiHA8HjGb6NMnT56kxGacvGsO\nuc5pmnA8HnF3d4f3vOc9OJ9TpGxqn2r3mhbBTl4ZTA1o274hbQJSj2mhWixV5ExKBL1WyytAWzJl\n6vNwVLyArJ96NlgTUg4bSV3MQPH3j7IM02uZy4RGKYAthBrINd1yjB4uTZxi2C2pjn0D5jpxaH+G\nDLl1eUs4eiL6MgD/OID/DcAfAvCdRPTNAD4K0fp/t3PPawBeA4AveN+rlceJanVbza3NidIEMnVo\nmkrLV8+dfh8yFUEkO1GJa+QZ5/M556SRdMEER5I0zKfI1dmXtLtaBoirwKZ5niuwthOa5ZXtNa12\n+hhX0UteJ7m8vMtL9iMFmEFUtPo21YOCfEub2HGXeAcy/ZAsmnJdmZgEsBkxTSZK4WzomBSlLJG3\nZectHbdL42A/hwy5ZXlpoCeiVwH8DIB/m5k/TUQ/DOD7IYj6/QB+EMC3tfcx8+sAXgeAD3/JFzNl\nJ++UY4VLgrN0farPwVI8Alr1NVY7Vf4//0b98rfgoi6b5yXgdDrhdBJja8mxHlPgkE+ukgI8x6mk\nHYjRZY1eXCmlXg0CUq3ZaunTNBlD6JZjbo3LfSk53cunuCbalYxGxsrvXCoAblIZS5AaUNIobEG+\nnpCj8Xd3zoEcw8GmeFCgJ3gPBN2qMIZqdZD77yhp6MCaGmu1+d6Y9Hj7IUNuWV4K6IlohoD8X2Tm\n/xYAmPm3zPm/AOCvPqKkvOcqY80gYyXGFIaPwufXmuTWIAcg5bex2l+dygBc78y0LALo9+cVp9OS\nja/FE2aG94TD5JL3jIDmwWj0zJLegKHJyUqbbDstP6/nHgL67mTWkbYe+70urxhnxbiNNKmVjJ+O\nShv3pD8RJYO6aTtBUziInYVSPhoESnRRiW62os/I7rHbri5s/UOjf2fKNTyvdyPd9zJeNwTgRwD8\nCjP/Z+b4BxN/DwB/BMAvP1QWM5LBD1A3v/ZxS1pd3TBco2MloIfXmkIQTTIZElHz/TEWt0lmyRip\nXh7Pnz/H+bwiRuD+fE5RpdKm4/EuafBC18zeY54LdXNw3tAKPmu7MiZ1hKalZ/S3/dT2tmDdgt+l\nlUp9rNRTjN5C3xh4ztc4R+AYoTtnOa+5c3SltVb12IlD88NlmmxD7ZRnJJNMmkCcTHrLwjXH7hhO\nN3Hhsp9tVW8DDlu6bsiQ25aX0ej/EIA/CuD/IKK/lY59L4BvIqKvhCDHbwL4Yw8XJYZYoRR0j1Sk\nzIciAvSWqy7UR+taQ0TZ2KgbgxcQ5cy3hxAQQ8kJL9kkRZtdVynf+yl7ykh+GQEZT4x59lmjn8mC\ntsb61kCvK4OSCXNLPbTA1HLWraZfxqMXK+CgE6LcbEA+UWRlzHKFIJRViKQnLhG8ajQvbaj7UNNM\n5bf2QTV6MGWgJyIQu7SSsHsBQLx2iHI/qr1qm3Fq5d2omQ0Z8mbkZbxu/hd0nRdx2Wd+R8TbOmmT\nHNLLXzxItkBvqlYQceV8z9davT3UyLquK9YlVq6T4hde+Oh5Fi+YeZ4z0McY4ZKhdZpEk59g6yp+\n6AJ0VoutKZdeG2ubwTYgqtXk+4DWTCJsqS6xcwgVVnvpiKeQBmTVwN3zAiJnqZPaEEtE8FPJXknE\nsll6TmpGOYul/lNa172AKImN0P7avP094NfrBtgPGXI1kbGUX1gxhkYw1y+pvuz5pY8BRNuEX0BD\nJzA2wKneM5IyOOTgHPWW8d5ndz8Ngmr92ZFpokn8wUEbbZNAGTAB5AyVdvNsK5cA/yGNtdbqSwpg\n+bI9J8ddNVXrSsmuNMT/3efcNN2nR6XvSpk5R3AOwsXnVVWEIyeGANJdptRX327s3vDtQGWzsXEN\n8ty2QD9AfsiQIlcC9AyOIYEEIxJJIisqzaPpDiEwAiPlgocAh553BK85d6cpcfEBFIHZeaxhRYwM\nLAG0yl6siBEcF6zLCespJKpGwu2fHA4pshWY/AQgwsUFnjz85OFcimYNBIqymYlMUgTnxLuGY8Qa\nS2TqukiOeu0zUPvHy3UxGZ4J8+wB1Nvu5RFLdgU7CUzTAd5PCKFk0LScuES6BojnywHOycYtWp62\nWQ2eMjE6yMYhhfNXz5wQAohLZktZARyy7SMGIKwKtgekLWzTRJKyUgaxiRAAz4yjdxKKl7OYJiov\nSloK5xzCGvHs2TOs64pDek4W0gfIf3bks2n/eLue557C9Vh5p/w7uwqgV81X2JiyPLfDXRsa0107\n1IXQPOkz1tyuUjfiTXPG6XSSoKW1plcUfDVidft7W2/VHirpBqrLSPuz1ciBfT/5S//4VJvX61qA\n109dNdWeKv0+6Bhotdb1Ua5TWmffXbUts9duiaatbQ+aB6gtszVm99ptz/eM1EOG3KJcBdDnXDcI\n2YtCl+wqAjxiHJTf5vbmhd4CffGs0b/z+YxlOec0wgSXk4gdsjafeHmzufaWvmFo4i9ti3DXbgN4\nVZc3IL/93l7foyd61yg1ZF0QrXtkZdg1MQj9dmxzyoi/f3JxdJY6uZB4rmMwLi6wZZ9ZjSnQFM72\nubVtu0Rr7Y3RkCG3KNcB9Mqja3QktkCvUrw40q0GPHovfIwllYAChvK7qiFO04TZSXi9pieweeOd\nI3h3KUgnQl0PBeQtoNX5Y9peXQJ2Pefc1lfctsF6u1jwU3uDlFFy6VScNtdlbdu1XTVVdJMBem1n\n2z47SdWrsDSeZhJotXVrF7ATbeuF02vfQ2M8ZMityFUAPYMRwgJ1pIlgxLgCbquxFg1VwN5SEbm8\n5AGiYKDArkCffegjZwC8m+9wOBxwd3eXt+mzbpBpD42tg3/TPqFsGmNo7ufWmNmCUk/rLcFL9X0W\nPFvqpgXMHt1hNfreGEu523w4uroSkLZ1bROrPdQOnbTrDVIKwOuzy0bydG80Sed6E3C7QcmQIbcs\nVwH0QHo5lXpB2IBQ0eQiAKUkAAFVAfXKwJkoBwvweTeopNVzMu4pTaPZJtssklYz1vILlUTJeyVW\n9ctn4xNek1HdMWhBFXhYI93zOGkpDltWT6Pv1VVWFa4B+sSZExugtWmQra/9fvulbajqsKst+4ws\ndZSNwXsT2JAhQ7JcDdATiZf8FhzTecd5Q+kayMRLpRUFtzoYasmeKqrVy0bUR9zdHTDPUwUoQKE8\ncgZGrt0MS34b2gC11a5Tq2wLd9vc1tNfsWw5agvmlsqx1EZrhH0RUJR+lgAmzedTp214GOizVp42\nDS/7027v0fFX6k1B3hpoe7RWr6whD8vbNUle6zN4kXbtUacvW+5nQ64M6EUjFMCvB1CAJCaf8OJD\nL0v+sKEvbA6bYnxdKqpHNXXl5mU3J8ppEuQ8YZoK2FQGX2PotOCvk1W+nnra82Wgr71jao+XPdDX\nFYZNrWCB8NJL3FJGpex9zn1dV5Az3jOVJ1HN7feMsblNqQp7jV6nk6xdndl790De8vnX9tINGfLZ\nlqsBet10OsaIiJTr3feMchpgsxqwp2wMBMQj53w64zOf+Qzu3xB/awsYIQRMhxlPnz7Nm4Usy1Kl\nwFUXP50cpqm1BZRMl3LMwe7OZKkFR2UrwZKUqwbkSxq4IqFqs9aYqv2Svmk59e5Veq+uZkQDT2kN\nQu3Pr1k088Yfrm5XWWnoJi0hj5H3VGX5bAFX67Dtl41LShBUD7RtNPP5fAYzY5p9XkWot45KO1lo\nX4YMuVW5CqBXw54GClHF9aokIydJ1KzcF8FcuGMFN1nirxKd6QnhvCLENWmPEu06Zw1eQGhyZX9W\nBcPSvprm6GmT6sK58fagLXD1jZtqf0A+1nqTtOXYz73zbb1bimT/HtsWe41clyYVKA2j4K2TV31/\nC+AtfbQdi/pauzJ5SEO3njgPrWSGDLkFuQ6gT252TvxvNnvAbm8Qg6y8wCFPEAr05/MZwfhdq7ZI\nRJinA47HI45PnuBwOGTgmlLyMgv0ChKVtihbZFT/AbKKqACFJLOmDYKydgW9tAahsiWfnusBZA/k\nFTjtPXtgb8+1XHbPiGvLsemeJT0CjKYf8grAtsuW8RDo7vXNpo5oJ+G9icGuRIYMuWW5CqAvZEjY\n1drkpRWt3mqSVptX/tbuBqXL9mzc84TDQcBeDazK1bd54au20NYAWPWhAXlbZ9uPku53C3ot0LZ+\n5O1nvULQa7dc+KaNWs6ORv9Q2xToOWn2ujVgmSzq9ltarFeupXesXBpz26bev5ch+/Iiq5wxliK9\ncXinrBavAugBRuS1A5aXwCaWzUpYXSjPAAron04nBMPPE1GOelVgB2oQamkKhiRiedCgaVIC1O2s\ngbbNM98UsrlPwK/uf698+VN6hisvoZYaqT/rch8rBWCtdm217W29uysKQqWlt+6Vqs3XBu86N32v\nbTadxZAhtyxXAvSPmxnlBS6AoYY4ZsayLDifBejP5zPOywnLsgIJDBTscxIsKmkBiAhkPEYuAaO2\nQ/n0lpZp/9rJQdpfp9a1/HlvLPa01dZmoDiogLvXD2ss1Sg1nYQsAPe060KJxGqlA7QTZt3uHr1S\naK2tp5CuxCzI2/ZY+4Wl7dpx6vVjyJBbk6sBeqVjhA5QEOlTH4AARYgr1rVESCrQ39/fY10WMGR3\nIn3RNY+NzVuT88HEegJh5grE6vptwjIF6dqVT8tvDaqtkbIFwJaqUcPnnlhaQ7NcOtefrLT8agLi\nGixbrbsP0ID60ZfNVTTvfq3Bt9Gp7WQqeYnKykon72VZABT7ivXIYZbMl23fbB/3KMAhQ25RXnbP\n2N8E8A8BBAArM38VEb0fwF8C8GWQHaa+gZl/9xFlZd5d3RFroN96aQgXX1wOlaM/nU6IIUgueXI5\n7QERZYNrGwhFuAwKBfxq6qGAST/opy6zpEdoy1YKygJU6wfeApedlGQMtobhHt1kNXOb1GwvZUCr\nFWtbY4wIUcZc94HlnF54y723E4ntt12BqI0FQKXNt6uevQlpbxeqIUNuVd4Kjf6fY+a/b35/N4Cf\nZ+Y/R0TfnX7/qUsFRAaeRQAp8jWsZfcgFUfHbPQDIHx0pOxffTqdcDqdpLz0op9OJxynGYfDDNkg\nRAB+Pkwg7yE8cszJuZY1+WhPE4iQc+HY0HwFdK1Hf1vqoTXkzvOc2lwyMspeqVsPmdazREDbeuMU\n/3SGTG6MABDyNochBDx7Jvnoy85Y2oYI3VKQqGw8wkyyIkDMHkPSd7mH44rI6lsv+eQ1KlbbE0IE\ns+SutykjdItAtTeoPcE5vU62EdTxd54wH9I/zYWBwJjI43wOOJ1POBwOxs9/GxSlm8UMjX7IEJG3\ng7r5egBfk77/OID/CQ8APQg59UAbTKNigdQu+xXQNd2wihpb52lOYHfE8XgsuWxMefr5GI69vcd+\nqrQGyEulHqItAAAgAElEQVTn7ffW66dPo5iVAnsQlY20e6kgiluiaUfXtVODvmrvIq3LZsG0k5Kl\nmtroYa1/a98ALB3VTnaXKaPtuPWekb321sH+RVY2tz5WLyp749Ub88/l2L4s0DOAv05EAcB/xcyv\nA/gAM38ynf97AD7wUCFkXnp9yW3+dD1uX27L3Z5Op40rpeaWV6C3Ccucc1UeSaGN9n3JLeDp9QC6\nWuOll8oaXXuGQlu+2g7UWKnXK5jncWKXJ7wQav695vtNfnqTXbPuExuNu7it2v7autsJTcVu8Vf6\n3K526j1223L3JlG1JTxETe3ZKIYMuUV5WaD/p5j5E0T0xQA+QkT/pz3JzExE3beMiF4D8BoAvP99\nr+7uo2quN4BWQN7yt63PtoK9dafMQFjamMt3jZaPzjX2XDsh9ADHfvb61FI+fWOs2aSDPWwELdiB\nI8m2fYbWatMl6Jypu2Npu6yLqdRZ2x+sm6NdabVePXbcgdogquXWY7sFejtx2n63Y97y9XsyAH7I\nEJGXAnpm/kT6/BQR/SyArwbwW0T0QWb+JBF9EMCndu59HcDrAPCPfOiLubd8b32r1UhnPyXHis+R\nr3qvJis7HA7J02auUh6gAe8WxG17egBu69LfPXqhNSBa6dXdaspybHf8q8lPvW5iDBut13tjS3AF\nXNu2sLUBmD60Wn4rfdqrbms7BpaOssfazx7Q21VfbzJox2nIkFuWN70zAxE9JaL36HcA/zyAXwbw\nVwB8S7rsWwD85YcLQ/XiWu1bpQX5dhMR+zLbjJTH4zGD/TzPlXbfGvIURFof9zZhVqvpmjHZXGdp\npkspdXvgVOqqPUmY61WETWfcHpdxilWkcBs5bFdTeyCuOeF1srTHe+Oo5/bGZK/vLafflq/frdeU\nDX6zZT6k8Q8ZcivyMhr9BwD8bHqZJgD/NTP/D0T0SwB+moi+HcDHAHzDgyVxrUFbbVLFbv/XbiZy\nPp8lZW56sRXoFdj33AZtfTAaZ4+qsW2xwHqJU241+h6YKyXVo4QsH90GCFmOW6/1vu5nGa9lA+Yx\nbBOwaQxDqaOUr2CqE1cvKnXL2ds+xM0kkutkaiYy3lxjgV5FqTm781T73Hrj/m6VF1m53MqYvN3y\nTlktvmmgZ+bfAPCPdY7/AwBf+4Jl5TTBGrXaDmAbFm815fP5DEklLN1RYNJyhLuuwaP3gCwQtjRE\nj6t+iCfucdi2nhao98BJAZCo3VLPQXL/+M1kVugkylROAdbtKkA1ZJtt03LydlJqE4zpeO4FR1lp\nJ0n5c12Q3163NYbvUTe9CWjIkFuVq4iMZdRUiL6krWHSbh33/PlzCYwyBljNvf7KK6/geDxWANWC\ngfXqWZYFMABkfcAtldDmNV/XtSqn1q59BUha9p6GrvfbDbz1GnVblGtr11KguJLa/VXV3XSaphxl\nuiwLnj9/jve+971lfA9S7hoWkAM4lvQJOvEqsOszsH0uxl63GadpKjRLGyyl/ZVyy/PQ1Zr1oHr6\n9CneeOON3A/V4q3Lp8Yq2PHWOoYMuXW5CqDX1AeWf+1pdq022Wp2+rJXXja8Xf5baY3A7XHVMh/K\nA2Pv29Pye0bd9p52wtu2S4+V83ZloFQIkQdzRAhr3qwDEBD+9Kc/ne/VCNTj8ZgN223yMP3sTcZa\nf5tawVJSD2vpdbyABXH7vPW8Hr/0TAc1MWRIkasAes1eeAmQVZtUTbUFSfWwAZATlwGAN5GrlVbf\nAYKWj+95nOzRObYteyB9SVpjaI/OqdvhMqiXY0jeNUJjqbFax0wn0E9/+tOZ337+/DkA4OnTpwBK\nVKmKjqNq373Vi7ZbtftCm8Xquj3u3RqXbZ0A8l6/bd12cmk5ets+rWvIkFuW6wB61CCpv3vXWC8R\nZs70iRpgAeQXn4jgGqDPZV1ogwVrlR7tYMHfTggtF2/vsZ9tf9t6K79/3nqs6DVtu6yHTPGtL+16\n4403MM9zpdHrRGnL29PKeyBqJz4da0mJsAX6Sxq93m/tLWpsn+d5MxZ7E+JjVl9DhtyKXAfQc609\n9wxzrZZtA6csLw40vvgmq2SrsVdlN9rmnjbYM6Ta32359nvPddGKBSW7+ui1ty3L2iMsjWNdVXsT\nqZaj+ftbV0Xl+fe08mqsm8jZOgCqr1nLuG2P2b4rZ295+MfQcW0bb1XeDf1/kcn63dDft1retB/9\nkCFDhgx5Z8h1aPTYuiK2WmtL2ehvoLgGWr7YarXAvjZpv7dUzd61e+1utfz2fBu9a8u1/W1XCXaF\nY8vS39p39VjR8dG2aOSwTf8bY5S9ddMY3t/fZ5pGeXots+XjeyuNh+gTNR63KzcRt+mrLasNomLm\nyn++N+ZDqxsypMjVAH27FG9fVBvJqYFSCnJt0EzL3/ZkA0QdoNprZ4//bQGslwvG+uf37sttMfVY\nwG4Nnwrqlm9Xw6V62uj4MHOVylnH0rqY3t/fAwDu7u66fH37W9tix0apFplgtve0lJEcL+OrfWvt\nHC2dp/3Sc73o5QH2Q4aIXAfQUx380nOvbJOZqXbcJixTaQHJSg/Ee5y8BRab/GvPQNpqnZfsDL0y\nLKCrT3mbkVPv1wmPmUFw8G7KXkkWxK2R+nQ6Vb715/M5n9OJ026zeDgcqnFsted2FaLn9dk456u+\n1xPxNnmbXZXYWIb22elY7Bm9957tkCG3KtcB9ADIA0wRawzwnRf3M595Bu8PYHYIAfD+gPNpFZB7\nRTbWmOcjAHGvJIZkdARXXjgZTIFa605ZIWNkyEbfCroxf8YIhLCAmRADA3CSSCwCkc/pOp/SAPsM\nesUNM61YnGr+aUN0jUZlAbd1OSFGcXv0U5lY1vVsKJW10DTxOc6rQwzTBnxVE1ZN33uPZ8+e4f7+\nvtqyT/MBvfHGG/m6p0+fZq8m3ezDTnRAmQDz5AuHeRb3Vob48Wu2Td0A3jkHcAm0miYHZnEDJQJE\nUU97xvI9It/j7onDunwG67riyZMnIHgQgIk8Jj+BeC3/luSBgDmCnOt6WL3T5UWMk7cmlxS5z5Zc\nm3JxNUCfJTICysbQKuqdYTlcP1HeBzZvKAJkILc0Sj7eoQ8E6EvOdEmm2NPONThI25LsBpEQE9AQ\nxUKJmFzxWg4RATk6VvzckSYVZI2/JEAL0WrPq2lP4eixpvLCVEXLKo2iG7PYCUAjaXXFoDn9tY2n\n0wn39/eVFq5BaT16LY8xbLwCVZ+MetWiZeqEJH9bu4b10bfPkYgA108f0bMjDBlyq3IlQM+ynyCh\n2vS55X8tT5tpiYOvd45C0tRDXU7t6kfgJncNmXrkR03h2MyRMQJh1bTJCkLWOFo2OOHYBFZR6YOs\nDjgDPUfVlssqgFLWBeH3beqD4t8eeU0TzpzHRoH8dDrlTdPtWKpoPTbNgfc+8/k295CWaxOcqeSA\nN+plkpxAlJ6vSf2QtXsAHAlgB+ZQrYLCWlxppTxn/hKgN3VaSqc15A4ZcotyHUCf30NZwu8tvVpD\npYbK94yubX6VHme/59euWmhrYJVPSYsTYsCySArgEAIYK2LQSWjBPB83W+vZlUSMazG0OlRADxjt\nVVkd5gz0nFYfefgMv289kLT/Nrq1BWwLulZT1j5bo2sL7lpH+4zyuCGahyu2BNnj3RhnI2VrrNTl\nEONSTY4xoOpfZbRHss9UQK8rG07/nrqPeciQm5HrAHphzLdHNzzXNhLSglILQt77XGpbVrvUd0mL\nl+OXXCmBEBLAx+IJBLJavwfRGbIbVM+wvFZ0BGk/Ytl4O99jgD7GYPq/NRY78lkD1/J1MrS5YfQe\nNdwCyOkSdFxsdkq9z06q7QorTxiwIK/Pp9g5mBlscsPZDcikLlRjlj8jgTxA3oF8ahN8Bn8ik4gO\nDCYZOv0cMuSW5UqAvkiltcetL3UP0IGtH372XkG9Gmg113wt6omgLUu1Wc3tHsJaDLcOCYRdAWFE\nxEjQbJNA7bljwVFtA66JBdBzpU22L9vkZxoFbMfLrioOh0O5Nmn06lK5LAuWZcE8z9nzRd0k7Qpq\nz7vJ/o4xVjtVaXt0pZSeKGRyl52oJN+Ny+NVyjX5baIDeQ9HE8hNgFOO3m/a1PL5Q4bcslwH0JMC\nu3LlArzBgIjdDcm6HGpelN7mE845eAPmbfCPTcfrTNBOj+pRKgiwxlLhzQXAezEADAZnbHM+TTbR\nrEZg9rmF8uHFv915S4dY76FiE2A1XoeiCbd+5i31oemHrQHbgrY1IttzrXupHUvpHAk1w3aCJAB1\nxlEdL9u+ihoLNs5A7CLOtfEWHkQeTADIbDsZjO8/uQH0LynDxvHOl6sAegLguCyxlatl4zKn1MK6\nLkk7LD731hNkU3aH5rHHC9dbrwxqjbvmrtXrRagkl2kU5wjO6XXbTawzGDqAIoDAyC6XcHBOIz2N\nv/hkk7FZm0KsgJ6ZsS4uUzEa8GSjYXubqmvZ0zRlg3YvcG3vZW958xgjiEt/dTINwXg1AVW5AvDF\nq6p2SbXBZUrx6Z+2ofa5tzRVuwIZMuQW5U0DPRH9owD+kjn0+wD8+wDeB+A7APx2Ov69zPxzD5Vn\ngUfBTn28AeDJkyc58vN0OsE5h+PxiPv753jjjRmvvPIK7u7uAIg2i9jP8c4sPuURdeKwrJl6QoyU\nNUpKwVyZz6aIw1FAUd0RnXMIMWWCZOTtC4sWXDIxCtCWFAXNmDb0Rq3B2gnAgldMW/E5kn1y7aRk\nJyedFDV1sd3gw66U7u7u8IVf+IV4//vfnwOqZIP14rJp25vH1NgECsCnfuZUxLq64jx5A2pjcFiJ\nEUF4fhJPoXVdsAYGkwc7QiRgvjvicHcHR7IqWWOE9+Wf8nS8k7pjxBouJ5IbMuQW5GW2EvxVAF8J\nACQq1ScA/CyAbwXwQ8z8Ay9SnthCPdQwqwY8U1/Fz7fUgdXWBVhU28XmntTmhkoQKXW4rLVnw6QH\nXPKNB0VI4KcYbx2K5u280AwKaq6qw0OCgabSfqj2qdRN6Yv3tbZc+lE48DwRkIkjoOJ1s6eZq8YN\nIKcAnqYJT548wd3dXZW1sp0UW2qrrSOP9QVrqKV8Wkqn2DM8nGM4F8V1kyYQPMDOuMHW3j9CjQlt\nA8LmGQ8ZcmvyVlE3Xwvg15n5Y2/qpWIA7FKcUgqPb0BYNclegi89r7+FPigGwJ7mzM19rFGyqMsN\nQWma4smjQKs0jZwDBMQFnPU6ajhiF9UQqsZIwJFy70rdFOrCeQuGxY9emyirkHRfnNFOZlZKUFLZ\nP1Y5+uPxmOmb97znPXj69GmV3KyXHK4t304G5RnZVYilWITqsquOYkvweRXkfXnWOq7lmad+gUFc\n57qpufwB9ENuW94qoP9GAD9pfn8nEX0zgI8C+C5m/t32BiJ6DcBrAPD5n/dqdS5TDR1PGAvmbeRs\n5dPNybsF9a5N9lPvy2VjuwG5lkvORnEGMKu7obanuAp6Lwm9gNbHP9EokcA8QbcFpGyYVA8T46Hj\nLQ1TKJHM7ZOkVJDi/SaK1PbF8t8KhnaDDyLC4XDAq6++iidPnlQBaG3uGRug1BvTUv/W+AoWF1KA\nQa4EbMUYEdbWPuJBpHl7Jng/gciBiQBSt3wnRu/SOhSX3ZoKG/Licg0T5YsYhK+hvdcmL52PnogO\nAP4VAP9NOvTDEL7+KwF8EsAP9u5j5teZ+auY+atefeXOHu8+VKudteczSIQ6qnIffLrtqamfbT/T\nn1IOQs8oJ22zaJaAJL1Of7vq2mlyFadtaRYFV3tOE5QdDodNNHA7MbUJ4No/7ZPWcTgc8PTpUzx9\n+hRPnjypDLOXEoi1K4iteySg/8wkP01NL+kxjpTbxrEc15WAc2lsJ/GyEVrIVdeU6ghMQOCIwJI/\naciQW5a3QqP/FwH8TWb+LQDQTwAgor8A4K8+VMBj5l8bBZvKrpb8FuCdK9xsdnTcoTNU2EajGg28\neJWUazVAxzmHkqDRpWRlUhoogkiHVyeexMHnVnkwU2Wo1Pot/VCoD8urFzDPtEesvWqsP7waXzW5\nmfWWAYS6eeWVV3Jys3bCsTaC3oqoPI9y3JEvPLqhVpyb0n1NTiN2AELVrhitRj/Du1muyxy9aP3B\naPTWAP0imuCQIe9WeSuA/ptgaBsi+iAzfzL9/CMAfvkxhTgofdIHY+WPewE7rSZvQdLhcqqDVotv\njZZZ03Z2Mw+Gcz75SFIG3hjre+U8MsjJvcnlmy2tUBs3sxG34xpYuOyykUhIWTDDig34K7jrn3oK\nqWeQ5eifPHlSedeoxq8xBHasLJhuxy9AUdj2pzxp/e3zeOpE0cZDMDMip8nOtRy90GlR7OHV8fb5\nDhlyy/JSQE9ETwH8YQB/zBz+T4joKyGv3m825/ZKyt+UvW3F0hip7gwIPV66UANbj5CeQZFj3IBq\noVJEQ1dgy9rrdjw23HjtoRIrbr2AvNYFc3wrFlRjDNk9UhPB5d9VAFKsVjxtbhoFTs0CejgcNoFl\n7eSn3/c0Zjlm4xSQA6hq0Nfry8TmycG5QrVI6mMPUK3pq0RyW/7Rrh4GXztkyMsBPTO/AeALmmN/\n9E2UBA5ncE4hIM1yU2ne+bwghIhpmuH9lPy2BVDm+YB5PmRfauc8mIF1DfBUAC2n/g0B1GiOqMDM\nUhSitYMd5ukIG5Kfc7IwJAc6MyZHFZ9NRHCz5mxfoJ4oyMDOQMpP73ya5lg12YC41gZn78WvfV0j\nYhBuOwaP8/mcE6UpPaMavHrZrEvA4XDEq6++ive+973Zpz73leUaHa/Jz/DOZzC3Y6j32JgB5f61\njSFaf3uHoGmZg9A3koxM8gyFZPCmc0AMAazbRCJimiVAynkAFDEfZomShQPHAI4he+lIhbGaZIZx\n7p0jw+j69shVRMYC6aE1mnK7TFfgUoBR3nhZlirk3zmXsz5aF0grLe2g1JH8roG+1tS5+V33ofay\n6fezXXmYs13aobdysd+VrlmWc6FzDIWjbZrnORuM2/o1LYI1/FrqqF1JtbYSW5a1MeyNk30OuuIQ\ne0L6DLrxyIqAbYpkvffSmF8yrg8ZcktyVUBPruSbaVPPKqCLVls8VxTsH5N3vEetZDDoHTPafUv5\n9Mpuv7fteQj4LK1R6qtBTCkY/RQOXvaIvb9/3h0LHS/1lVce3gKkeAHVQG89eqxxdo+ff2is23HS\nY3bjckn9vGJJ0beMAPgC8NrGEAKQ/i2UfDrbugfIDxlyNUBfa7iO+q587ZaANpjHavTee3inGmld\nU9aOW43eTDK27sesDu0k0AP8PenZBOo+1xkgAVSeNOsSckqI83nNmSitp4yCowV4C+oq1i20zV0j\ncQRbW0c7Vq3R215D2O43q582kKt1AZUI5JKITdup57WvmrJir/4hQ25ZrgToE8AZTZ6Z8/IdQM5t\no7smqXbfeoTotT0QVdkDgHKNBe3WpXBLtTyWK9waIbfRv/azd6/y8MrB65Z/yxJyMi8L6hbo7+7u\nqomxnZSsgdVSKiEEONQgX+XTbzR8W94luWzMLeW0MQq1Mbtf7qXyhwy5NbkSoEfFzfdEXf6A4lIJ\nFI2uTdAFUl536zlivWe0bpuErKZLWv7Xrjb2eWH7veWu9dyWYrCAWXvHqLTab3GZjBkQj8djlavG\nUjdan9Xa23Hp/cXI3XFs+763siHUY9BGNWub7PMAAD8V28KWm++v/PaODRlyq3JVQA+ql/WWWrDU\nSktntHSAAj0A48mz1TAt/WOvacF4r70tp27LfYxs77Eui9vyLMgzcw7ysmA+z5LB8ng8blIYKOWh\nQNsaNjWQqh1nCUDbuqVaQ21vAqkmOq6Bvv2z5/I9KLYD7YtN1KYUn4zHpef07gP9S8bnd7I8dnV8\nLfJOae9VAL36ukO1PS4URL7G8sVU74PabjwiILhN7nUJtFuNfp9GqbXWHsDvTRQKUG1glwV31eb3\njJettg/YFAu+AsY6qraMTQhmYxMD9Hs5gcRQvqXC2jK0Hn0W1cqJt2Nl/f1VLB2kZfdSMGyNwnVS\nM/t83ikv45Ahb5dcBdBneM2addI+zfJ+nmfc3d3h+fPnAJC1T910JATxPAEkd/p8SBtpONoYCUUb\nLECi9+s5C1QhLIWnbqgOW54FPtWarRYKSLIzvc66SWoZMoltNzyRKFhku4Scl0/dYORwOORzWr5O\nllqe9sH6vmvOfzvZlfaaPDrRJnWrVxIWlHUybieBGHmTh8gCtdJQh8NB+mnGVs97Y1uYpglgV9lq\nVGw9dpIYMuRW5SqAHqg5+qztNectRWH3Qm01Q3s88FYrF42/0cgrvhybcop2erkfPeqnUC/bVUKt\n8feW3tF81vnlnXPwk+SUUZBuvWasRv8Qx25lTwu22nJPY27r2PaxZ9Su6SkWTmxzrn2+eSXYaW87\nWQ4ZcstyFUBvX0OlXdqX0wLYNE1VQJANotIy1FgJqv3A0dSVv6MGP2s07AGicvT17+3mKHY10QP6\nfrmdiYnVw0iv46zdejdVe8D2UkXoxHAJ6B9Lc7Tgasvp5SJq70u/Ns+konq6k0ML+v2grHZiGRr9\nkFuXqwB6oOHVqaZXgPrlVd7WZmhU2mBbZn1/LqfhixXoAWQPkx7Q7xnBar64HLefqrFfAsq6jpR+\nAcVwnNtrQFniBias61r5yNtsnm0A1F7O+tp9cSu9SbO3k1Xd757WvU0VIX+afbT0bW81QtTX2LXv\n7eQ2ZMjLyJ7C906QqwD6dviyFmY0se1Lrn7lMWuyew9CKZ9cHzNcRxPUc9sWbQFF2tF/0HttfUi2\noFXTIwJ6khhNri9BTmrs1Cyflku3KZ71uALhtu8P0y/2+N6xh2khMvy93RYxTQqG4y9//QmoB/RD\nhgwpchVAb8Uu2y8BZMvpXtK6bdbDfN0OgPXa09NWbTssGLda96Zv5vtjwLBHvejOVLoZR1lNFOC3\nHLoes+Cun5e8bvYmqpbi6mn3Panvc5v+y+9tLh1bzx4t1I6dHd9L4zxkyC3IVQB9616p0vOWaUFl\nL9c8UANH+b0FYSIJ5+lNFi3Xbe/vafS6clD7QV3GXht7wUYRLdBr2eq9I1vryTGlr/YAkpmrNvUm\nplYTtuPkqHZntLSQHYPWO2nPJtA+i/ws0566OhHHGOF4a28gImiWT6VxVGzMxQD5IUOuBOiVk1dP\nmKgccqO97Wl1e0v1Hgdt/1qQbrXsnlZdFgSMFkOsJh3Cdvs666v/GABqgU2BWd0X9a+MTX/VoHX1\n2tS7vp3UeqCtmvue4Xhv9dOW0wN6oOSRZ95OEnXMQ3/cevUOGXKrchVAzyDE6VAAJcYqzw2ASoO0\nf9572VA6Nvw5iVF3RRDjqmPjPw4gKAhGRGYE2Ysqx91EjgBHeNReG/LVgx0BOiFFwmSMn4VPp2rV\nsa6oMm6GsALYZorkWFYEAu4zJu9y4q5JzQ3swGmYJjeDpgKcIWzB33sPZIOwjPwajB+9JwQGVg7g\n5CFEngBHWMEIyzlfi8ljhWjSgeQzxghCACKDw5qeke4TS2DnsSwBCBGHyQPkcV4XydNDADsCuwmB\nGV43Wp8OmGdZufj0eBwS9QPGGlZEMJw3k/ryDI4ZbDZhGTLkluVBqxUR/SgRfYqIftkcez8RfYSI\n/u/0+fnm3PcQ0a8R0a8S0b/w2IZYrr33Ylpgb90ENcXt+XzG+XzO2+b1km61fw/0vdKkrcau3x/j\n1VGvCPoacHWctlq1TBZ1sBE5rv6aEQVz2O3rQ7aHx2jFtso9e0SPzycizYiwWVE81K62P5fGdciQ\nISKPcU/4MQBf1xz7bgA/z8xfDuDn028Q0R8A8I0AviLd818SkccjxC7fe54e1kWwTXCluepPp1P+\ns0nOWqCv8sU8YNBrDYM98G/TJ2sdPYNlew6oc9jk+p382fNyjwZOMdQxqZQv55g72wlyADcTRRtc\ndRH4myCyzN3vHG+P7YF9b/xbaq5VAjb2lUHVDBlyUR4Eemb+BQC/0xz+egA/nr7/OIB/1Rz/KWY+\nMfPfBfBrAL76EXVsALJ9gZUO0TwubZpdm1K3DVqy3x8D8nvt2/PwabXXvcmqbc9e2RXgEgMk2wrq\ndxCXiYAY5CJA9XmQJD1TgN/T6O3v3vde2zdjEDn/tWXsgjxr5HLEGgPWIH9tHXsg347VkCFD9uXN\ncvQfYOZPpu9/D8AH0vcPAfhFc93H07EHxRrwmHlDiVitusfTW7Evf9aUGyzYMxja80AJuKpdE7PN\nUGu82LfW5bIFqEt0xJaWiJD0vOqVY33/xeag3+uUyvX49PprQbR3jhjgyIgdaqncX+IQpO3b8dCV\nS9k+cEWMGukcumXbtuhnnjg6bbbjOGTIrctLR5ZwebtfSIjoNSL6KBF99DPP7l+2GUOGDBkyZEfe\nrEb/W0T0QWb+JBF9EMCn0vFPAPhSc92XpGMbYebXAbwOAL/393wRt1ps5/oNZZL3jYWvXAd7y3/3\nKEtBuT/z7di6XTJH8bopB8AXUvy2Zevx1pBYa6vyBwgXz6weP1pezNpyaVu7Aupp6L461lIheuVF\nWisyyJEwRI4qesUp25Ra48iBTSyDzYdvNzJXjZ528t4759BkSq40+va4ykOumLcgD9lBhtRyyW73\nTpU3q9H/FQDfkr5/C4C/bI5/IxEdiejDAL4cwP/+qIbsGAZVWn5dgV4327ApcpUOsHz9Yzw6VHrG\n116yrh7n3+PvL1ERVsRLRv/KhFC8jdT4aiYxXsEI8skKlPavHcdQtb2dFHs8eN3Ghn5KJgHERO10\naKipSvscNn+tTUMT16k9prXL9J5f77d9dkOG3LI8qNET0U8C+BoAX0hEHwfwfQD+HICfJqJvB/Ax\nAN8AAMz8t4nopwH8HQArgD/BzNsonX491Qu8Z8Rs85l77wEnecnV916215sTSGw9SuTvMe0BqBOS\nL7sZNZy2WQU8BEI9Ln7vfLmOAbMhCqg2OPfaL+dKYrS2/PZYu4J5yI7R3msns80Emcq2eXZ6xu12\no2btiPgAACAASURBVBH97r2Hq7h/M4Y7fR8yZIjIg0DPzN+0c+prd67/swD+7Is25Hw+Z02ul6Bs\nmkp2RqtJA8D9/T2ePXsGkPx+8uRJ5ZWjoKKblQjwtEmzJqzrOdcr5a9gcjkFsBgNI7yf4ebJgJgD\nMVdBTnbTkXVd83cFtnVdu7n1LYDVExtXNIR1vSz98NWmH2WMXK5Dyigga1M8MEuQEWgbIex0gjRt\niKlf1sWUnUsBUqV9zskWhudlyWNmAV+ek2wiU7ZDnAEgr2RkbMvkZl1siWQDFpU2BcKtUzdDhlxF\nZCzwsIdEj8pQ8NJjbDjp1icf2GY1bCkVe9x+tnXbclSbRwdMHtLst+cu8fpNGuRIqVprT6g9UlKv\nqz5oWZfbsZW91UNLcyGtgvRaSxEBfTdY9WDyfhsQ1/anPQ4grxSGDBnSl6sA+pYDf8gYYvdGZeb8\nmw2AaXIt5tJFBegQAsDFhVMAaRvgxEziTtgk8HLOS8i+Ai8DdEFz7E0YPe4ecEIXESUDbBvV68x8\nonXLMXJi9NTzMgkUg2y+yxzrgWYrLWe/6QdLXhonJFeesKK5l7lo77qayascKs9Gn2mrrWtfexOw\nThR7E9SgcYYMuRKgBx7n82xfdGu4dc5hnmdErqmaKggJ25e+BvomWAk1yOj1co9DZMMv72j0ts1t\nH9v+bnhncuiBNLMCpgCqcPB9UJeJQ+tQ+8G2Xe3Ylvr2/emFskI3PgFpyhVt3qwy0jVVPnxS6oea\njc2tV1NdxwbouZ/Ybs9mMmSIyiX71rtJrgboN/nisXWTaxObASWQyWr0vQAl+zwv0SR6zBEhiqqY\nj2WK6JGUb2tUbeto+9u7Russ4vO5+l4Hi7p2cuq1o0cP2bp6uekRGU64map9tj42FAoZusauTnKd\nruwi5n29H7B22T676t5m3Cx10/4bGhz9kFuXqwF6FQXmPQ3NfteXOIOTeaHbXC/q710ok37U6Ka+\nHQ3ctod5Cyg9quEh/rv9LROON/G5qL6jTG1A0uizFqtnUpZHRs1hX5pYexRN+rK5lixtA0KkQh+J\n1l/z9GoQd87lFgltU7h5+TONbdpV5QSifmTsnsIwZMgtylUB/WM8JBQIrNHOOcpcOaBAUAciMQs3\nn133uHUFdIn+qOtBQ7dYDVOviVwzGJf60VIjvWsFIGXVoZx9qbvQEZb31j4UuqkAqgBtqhNlL9Uy\nVn3tvDrHQEAN8Bnkrc3BE2K0k4S4eKrH0rIE05/tCm1PemOVnz/RJmCuN95DhtyqXBXQX5JWu6w2\n3og1LSDaYwsOtW+4bmVXVgQ79IoB2aotZEEPlR99297eSqCmleq+cQS4CXYS6gpQrR2AuDHqJMOU\nXS7lvN/UrZRVayuoz+9z2y3Ie6rTNgNIzFLZmIS4aPIhBCzLmve01R2i2klGn5d9Bnas2i0RB5YP\nGXJZrgLomRnLshgt0QGxcLwA4LOWyGAn5ybncJgmxLjgjef3WM/P5VrvcXhyB5/8r6XsA6bJA+yw\nLoBzDO8nOGcNtwJW68pgDxDNgAdCjIgcQWmbOzc5IG2OwixeP+pfHpPbIIJoq96E/9tAIOYIOA9v\nNGvHRYtW7b185+yTrpGxbmK4CcmL5YzJFWNmjIxlOWM5hezR4pxHGwzdatIrSywBK/gyScwBEwgM\nggdzQGQgUoRzDDjAOdXUHdg7RHDaF2DFeTnjvKzJFx45huA4Fb95bYbzK8hxJqVk16lkP/CEGBnO\nmY1dOCDEeivB0/l+E2V9C/JYavAW5TEr7HezXAXQq2RtOPuD1+czDcF1St/iqVFAKwdHJZDTpb0A\n87zRqPeW+61R09Zr0y7Y3Pfeezjd1xUXtHk0Gu0D/xit1isZOfc9Sqym/SIA0GrU24IjBPI13YLT\nJU2632rfqQ7WLQ8JQK2NC2Cncvb6rqunZpWRv3eor94zHDLkVuUqgL7moAXIxcjWpw50ya/HiqdG\nDfQChpSpghINKt1uA6ouAWIL9BZAFehtIJenRC0ZjLHaun4+xiZhPVbkT9td7s+RqVz7/ds+XKqr\nnHO53OIXn/khWB98tR8wAPXnhxOXyhDSZipB++HhXGyeg6406v62/VegJ0PntPRTj7tvbQ5Dhtyq\nXAXQA/XLWjT7WipvFMvRr4ImFdCDq+yICtQ2EKdNA9CKBceWs97bvKQFIdtuW6YA18Nj0v6ux6h2\nibTeLZf6Ysvu2R4A4f8pgzuq7+mCMkGjKNUxCLivS8S6RoQg+9faZ2dXMfIMt2Bc9b1x53yIjhlJ\nzIYMqeVqgB7YNwICNUA6Xye8Oq8nAPULHrlozyGEvCl330CJze920in8eG0Y1Hts+cyMkFRZiltj\nq353F8DK1m37X5Vlbtd8Mlajb/vWA9ttpQ7Jb6dub4rfymW4esWllE2IAuzrKh424m1TipcxkoCo\nonnb83VbnXMgpy6jpYx20tqbKIY2P2TIFQF9BSr6Spt3tKJM0Gj06bgmElO+XIF/XdecMMtSLvrZ\ny1n+EM1R++TXNEKMMafstUCv7cnuhPQilEo9VqJJ2zTDHe0cNQi23ir2s6wy9DnotWnSkKuQHwo5\nk9OGMk2zxpiBXv/sJD3Pc/rz1c5g5flfzpqpfW/TT7dafGX7GDLkxuXqgJ6IFFu61AYz19GXCTyc\nc4jnkqY4xJBBXSkcva4AtauAP5ffoThaO4I1dPYAtr2/7YOeu2QEbo2/5S9NKMxdCmmTdK1Tbyt5\n0ktup0QelHMcxGQIFW8hIg+kTVHAHswC8jEC65qM0ysjrJyOF/rHHSYcjzMOhwNKzvw0juj7wrft\ntLYI9XiysjcJDylyaXzf6TJWcVu5OqCXH+mzA5rMjBgKLaO0iaVwbEpivU5pDf3NzPDeVcBhxYJi\na9jTFYSdSCxdIquO9L2zKml/ZztBpw0busa2CWUzDyLOBtmWYurx9j0AVIAH6o1PmHt0jwOlDPGc\nMmnGGDLg9yYg7efhcMDxeEQIC2JczTOw7ahprt7zyZMgtpPkJXpuyJBbk6sBeqAGVWbebB2XvVaM\n54f3Hm5inE6n/EIfDgfwSjkXvPLn5/MZ0zTheKxz1WvZ1ltFfNHrHC0KNpYi0jZYLXNZluz7PVGJ\n+tRJIbe7iU7deg4VisjurgQkjxaO2cVzXRcwc161tGMq7S8Uh058+jv3Id3HnBInxABAfNm1rbKi\nUO+cQmXpQkujYGXFhKx5z/OEV155BURyTCbosr+Adz7TMjomy7LIlJK4ejv+dpNxmxZhmup/1kPD\nG3LrcjVAbwG1t9V4xcNSrem2HLj3Hp6V/2mzQloj4L7B7hJPrOct923bSURirCSCo74fe2tctBNP\nj86x7Wk1fGuvaFce9fdSh1JYl7hwJA1fNHopR1JDpzQHAMCUgrMkMO10PucUxBKQ5qqJTakntSso\nrWP7ZfsGoDL66nj3KK3uvUObHzLkUVsJ/iiAfwnAp5j5D6Zj/ymAfxnAGcCvA/hWZv7/iOjLAPwK\ngF9Nt/8iM//xxzYmv+jc94ZpNVq9x3LmACRlceKxQ5PczJal9+95qVxqY2+i0TK891mjV8MxUBtD\nZULb2gP2jK+bMTLn1B+9RzHVY1SDvAX6XAfVmT7lnE+0UgL+qJq8oL3aQM7nM06nU/ZQ0n0CdBIq\nu37Z1UHtstmlqZr2WI3fjkFvvAbQDxnyOI3+xwD8eQA/YY59BMD3MPNKRP8xgO8B8KfSuV9n5q98\n0Yb0jZrlJbX0h15lefcMsEmmlG9lTcv6ZVnyPQpEbQBTr009aQ2lem1rMAWEyd7rZwtWe4bTi1o3\nVKOfqnHqXdcrq7eSyMfYliVAH1bRwkPUcYgpVmGpjKAK7PM8V2AvFEwN9K2dxAK3Toj6vaXR9lZL\ntk+3DvaXxuixxx9b7tsl19qud4o8Zs/YX0iauj32P5qfvwjgX3+ZRuzRFo