Code Finalize-checkpoint.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Data Preprocessing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Python Standard Libraries for importing data from binary file\n",
    "import os.path #for accessing the file path\n",
    "import struct  #for unpacking the binary data\n",
    "\n",
    "import time    #for calculating time\n",
    "\n",
    "#core packages\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Loading data from binary file downloaded from LuCan's site"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "def load_data(dataset=\"training\"):\n",
    "    \"\"\"\n",
    "        Loads MNIST datasate from  the binary file into 3D numpy arrays        \n",
    "        \n",
    "        Dataset Obtained From:\n",
    "            http://yann.lecun.com/exdb/mnist/\n",
    "        Dataset retrival code adapted from(but modified to our need):\n",
    "            https://www.cs.virginia.edu/~connelly/class/2015/large_scale/proj2/mnist_python\n",
    "            \n",
    "        Argument:\n",
    "            dataset - type of dataset to be loaded. may be either 'training' or 'testing'\n",
    "        Returns:\n",
    "            images - 3D array consisting of no. of examples, rows, columns of images\n",
    "            labels - array  containing labels for each images\n",
    "    \"\"\"\n",
    "    digits = np.arange(10)\n",
    "    path = \"dataset/\"\n",
    "    size = 60000\n",
    "    \n",
    "    #setting file path based on the dataset\n",
    "    if dataset == \"training\":\n",
    "        img_file_path = os.path.join(path, 'train-images-idx3-ubyte')\n",
    "        lbl_file_path = os.path.join(path, 'train-labels-idx1-ubyte')\n",
    "    elif dataset == \"testing\":\n",
    "        img_file_path = os.path.join(path, 't10k-images-idx3-ubyte')\n",
    "        lbl_file_path = os.path.join(path, 't10k-labels-idx1-ubyte')\n",
    "    else:\n",
    "        raise ValueError(\"Dataset must be 'testing' or 'training'\")\n",
    "    \n",
    "    #retriving the data\n",
    "    with open(lbl_file_path, 'rb') as flbl:\n",
    "        _, size = struct.unpack(\">II\", flbl.read(8))\n",
    "        labels = np.frombuffer(flbl.read(), dtype=np.int8).reshape(1,size)\n",
    "\n",
    "    with open(img_file_path, 'rb') as fimg:\n",
    "        _, _, rows, cols = struct.unpack(\">IIII\", fimg.read(16))\n",
    "        images = np.frombuffer(fimg.read(),dtype=np.uint8).reshape(size, rows, cols)\n",
    "\n",
    "    return images, labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Time to load training and test set from binary file using numpy: 73.27628135681152ms\n"
     ]
    }
   ],
   "source": [
    "toc = time.time()\n",
    "#loading the original data\n",
    "training_x_orig, training_y_orig = load_data(dataset=\"training\")\n",
    "testing_x_orig, testing_y_orig = load_data(dataset=\"testing\")\n",
    "\n",
    "tic = time.time()\n",
    "print(\"Time to load training and test set from binary file using numpy: \" + str(1000*(tic-toc)) + \"ms\")\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Confirming the shapes of the original data in the dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Data Types of the obtained data: \n",
      "==================================\n",
      "Training Set Input Type:  <class 'numpy.ndarray'>\n",
      "Training Set Output Type: <class 'numpy.ndarray'>\n",
      "Test Set Input Type:      <class 'numpy.ndarray'>\n",
      "Test Set Output Type:     <class 'numpy.ndarray'>\n",
      "\n",
      "Shape of the obtained data: \n",
      "==================================\n",
      "Training Set Input Shape:  (60000, 28, 28)\n",
      "Training Set Output Shape: (1, 60000)\n",
      "Test Set Input Shape:      (10000, 28, 28)\n",
      "Test Set Output Shape:     (1, 10000)\n"
     ]
    }
   ],
   "source": [
    "print(\"Data Types of the obtained data: \")\n",
    "print(\"==================================\")\n",
    "print(\"Training Set Input Type:  \" + str(type(training_x_orig)))\n",
    "print(\"Training Set Output Type: \" + str(type(training_y_orig)))\n",
    "print(\"Test Set Input Type:      \" + str(type(testing_x_orig)))\n",
    "print(\"Test Set Output Type:     \" + str(type(testing_y_orig)))\n",
    "\n",
    "print(\"\\nShape of the obtained data: \")\n",
    "print(\"==================================\")\n",
    "print(\"Training Set Input Shape:  \" + str(training_x_orig.shape))\n",
    "print(\"Training Set Output Shape: \" + str(training_y_orig.shape))\n",
    "print(\"Test Set Input Shape:      \" + str(testing_x_orig.shape))\n",
    "print(\"Test Set Output Shape:     \" + str(testing_y_orig.shape))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Visualizing Some of the sample digits from the dataset with their label"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "def visualize_orig_sample(x_orig, y_orig):\n",
    "    fig, axes = plt.subplots(nrows=2, ncols=5,figsize=(16,8))\n",
    "    fig.subplots_adjust(hspace=.2)\n",
    "    fig.suptitle(\"Sample Digits\")\n",
    "\n",
    "    for ax,i in zip(axes.flatten(),range(11040,11050)):\n",
    "        ax.imshow(x_orig[i].squeeze(),cmap=\"Greys\")\n",
    "        ax.set(title = \"Label: \"+ str(y_orig[0,i]))\n",
    "        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1152x576 with 10 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "visualize_orig_sample(training_x_orig,training_y_orig)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Preparing Input by flattenning/reshaping and normalizing the data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "def prepare_input(training_x_orig,testing_x_orig):\n",
    "    \"\"\"\n",
    "        Flattens the 3D numpy array of the data and normalizes the data in the range between 0 and 1\n",
    "        Arguement:\n",
    "            training_x_orig - image data loaded directly from training set file\n",
    "            testing_x_orig - image data loaded directly from test set file\n",
    "        Returns:\n",
    "            train_x -flattened and normalized training set input data\n",
    "            test_x - flattened and normalized test set input data\n",