8BYBI1dQMOFGuNXemE\nZVnM3q61Nv3QP5A9g22rGYsm3ze+2u/6s/S77w2z1ehTfWTdCLcuolt66nL/tMzqN8R9Ur4SQlgR\ng2y8EoKM6/1yxvl8lsA07xJtxQAV+4KCvaROcJuxUxsCudotUs71g8DyOHbGtr1myJBblrcihPDb\nAPz35veHiehvEdH/TET/9KMb0klAZV9Qy8u22lxrxLTnVJRekERbS46YvbTXaI/H79E1vTpbrV8n\nnJZO2YL/Np/OHo3VO7aZcHb+2vbmc46ba71EyWJO2TInAA4hMM7nFc+fn/D8+XPc398n2mYFs3GT\nRBDAdwznrZ1Er9hG9fbautdPO272994zGjLkFuWljLFE9KcBrAD+Yjr0SQC/l5n/ARH9EwD+OyL6\nCmb+dOfe1wC8BgDvf9976hea+lqZgkCkOgjJYbs8t4DRpggoXPm+4fMFxqBbNzK4b7n8osHa7za3\njLa/lL83IZW2PrwaseVZKsy2SX5EoMrN71E2NhFPnOUsnLydMGWlNCXahuD91vBd6ty6tcYYc9Rz\nd4I1VJVeb1dPe89kLOWH3Lq8aaAnon8TYqT9Wk5vGTOfAJzS979BRL8O4PcD+Gh7PzO/DuB1APiy\nL/kA94B+T0vjZh2ytxKwgGalcMk1qHT6WJXXnrs0IeikQo1Xyd6KZa+8nvZdrq/ru9QWBXqgHwnc\n1sf6nX22iRA5cAqGWtcVYS25gsQ+UtxElZu3G333bBqphdVEXrWjQ0f1PIZ6q5PeuA0ZcovypoCe\niL4OwL8L4J9l5mfm+BcB+B1mDkT0+wB8OYDfeESJiIGyx0fv1dSdkYhYUqQzCx1AAGKAn4EZ6l+9\nwq0RHowJhDVy0jgZ9/E5ZppwePoUEwAKAY7EG2Rdl5QqYUq+8OKTvQbOOdS9n7KGHRWcg/DmDMmW\n6ZyDIw/vPNjudIUayLwxSorGOkmUJxGIVgQwYlwQIUZP773k0GFgchNAkryNiADnEwduaRDkeuEc\nyHucJZopb8BtDdJEBISIaTrkXP4hnAE4wMs1z/kez8NzvHF+I8clHI4HzHAIwWGa7nB3d4e7u0NO\nQewcg5wYw50HIq/QoFvdHcx54ODLpGB3jMpZQXW7wsDgCCDl14ltXh647KI72JshQx7nXvmTAL4G\nwBcS0ccBfB/Ey+YI4CPpBVM3yn8GwH9ARAtEvfvjzPw7j2kIERkt8s29nZf46ZYr14yWNleOpRmU\n9895aSBAZ/PaW3pou/rYUio97fIxNFGPZrJ1PzQebX6fMg49F8zSR6FkOLlTlnGzwVbMjPP5DO8p\nBaMdN0Fb9pn0+v8YykyNrvae1lbz0DgMKfIiY3LJEP7ZkvEMX04e43XzTZ3DP7Jz7c8A+JmXbdSF\ntgDN9tg949y+EbKAtXrhAKIBKnDYfVhbA2Fbfg+kyj/+rd2gpUxaeqq+v+6jThgtyNn22DKU2lGq\nQ7NbXqpHj9uNvGOUfDXKxd/fi4dNtpdkm8eEeZ5xPB7TRjDanm3swl69e7RW1s6b/lvqp1fWoG+G\nDBG5msjYh8QCaAtRFsyAOlLUucIh62bUur1gvpd89qtXTV/LVDCzddhcMm2ahNLWpG1iy7MXgN96\n8lQTB9UrnTZIbG+c2rbYQK32OjuGcrwEJMl4RCzngNPphGVZcH9/rurXcpSPV06eCBUgtxz8JdkD\n/fiIVUxbxgD5IUOuCOgV2C6el2/GOFioCWbOAVPqty1aqd0Nqo6S7aUuULDXfClt1GxPk7YUggUY\noaEu9aXuR7WvB4mXi4wLXQR3W2ZLJ63rmvvVS+Xb24Uq7w4VI9Y14Hxesvvk+ay2Cp/vlyRlBxP5\nasenZAu1fX0sADNzpvKyzaEzDj1a67HUzpAh73a5EqCvNdxLHL1ytTUfj8yrA5ICoVxvX/LCL4cQ\ncuIyBUHnXAZGSaNL1V6wlqO2/D6wBbACoPv+3vZ6uTJNDomqUf8Xu3Jo71FptfbWLtGzI7THSv84\nu0wui7hRqitlzIbcksOmNr7WE+J2pVNPiHtyacXyWG54S6cNGXKbchVAL0pbTXdsr6k1+hYklD7Q\n72XZblMJl8ApGywVA+Pu7q6ifrQsu4lJm0yrR7m09UsK38a7ZtPGdF8D9DD3tTnWe2Njy7aaut2U\nxd5jJwcF+RjXtGHImsZKaC69Vjdin6apcqHU397PuRyby6Ztkx7rPedLE0C3753LB7i/dTJWRO98\nuQqgBwrQP/SPylIZtVbf52NjhAHnmOkMqyWvU8Arr7yCaZo2mrFNIWx5e/3dGmtr46+Cdml7D8h6\nxlVLU1gtdo/6sEbkHsjP81wZoG3ZQJkAmQXcFeiXZa0mO0r2DO89jsdjpnAAJMCf09jq9ovajssZ\nMx8DzA9NcD0ZgD9kyJUAPSFRHdmJQl5OC1jKuQOElMamogmssVIpBi3DepwcDgdM05TpCADwbsrg\nr+D17JmEByiQKdXT0j3FO6XUp/y6cPQ1B26NxpHX3HbZeCQlagspbYB3kKAugnOUwVcnkdZtsjVK\n6+8YI54/f15x98yScvj+/j6373Q6AYg4n6Vdki5ixbpIUrh5lp2h9E8nPDHAInnjFDuJtIHzZ4wR\n5/M5r5j0GjuZ9YDZp5z+miCupaXaSOpLq58hQ25RrgLo1W3usRp9CwV6fw8k9GWXiSJsJgDV8D/z\nmc9kuuZwOGTjbkXDmPJbrblXr22zva6nnTOLv3o6sKFw9rxk9urr1d96y1gjqU4aIcQ8cUmisgjA\nZV6+eNUUkJcUycXAXdFwVNtR9nLnt+O710/bp/zXOfeYMoYMuRW5CqAH9GVPny9wDyBud73jFljm\neUYIBXCtVw0z4/7+fpNSty1Hr2016Uvcczs51DRMs7+rgSwi6jns5HN7GqsFz2L7QNVXBfh1XfLK\nhYiSC6V41oQQ0oqHcDw8yasZDYbS3wr0zKGaOOykbVc49U5Zpq9mcrpkiNXPKsndhWuHDBlyRUAP\nPGyIs9ep2GV8KwLCBewVnACrwQYQDJ0Si0ZrQV3pH0u1tN/btinVZH9XbU2ApXUrPRVZUh7IZfv8\n/CXDbBqB6rj2SfsIcKZuiCh51pyxrgVENdpV+6lArxudtGNfpYom4yKKelXS9ql9bq3oxNWO455G\nP7T4IUOKXBXQv1m5RNk4J1qkcuwZTA3PTnDZELssS77X+uUrF6/19Qyke+2xk4ClgdqNOuZpzhx9\nZE7b7MUcONX2rTcONfDWAVo2EIrTZGLdR1V0UpymCYfDEa88eSVfo9y6nSjlXDHIWlDW2AU7YQE1\nj94+t3Yip53Jcm/8B8gPGVLLW5GPfsiQIUOGXLFclUb/WE3sMdcVXh2GuoGhLUxZKezful0CRTu1\nHiLWg8dq560muk/lbKNsraZv6ZDInN0T27pasVq71eitJq8ulEWzDxvvonU9g0i1+QOOx2NlnLYb\nflvN2hpba61bWqLt0nparf0xz7enyRMR2PXvH543Q4aIXA3Q1y6JjzOktZx177im4hW3yNoNUeka\njuWeOs9LyOCkhlqg7FVreeqetBx9j5+2AK7uk3peAFXdB2uqozdZWA5ejpUJTMvVLf+kf2tyqUTF\nvSNRWdpnrUspHdvmcqydaBq7QhoHdU9tx6Aes52x7EwQzjmw628gM4B+yBCRqwB6Qno5H6mp62WX\nPE5aTltAM5rvxpMmOYq0vLF6ntiIWQUqradXXwvwth1WI3W+ToG8rEs2LpKj5GZZvHH2vFJsmVbr\njbFk6tRgKQV9LU9XKnd3d5jnGXd3B8SIFOXqQST+/XbyUan94Ut6iDSa3Wel9fVy3/RWRrri6o1l\nflY7XP8A+iFDRK4C6CVSSIKXBIAB17zwuuSPQXY5IvKICYC8PyAwZ0DQzakDM3gNiAx4P+Pp0yPe\n+95XEWPEs9M9nj17hrAEwAG8LvATg5kQOSSwrYFT870ABfQ1/7qNEM3gDUYIZ8Rgj8vm2Y4mxAgs\na4Rz4qLop3uZfCCpgSMxnJvAEQhhhaNDpl5sYFTO3xOfZ7dJG8Sl7VqWBSRqPtYgE9j7P//zAEjs\ngHMOB90z1lHaDnDK7WZmTK7EF6wcwXxGSAG3zIxZJziXtH6dnNLmLssiz9mmddCVEnOsJj79TKZc\nMElUtPceumaIYHBFVwE8ubyBS0gU1ZAhtyxXAvRFswSSBwdvNVfVEIUGkEAeoKPhIuTUB5YGsr7X\nACq+WNwZS+g+UdHUW08ZW+fpdMqbjpc2pj9Xa/uW0mDdYikdF004lU2JcuEI5oAQlYKJFdDbNAwA\nEKLmp1mq6GCVEAIEKym7mlrPImn3dm9aaX+9n61MLCGDc44tiEkr7zzljTfNjt1hz4sqfau0/q2n\n0VaGRj/k1uUxO0z9KGRv2E8x8x9Mx/4MgO8A8Nvpsu9l5p9L574HwLcDCAD+LWb+aw83w/pYi/84\nR94AlYAnw7GmEKjznJeXPwIpytSh5t+ZC21hOfYITdFbjIsw97ZgYqNqAeRAK+t6SE7dA61zk6xG\nGOqqSYhxQQgOMS7SJu8yZSFt0XKnqg2tu+Sy3lcafTtRERF8btuUbQ9AAXqv7UK7aYvTR9UYW1vj\nqNJwnafcTHo9WqU1Yut1JUiNqufZTt723iFDhog8RqP/MQB/HsBPNMd/iJl/wB4goj8A4BsBfAWA\n3wPgrxPR72dF1x1RwMzGRwAcYvPCyi5GApxaH4Pywh4FXMxt5ADnvJkM5LgN5w8hwHEyzLImNatB\nSYHTTiqqTeuf5Z+ZGUh9caR5XxQ0I3yaBHQFIfdLlsjpMANMya8/ILIGKxV/fq3Dfl/W+9TeWAFp\ntS0iUgCZxyaADAB0XcWoUxXYPudyQWDeRrna74xCvzT/VjZa/CXDdrl+e1/vnr2JZMiQW5THbCX4\nC0T0ZY8s7+sB/BQznwD8XSL6NQBfDeB/fdGG9V5QmS8EhFuVcXO9UsWujlwlSnSNa6iI7DKoVIiG\nYdZgbpOZEVG1h6pNkxxjRIhrLgOwWv+M4AIowarmgHfujGVZ4NcFBIf75Yz7+zOWVbnsw8aYaicf\nUL07ls1Jk39zsi9M/QRomUAjvcfnCVJXOqV8AnOrVTcgbL63z6s1TveepQVsGfd+ea0MgB8ypMjL\ncPTfSUTfDOCjAL6LmX8XwIcA/KK55uPp2EUpYIRCEVB/P9UYOW8pB+prk0Ujh9H+xbjInLRiFAMp\nM4MSSDun/H2qMyJvRmLphMyLG08TC7pW01YpRtQVRGY3JnapvLQX69nBkcc5rLJH6xIS0BVKRsHb\n9tt7jRcoWrilPerVUO2RUjT+bNGuJoIWV4X+EeCt+tmdoDk/iN7zqiYr0zaVuh01XWQn8aptFyaS\nIUNuTd4s0P8wgO+HqL3fD+AHAXzbixRARK8BeA0A3v95r6aX0qSXdQTL+OjLGmOhMpA8MnW7ugoc\nfAJmLqkGvBewsIZK5eod6/6wusmIlBPWWAGQaX/VLssb62cLMHZ3qxDuAZQcPEQEjmmTDzAmP2f/\ncKuhq7RAT0RGS6/97W2uH0ROtFWpV6/z3oNCMgijpUVaw7FcZTX6ajLMVxS5BOZ2XHtJ4ixoW9vN\n3vW2zgH0Q25d3hTQM/P/3975+0hyVHH8862eWQeIAHT4ZNmWcXCJSRwgIoIjARMdJMiOjOSAwPwB\nRkg4skREBkKHsOzEti6xODkAGSeEtpGQuLOwOIERdwIfiAwhz27XI6iq7uqenv0l3/Xs7PtIq+2u\n6al+3a3+Vs2rV68+LtuSfgG8lXfvAI9Whz6Sy6bquApcBfjiww9aiTopLpDkq6+FoM5WGDEMUVwl\nXa3Fpn4hvs7dki53LNBdT9aKa6B8nv4fKIlaLS7Dwcf1xT7G+dDLOUsvv0xiijFmd8+yy8ezWq04\niC3LRUSLhhj7LJTj845Fs08BrDUBL7lriutmsSwLpvd1hhC6FbH6hrU0dKnO2Nb3IXZCX9+L0UMZ\n3LOasU9+PA5Ql4+Ffnx/p453HCdxKqGX9JCZ/SPvfhu4kbevA69J+glpMPYS8O5R9aWXvQ/TK4Oi\n42NK2GPIPXiLB2szLbN9nQsoCXC/IEn5fG9vr5v9ulqtWIgusqXMojWzbnboarWiaZpOoOuGothX\nyuoFNYqwl/22bTEiCrBoGsBY7X+CrYyAWC6XLMNeGhsQLBYNir2LqIh2ySRZIofatqWN/8vn75f3\nGzdspTw06786Yoxr8xfGAhpC7+qJYtAIFV9+OmG+N/kemIZ1TaWiqAe4a+pZymZ0YyGDe1rZXO5V\nN+4wWkbRcc4bxwmvfB24DFyQdBt4Ebgs6UlSF/oj4HsAZnZT0jXgA+AAeP6oiJt8lm4rvezrvb9h\nDngwSxNihr3wUY9eEdrpnnipsxPnTuiLOyMLt/UpFGr3zNhXbDksFDtCKBvRx7YU0ncD2YWj3iee\nGq1+Gb8i1KWH3q/kJBr2Oh/9eAWn+h6W61t7ChvGPOr9cc78+nuSuoZC5VzlOA175selHhcpQj92\nAdW/TIDBPXEc53hRN89MFP/ykONfAl46mRnDmOmOavKOQlqQJEWqlBjJnCIglnPnYwVimL98ZCPF\nFVF6xw1lxmcYCH1srfPjl8HU8QIe6Qoq8Y/9eWphSz8ySv11vpeUQiCQYtpbi0Qsi2PIg55plmxZ\nzm8wSSmfKzS9j378WXEv9eWbV6yqXV+DsokGovaRl+8lYe8Hf8ts3/Hg8dSzmWoIho1qdT9H7qtC\nP9P26MlUjnMe2I6ZsZWmdCGUxsB90zUCiojh6k+Hvcb1S77WkFD1Skdl6VfDsPdahy7WQp+iSsJA\n6Gtx7fzdpkoYi9BX/mbLSdMs0h5ELIgQFpBDQovQlwRkY/sWy9KTHd1eVeGdR4WlMhTsWkiTu2aY\nmbJ2uQzuuU0PXk/55Ws2iXJ/3HQjUddTP5/D6rxP/Bv425wGODvNY8c5aCuEvusFVpgZsRL65354\n9X6b5ZxR6jGRqein+4mZfWG2kztOZiuEvlB60slP3faLZTvOCahdNu6nd5wtFPouNDC7K37+a3xB\nmgAABF9JREFUo+9iFmmWizzoBm0JZ8wRILEdL7pd5XTJjUWJwugGVDV05TQUf3uJV+9TGB82gFin\nO+hssOn0uNH6ZffSX3JFFRdV3M+LejQh1bHI+W2yLU2zHIQf1hkgm6YBrbINw9j2OgLH2vwZ7Vps\nPkCI+/nXVLG/RC+lOvbzwuExRmhCF4XTRfVkH3qJsmnLOEboffmbQi3rz6Z893lr8phuMDiElKGz\nKnOc887WCH3/ouaX0+p46j5GPEYb5K/vRNgsqfeIOl3B0Oc8ff5htItyyuOhuNSi1G3XMfwKa+cD\nkNWhpClBm9SHcg6iV0ID5ZqrhqOOGBr7z411P/aUHalhWr/+w3znm0RzPBgayqpWE4vIjMdLxufZ\nNBA7ptyrwyZK1dftYu+cd7ZC6NO73M9MLZEdoVmyUEO0g5T6oI3dQKFFy2GEezSK7B/sE2OfSbLJ\ns0qj7adJQiVkM+bkXd3LX7IyBgJNinTJ9RNTDvSUPbJkq2wxyymOo3X5bJqmxO6n3jj0whWtX4yj\nWcSBoImGRbOXTFDpbacoHkwp9l+iROaUXvbQNRGJ0Thg1fXkF9nmIBFkhNjmHnZLkyejyejCTxUC\nTQhY2Ev5/ukbk3Sulv39FmQ0C2jKgPgoerbztmWNT/MFusQKlFxC40alnql8mDiXRnSqQSj1LpfL\n7vt1Xn7HOa9oG0LPJP0L+C8pQmEXuIBfy7bxmA+MOueVrRB6AEnvm9mX57bj08CvxXGcbSIcfYjj\nOI5zlnGhdxzH2XG2Seh3aUaUX4vjOFvD1vjoHcdxnHvDNvXoHcdxnHvA7EIv6SlJH0q6JemFue05\nKZI+kvRHSX+Q9H4u+7yktyX9Of//3Nx2bkLSy5LuSrpRlW20X9IP8rP6UNI35rHacZyTMKvQK80C\n+inwTeAJ4BlJT8xp0yn5mpk9WYUhvgC8Y2aXgHfy/rbyCvDUqGzS/vxsnga+lL/zs/wMHcfZYubu\n0X8FuGVmfzGzFfAGcGVmmz4NrgCv5u1XgW/NaMuhmNnvgP+MijfZfwV4w8w+MbO/ArdIz9BxnC1m\nbqF/GPh7tX87l50lDPitpN8rLXgOcNH6pRb/CVycx7RTs8n+XXhejnPu2IpcN2ecr5rZHUkPAm9L\n+lP9oZmZptZGPCOcdfsdx5m/R38HeLTafySXnRnM7E7+fxd4k+TK+FjSQwD5/935LDwVm+w/88/L\ncc4jcwv9e8AlSY9L2iMN9F2f2aZjI+kzkj5btoGvAzdI1/BsPuxZ4FfzWHhqNtl/HXha0gOSHgcu\nAe/OYJ/jOCdgVteNmR1I+j7wG6ABXjazm3PadEIuAm/mlLgL4DUz+7Wk94Brkp4jrRf6nRltPBRJ\nrwOXgQuSbgMvAj9mwn4zuynpGvABcAA8bzbKU+w4ztbhM2Mdx3F2nLldN47jOM49xoXecRxnx3Gh\ndxzH2XFc6B3HcXYcF3rHcZwdx4XecRxnx3GhdxzH2XFc6B3HcXac/wNPAbxSsOLOKQAAAABJRU5E\nrkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x10eef2e8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Load an image\n",
    "img = load_img(4, 1038)\n",
    "\n",
    "try:\n",
    "    \n",
    "    # Resize the image\n",
    "    img1, box = mnist_resize(img)\n",
    "    #img1 = edge_detect_and_resize(img)\n",
    "    \n",
    "    # create figure and axes\n",
    "    ax = plt.subplot(1, 2, 1)\n",
    "    \n",
    "    # Display the image\n",
    "    ax.imshow(img)\n",
    "    \n",
    "    # TODO:  Plot the original image, img, along with a red box around the captured circuit component.\n",
    "    # Use the mpatches.Rectangle and ax.add_patch methods to construct the rectangle.\n",
    "    minr, minc, maxr, maxc = box\n",
    "    rect = mpatches.Rectangle((minc, minr), maxc - minc, maxr - minr, fill=False, edgecolor='red', linewidth=2)\n",
    "    ax.add_patch(rect)\n",
    "    \n",
    "    # TODO:  Plot the resized 28 x 28 image, img1.  You can use the plt_circuit(img1) command \n",
    "    plt.subplot(1, 2, 2)\n",
    "    plt_circuit(img1)\n",
    "    \n",
    "except ImgException as e:\n",
    "    print(e.msg)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Step 3: Randomly Capture and Appropriately Label 8000 Images"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "images captured = 500\n",
      "images captured = 1000\n",
      "images captured = 1500\n",
      "images captured = 2000\n",
      "images captured = 2500\n",
      "images captured = 3000\n",
      "images captured = 3500\n",
      "images captured = 4000\n",
      "images captured = 4500\n",