    "            \n",
    "    \"\"\"\n",
    "    \n",
    "    train_x_flatten = training_x_orig.reshape(training_x_orig.shape[0], -1).T   # The \"-1\" makes reshape flatten the remaining dimensions\n",
    "    test_x_flatten = testing_x_orig.reshape(testing_x_orig.shape[0], -1).T\n",
    "\n",
    "    # Normalizing the data into the range between 0 and 1.\n",
    "    train_x_norm = train_x_flatten/255.\n",
    "    test_x_norm = test_x_flatten/255.\n",
    "    \n",
    "    assert(train_x_norm.shape == (784,60000) )\n",
    "    assert(test_x_norm.shape == (784,10000) )\n",
    "    \n",
    "    return train_x_norm,test_x_norm\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape of Input Training set: (784, 60000)\n",
      "Shape of Input Test set:     (784, 10000)\n"
     ]
    }
   ],
   "source": [
    "train_x_norm,test_x_norm = prepare_input(training_x_orig,testing_x_orig)\n",
    "print (\"Shape of Input Training set: \" + str(train_x_norm.shape))\n",
    "print (\"Shape of Input Test set:     \" + str(test_x_norm.shape))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Preparing output by transforming the labels into one-hot representation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "#encoding method 1..... \n",
    "def one_hot_encoding(raw_y):\n",
    "    \"\"\"\n",
    "    \n",
    "    \"\"\"\n",
    "    classes = np.arange(10).reshape(1,10)\n",
    "\n",
    "    encoded_y = np.array(classes == raw_y[0,0]).astype(int)\n",
    "    for i in range(1,raw_y.shape[1]):\n",
    "        temp = np.array(classes == raw_y[0,i]).astype(int) \n",
    "        encoded_y = np.concatenate((encoded_y,temp),axis=0)\n",
    "    assert(encoded_y.T.shape == (10,raw_y.shape[1]))\n",
    "    \n",
    "    return encoded_y.T"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape of Output Training set: (10, 60000)\n",
      "Shape of Output Test set:     (10, 10000)\n",
      "Time to encode: 14581.954002380371 ms\n"
     ]
    }
   ],
   "source": [
    "toc = time.time()\n",
    "atrain_y = one_hot_encoding(training_y_orig)\n",
    "atest_y = one_hot_encoding(testing_y_orig)\n",
    "tic = time.time()\n",
    "print (\"Shape of Output Training set: \" + str(atrain_y.shape))\n",
    "print (\"Shape of Output Test set:     \" + str(atest_y.shape))\n",
    "print(\"Time to encode: \" + str(1000*(tic-toc)) + \" ms\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "this seems to be taking a lot of time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[0 1 0 0 0 0 0 0 0 0]\n",
      " [0 0 0 1 0 0 1 0 1 0]\n",
      " [0 0 0 0 0 1 0 0 0 0]\n",
      " [0 0 0 0 0 0 0 1 0 0]\n",
      " [0 0 1 0 0 0 0 0 0 1]\n",
      " [1 0 0 0 0 0 0 0 0 0]\n",
      " [0 0 0 0 0 0 0 0 0 0]\n",
      " [0 0 0 0 0 0 0 0 0 0]\n",
      " [0 0 0 0 0 0 0 0 0 0]\n",
      " [0 0 0 0 1 0 0 0 0 0]] [5 0 4 1 9 2 1 3 1 4]\n",
      "<class 'numpy.int64'>\n",
      "[[0 0 0 1 0 0 0 0 0 0]\n",
      " [0 0 1 0 0 1 0 0 0 0]\n",
      " [0 1 0 0 0 0 0 0 0 0]\n",
      " [0 0 0 0 0 0 0 0 0 0]\n",
      " [0 0 0 0 1 0 1 0 0 0]\n",
      " [0 0 0 0 0 0 0 0 1 0]\n",
      " [0 0 0 0 0 0 0 0 0 0]\n",
      " [1 0 0 0 0 0 0 0 0 0]\n",
      " [0 0 0 0 0 0 0 0 0 0]\n",
      " [0 0 0 0 0 0 0 1 0 1]] [7 2 1 0 4 1 4 9 5 9]\n",
      "<class 'numpy.int64'>\n"
     ]
    }
   ],
   "source": [
    "print(atrain_y[:,0:10],training_y_orig[0,0:10])\n",
    "print(type(atrain_y[0,0]))\n",
    "print(atest_y[:,0:10],testing_y_orig[0,0:10])\n",
    "print(type(atest_y[0,0]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "#encoding method 2....\n",
    "def one_hot_encoding_two(raw_y,no_of_classes = 10):\n",
    "    \"\"\"\n",
    "        Transform the output labels into the one-hot encoding representation\n",
    "        Arguments:\n",
    "            raw_y - raw labels loaded directly from the binary file\n",
    "            no_of_classes - number of the classes based on which the transformation is to be made\n",
    "        Returns:\n",
    "            y_out - encoded ndarray of the labels with data elements of int type\n",
    "    \"\"\"\n",
    "    classes = np.arange(no_of_classes).reshape(no_of_classes,1)\n",
    "    temp = np.zeros((classes.shape[0],raw_y.shape[1]))\n",
    "    y_out = temp + classes\n",
    "    \n",
    "    for i in range(raw_y.shape[1]):\n",
    "        y_out[:,i] = np.equal(y_out[:,i],raw_y[0,i])\n",
    "\n",
    "    assert(y_out.shape == (classes.shape[0],raw_y.shape[1]))\n",
    "    return y_out.astype(int)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape of Output Training set: (10, 60000)\n",
      "Shape of Output Test set:     (10, 10000)\n",
      "Time to encode: 191.2252902984619 ms\n"
     ]
    }
   ],
   "source": [
    "toc = time.time()\n",
    "##encoding the output of the training and the test dataset\n",
    "train_y_onehot = one_hot_encoding_two(training_y_orig)\n",
    "test_y_onehot = one_hot_encoding_two(testing_y_orig)\n",
    "\n",
    "tic = time.time()\n",
    "\n",
    "print (\"Shape of Output Training set: \" + str(train_y_onehot.shape))\n",
    "print (\"Shape of Output Test set:     \" + str(test_y_onehot.shape))\n",
    "print(\"Time to encode: \" + str(1000*(tic-toc)) + \" ms\")\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The time taken seems reasonable"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Validating the encoding"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[0 0 0 0 0 0 0 0 0 0]\n",
      " [0 0 0 0 0 0 0 0 0 0]\n",
      " [0 0 0 0 0 0 0 0 0 0]\n",
      " [1 0 0 0 0 0 0 1 0 0]\n",
      " [0 0 0 0 0 1 0 0 0 1]\n",
      " [0 0 0 0 0 0 1 0 0 0]\n",
      " [0 0 1 0 1 0 0 0 0 0]\n",
      " [0 0 0 0 0 0 0 0 0 0]\n",
      " [0 1 0 0 0 0 0 0 1 0]\n",
      " [0 0 0 1 0 0 0 0 0 0]] [3 8 6 9 6 4 5 3 8 4]\n",
      "<class 'numpy.int64'>\n",
      "[[0 0 0 1 0 0 0 0 0 0]\n",
      " [0 0 1 0 0 1 0 0 0 0]\n",
      " [0 1 0 0 0 0 0 0 0 0]\n",
      " [0 0 0 0 0 0 0 0 0 0]\n",
      " [0 0 0 0 1 0 1 0 0 0]\n",
      " [0 0 0 0 0 0 0 0 1 0]\n",
      " [0 0 0 0 0 0 0 0 0 0]\n",
      " [1 0 0 0 0 0 0 0 0 0]\n",
      " [0 0 0 0 0 0 0 0 0 0]\n",
      " [0 0 0 0 0 0 0 1 0 1]] [7 2 1 0 4 1 4 9 5 9]\n",
      "<class 'numpy.int64'>\n"
     ]
    }
   ],