      "images captured = 5000\n",
      "images captured = 5500\n",
      "images captured = 6000\n",
      "images captured = 6500\n",
      "images captured = 7000\n",
      "images captured = 7500\n",
      "images captured = 8000\n"
     ]
    }
   ],
   "source": [
    "# Dimensions\n",
    "ncirc = 8000   # 5 different circuit components * 1600 samples for each component\n",
    "nrow = 28\n",
    "ncol = 28\n",
    "npix = nrow*ncol   # npix = 28*28 = 784\n",
    "xcirc = np.zeros((ncirc, npix))\n",
    "ylabels = np.array([]) \n",
    "\n",
    "i = 0\n",
    "\n",
    "while i < ncirc:\n",
    "    \n",
    "    # TODO:  Generate a random circuit and sample    \n",
    "    circ_ind = np.random.randint(1, 6)  # Get a random circuit component\n",
    "    ylabels = np.append(ylabels, (circ_ind - 1)) # Get the labels for each circuit component\n",
    "    \n",
    "    # samp_ind = random number from 1 to 1600\n",
    "    samp_ind = np.random.randint(1, 1600) #Each circuit component has 1600 samples to choose from\n",
    "        \n",
    "    try:\n",
    "        # TODO:  Load the image with load_img function\n",
    "        img = load_img(circ_ind, samp_ind)\n",
    "        \n",
    "        # TODO:  Reize the image with mnist_resize function\n",
    "        img1, box = mnist_resize(img)\n",
    "\n",
    "        # TODO:  Store the image in a row of xcirc[i,:] and increment i\n",
    "        img2 = skimage.transform.resize(img1, (npix, 1), mode = 'constant') # Resize the img from (28, 28) to (784, 1)\n",
    "        img2 = np.squeeze(img2) # Change image shape from (784, 1) to (784)\n",
    "        xcirc[i, :] = img2\n",
    "        i += 1\n",
    "        \n",
    "        # Print progress\n",
    "        if (i % 500 == 0):\n",
    "            print ('images captured = {0:d}'.format(i))\n",
    "    except ImgException:\n",
    "        # Skip if image loading or resizing failed\n",
    "        pass"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# since this takes a long time to generate, save the matrix xcirc to a file xcirc.p using the pickle.dump command. \n",
    "\n",
    "import pickle\n",
    "with open(\"xcirc.p\", \"wb\") as fp:\n",
    "    pickle.dump(xcirc, fp)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Reload the data xcirc from the file xcirc.p\n",
    "\n",
    "import pickle\n",
    "with open( \"xcirc.p\", \"rb\" ) as fp:\n",
    "    xcirc = pickle.load(fp)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Step 4: Rescale the Data and Create Training and Testing Sets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Rescale the circuit components xcirc to a new matrix xcirc = 2*xcircs-1 to make the pixel values go from -1 to 1.  \n",
    "xcircs = 2*xcirc - 1\n",
    "\n",
    "# Reassign the variables for convience\n",
    "x = xcircs\n",
    "y = ylabels\n",
    "\n",
    "# Get 7000 training samples Xtr,ytr and 1000 test samples Xts,yts. Remember to randomly select them.\n",
    "import random\n",
    "rand = random.sample(range(8000), 8000)\n",
    "tr_index = rand[0:7000]\n",
    "ts_index = rand[7000:8000]\n",
    "\n",
    "# TODO\n",
    "xtr = x[tr_index, :]\n",
    "ytr = y[tr_index]\n",
    "xts = x[ts_index, :]\n",
    "yts = y[ts_index]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Step 5: Create the Neural Network and its Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from keras.models import Model, Sequential\n",
    "from keras.layers import Dense, Activation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# TODO clear session\n",
    "import keras.backend as K\n",
    "K.clear_session()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# TODO: construct the model\n",
    "nin = x.shape[1]           # dimension of input data\n",
    "nh = 306                   # number of hidden units\n",
    "nout = int(np.max(y)+1)    # number of outputs since ys = 0,1,...,nclasses-1\n",
    "model = Sequential()\n",
    "model.add(Dense(nh, input_shape=(nin,), activation='sigmoid', name='hidden'))\n",
    "model.add(Dense(nout, activation='softmax', name='output'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "hidden (Dense)               (None, 306)               240210    \n",
      "_________________________________________________________________\n",
      "output (Dense)               (None, 5)                 1535      \n",
      "=================================================================\n",
      "Total params: 241,745\n",
      "Trainable params: 241,745\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "# TODO:  Print the model summary\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "# A callback is a class that is passed to the fit method. \n",
    "# Complete the LoadHistory callback class below to save the loss and validation accuracy.\n",
    "\n",
    "class LossHistory(keras.callbacks.Callback):\n",
    "    def on_train_begin(self, logs={}):\n",
    "        # TODO:  Create two empty lists, self.loss and self.val_acc\n",
    "        self.loss = np.array([])\n",
    "        self.val_acc = np.array([])\n",
    "        self.prt_period = 10\n",
    "\n",