   "source": [
    "print(train_y_onehot[:,50000:50010],training_y_orig[0,50000:50010])\n",
    "print(type(train_y_onehot[0,0]))\n",
    "print(test_y_onehot[:,0:10],testing_y_orig[0,0:10])\n",
    "print(type(test_y_onehot[0,0]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Dev - Test set split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Shape of the Final data: \n",
      "==================================\n",
      "Training Set Input Shape:   Processed(784, 60000), Original(60000, 28, 28)\n",
      "Training Set Output Shape:  Processed(10, 60000) , Original(1, 60000)\n",
      "Dev Set Input Shape:        Processed(784, 5000) , Original(5000, 28, 28)\n",
      "Dev Set Output Shape:       Processed(10, 5000)  , Original(1, 5000)\n",
      "Test Set Input Shape:       Processed(784, 5000) , Original(5000, 28, 28)\n",
      "Test Set Output Shape:      Processed(10, 5000)  , Original(1, 5000)\n"
     ]
    }
   ],
   "source": [
    "train_x = train_x_norm\n",
    "train_y = train_y_onehot\n",
    "train_x_orig = training_x_orig\n",
    "train_y_orig = training_y_orig\n",
    "\n",
    "# def test_dev_split():\n",
    "#original dev set data\n",
    "dev_x_orig = testing_x_orig[0:5000, :, :] #original data is 3d ndarray\n",
    "dev_y_orig = testing_y_orig[:, 0:5000]\n",
    "#processed devset data\n",
    "dev_x = test_x_norm[:, 0:5000]\n",
    "dev_y = test_y_onehot[:, 0:5000]\n",
    "\n",
    "#original test set data \n",
    "test_x_orig = testing_x_orig[5000:10000, :, :] #original data is 3d ndarray\n",
    "test_y_orig = testing_y_orig[:, 5000:10000]\n",
    "#processed test set data\n",
    "test_x = test_x_norm[:, 5000:10000]\n",
    "test_y = test_y_onehot[:, 5000:10000]\n",
    "\n",
    "print(\"\\nShape of the Final data: \")\n",
    "print(\"==================================\")\n",
    "print(\"Training Set Input Shape:  \",\"Processed\"+ str(train_x.shape) + \", Original\"+str(train_x_orig.shape))\n",
    "print(\"Training Set Output Shape: \",\"Processed\" + str(train_y.shape) + \" , Original\"+str(train_y_orig.shape))\n",
    "print(\"Dev Set Input Shape:       \",\"Processed\" + str(dev_x.shape) + \" , Original\"+str(dev_x_orig.shape))\n",
    "print(\"Dev Set Output Shape:      \",\"Processed\" + str(dev_y.shape) + \"  , Original\"+str(dev_y_orig.shape))\n",
    "print(\"Test Set Input Shape:      \",\"Processed\" + str(test_x.shape) + \" , Original\"+str(test_x_orig.shape))\n",
    "print(\"Test Set Output Shape:     \",\"Processed\" + str(test_y.shape) + \"  , Original\"+str(test_y_orig.shape))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "def visualize_processed_sample(x_orig, y_orig, y, encoding = False, dataset = \"training\"):\n",
    "    if(dataset == \"training\"):\n",
    "        visual_title = \"Sample Training Data Set\"\n",
    "    elif(dataset == \"dev\"):\n",
    "        visual_title = \"Sample Dev Data Set\"\n",
    "    elif(dataset == \"test\"):\n",
    "        visual_title = \"Sample Test Data Set\"\n",
    "    else:\n",
    "        raise ValueError(\"Dataset set must be training or dev or test set\")\n",