    "    def on_batch_end(self, batch, logs={}):\n",
    "        # TODO:  This is called at the end of each batch.  \n",
    "        # Add the loss in logs.get('loss') to the loss list\n",
    "        self.loss = np.append(self.loss, logs.get('loss'))\n",
    "        \n",
    "    def on_epoch_end(self, epoch, logs):\n",
    "        # TODO:  This is called at the end of each epoch.  \n",
    "        # Add the test accuracy in logs.get('val_acc') to the val_acc list\n",
    "        v = logs.get('val_acc')\n",
    "        self.val_acc = np.append(self.val_acc, v)\n",
    "        if (epoch % self.prt_period == 0):\n",
    "            print('epoch = %d val_acc=%12.4e' % (epoch, v))\n",
    "        \n",
    "# Create an instance of the history callback\n",
    "history_cb = LossHistory()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#  Step 6: Test the Neural Network on the Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Create an optimizer and compile the model. Select the appropriate loss function and metrics. \n",
    "# For the optimizer, use the Adam optimizer with a learning rate of 0.001\n",
    "\n",
    "from keras import optimizers\n",
    "\n",
    "# Since this is a multi-class classification problem, we select the sparse_categorial_crossentropy loss.\n",
    "# It is sparse because the labels vector y is of shape [batch_size] instead of shape [batch_size, num_classes]\n",
    "opt = optimizers.Adam(lr=0.001) # beta_1=0.9, beta_2=0.999, epsilon=1e-08, decay=0.0)\n",
    "model.compile(optimizer=opt,\n",
    "              loss='sparse_categorical_crossentropy', \n",
    "              metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch = 0 val_acc=  5.7200e-01\n",
      "epoch = 10 val_acc=  6.6700e-01\n",
      "epoch = 20 val_acc=  7.1300e-01\n",
      "epoch = 30 val_acc=  7.2800e-01\n",
      "epoch = 40 val_acc=  7.5900e-01\n",
      "epoch = 50 val_acc=  7.6800e-01\n",
      "epoch = 60 val_acc=  7.8000e-01\n",
      "epoch = 70 val_acc=  7.8300e-01\n",
      "epoch = 80 val_acc=  7.9900e-01\n",
      "epoch = 90 val_acc=  7.8700e-01\n",
      "epoch = 100 val_acc=  7.9200e-01\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x10ec0e80>"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Fit the model for 10 epochs using the scaled data for both the training and validation.  \n",
    "# Use the validation_data option to pass the test data. \n",
    "# Also, pass the callback class created above. Use a batch size of 100. \n",
    "# Your final accuracy should be >99%. \n",
    "batch_size = 100\n",
    "model.fit(xtr, ytr, epochs=110, batch_size=batch_size, \n",
    "          validation_data=(xts, yts), callbacks=[history_cb],verbose=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYwAAAEKCAYAAAAB0GKPAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x10e293c8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot the validation accuracy saved in the history_cb. This gives one accuracy value per epoch. \n",
    "plt.plot( history_cb.val_acc, '-')\n",
    "plt.xlabel('Epoch')\n",
    "plt.ylabel('Validation Accuracy')\n",
    "plt.grid()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Step 7: Create a Support Vector Machine (SVM) and Test it on the Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# First create the SVM classifer. Use an \"rbf\" classifier with C=2.8 and gamma=.0073.\n",
    "\n",
    "from sklearn import svm\n",
    "\n",
    "# TODO:  Create a classifier: a support vector classifier\n",
    "svc = svm.SVC(probability=False,  kernel=\"rbf\", C=2.8, gamma=.0073, verbose=10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LibSVM]"
     ]
    },
    {
     "data": {
      "text/plain": [
       "SVC(C=2.8, cache_size=200, class_weight=None, coef0=0.0,\n",
       "  decision_function_shape='ovr', degree=3, gamma=0.0073, kernel='rbf',\n",
       "  max_iter=-1, probability=False, random_state=None, shrinking=True,\n",
       "  tol=0.001, verbose=10)"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Use the svc.fit command to fit on the training data.\n",
    "svc.fit(xtr,ytr)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuaracy = 0.773000\n"
     ]
    }
   ],
   "source": [
    "# Measure the accuracy on the test samples. You should get about 96% accuracy.\n",
    "yhat_ts = svc.predict(xts)\n",
    "acc = np.mean(yhat_ts == yts)\n",
    "print('Accuaracy = {0:f}'.format(acc))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "# Evaluation \n",
    "\n",
    "As can be seen above, both the Convolutional Neural Network and the Support Vector Machine were unable to obtain an accuracy greater than 80 percent! \n",
    "\n",
    "This may be do to several factors such as poor data due to inaequate pictures of the circuit components and poor thresholding of the data images!  \n",
    "\n",
    "Ultimately, the this project could be improved by maybe finding a better way to threshold the images to black and white, resizing of the images differently to add resolution, and possibly scaling the data more efficiently.  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "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": 2
}