    "    fig, axes = plt.subplots(nrows=2, ncols=5,figsize=(16,8))\n",
    "    fig.subplots_adjust(hspace=.1)\n",
    "    fig.suptitle(visual_title)\n",
    "\n",
    "    for ax,i in zip(axes.flatten(),range(1040,1050)):\n",
    "        ax.imshow(x_orig[i].squeeze(),cmap=\"Greys\")\n",
    "        ax.set(title = \"Label: \"+ str(y_orig[0,i]))\n",
    "        ax.set_xlim([0, 28])\n",
    "        ax.set_ylim([28, 0])\n",
    "        if(encoding == True):\n",
    "            ax.set(xlabel= \"Encoding: \"+ str(y[:,i]) )\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1152x576 with 10 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "visualize_processed_sample(train_x_orig, train_y_orig, train_y, encoding = True, dataset = \"training\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1152x576 with 10 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "visualize_processed_sample(dev_x_orig, dev_y_orig, dev_y, encoding = True, dataset = \"dev\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA54AAAH/CAYAAAA/uAIxAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjMsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+AADFEAAAgAElEQVR4nOzdebyc8/n/8feVRSKyE3uQIt+UFJEgrbUtilZDqa89aWmq+CklFXuoLaG1VatpRRaEVGzVKL5BUUsWS1BCaqkQSVQjCUGW6/fH3NHTc3/unDkzc8+Z+57X8/E4jzPnOp+57889eZ/JXHPPfMbcXQAAAAAApKVVS08AAAAAAJBvNJ4AAAAAgFTReAIAAAAAUkXjCQAAAABIFY0nAAAAACBVNJ4AAAAAgFTReAIAMsHMRpjZzS09DwAA0Hw0ngCANTKz3czsSTP7yMw+NLO/mdlOLT2vYpnZ0gZfq8xsWYOfjypju0+b2dFr+H0fM/MG+3rfzO41s683Yx8nmNn/lTFHM7MLzOytaA7vmNn4auwbAICGaDwBAInMrLOk+yRdJ6m7pE0kXSjps5acV3O4e8fVX5L+KenABrVbUt79ygb77ifpMUn3mdnhKe93taGSDpH09WgOu0RzAACgqmg8AQBr0luS3H2iu69092Xu/qC7z5IkM9vSzB42s3+Z2QdmdouZdV195ehM2zAzm2VmH5vZjWa2gZndb2ZLzOz/zKxbNHaL6AzhUDN7z8zmmdnpSRMzs4HRmdhFZvaCme1VygGaWWszO8/M3mh8DGa2jpndFp3pXWRmz5hZNzP7paSdJP0hOpP4y6b24+7z3P1KSZdJuqLB/s83szej2+MlM/t2VO8n6WpJe60+YxrVD46Od7GZvW1mZ69htztJmuLub0ZzeM/d/9Bg393NbHx0Nvad6Oxoq6R9AwBQKhpPAMCavCZppZmNM7P9VzeJDZgKjdTGkr4sqaekEY3GHCJpHxWa2AMl3S/pbEnrqfD/0CmNxn9d0taS9pU03Mz2bjwpM9tE0p8lXazCmdgzJE02sx4lHOOwaF+7SdpU0nJJV0W/O15SGxXO9K4n6WRJn7v76ZKmSzo+OqOZ2CAH3ClpUzPrFf08W9LXJHWRNFLSbWa2nrs/J+lUSY9G+9gwGr9Y0pGSuko6WNIZZrZfwr6elnScmf3MzHY0s9aNfn+LpI8kfUnSzpIOknTMGvYNAEBJaDwBAIncfbEKDZlL+r2khdH7FDeIfj/H3R9y98/cfaGkX0nas9FmrnP3+e7+rqTHJT3j7s+5+2eS7lLhJagNXejuH7v7i5JuknREYGpHq3Amb4q7r3L3hyTNkHRACYf5Y0nDo7OBn6rwUuL/NTNToQntIWlLd1/h7tPd/eMS9tHQe9H37pLk7rdHZ0NXufsESe9K6p90ZXef6u4vR+OflTRJ8dt8tRtVaMoPlPSEpPlmdpokmdnmkvaQ9DN3/8Td50m6VlK1XgYMAKgjbVp6AgCA2ubur0gaIhUWzJF0swovwzzCzNZXoVnZXVInFZ7Q/HejTcxvcHlZ4OeOjca/0+Dy25K+EpjW5pK+b2YHNqi1lfRI00f0H1Fz2VPSFDPzBr9qJWldFRq3DSXdYWYdJY2XdJ67r2zOfhrZJPr+YTSH4yT9VNJmUb2jCmdXk+a8q6RLJW0jaS1J7SRNCI11d5c0TtI4M1tL0qHR5WdVeDKhvQpPJqy+SitJc0o9MAAAknDGEwBQNHd/VdJYSX2j0mUqNDDbuXtnFc5EWvjaRevZ4PJm+s8ZwobekTTB3bs2+FrH3S9vzo6ixuxdSd9otK327v5BdCb3fHfvo8LZwe/rP2cEPWm7TThY0lx3f9PMequwcNNQSd3dvasKjd/q2zC0j0mSbpfU0927qPDv0eRt7u6fu/utKry0t68Kt+FSSd0aHHdnd9+xzOMDACCGxhMAkCj6SJDTzWzT6OeeKrz09eloSCcVmpdF0fsuh1Vgt+eZWQcz21bSD1Roshq7WdKBZvataHGg9ma21+p5NtMNki6Pjk1mtv7qM6lmtreZbWNmrVR4b+UKSavPds5X4b2RRTGzDaOXuZ4laXhU7ihplaSFklqZ2QmStmpwtfmSeppZ22gbFl3nX+7+qZl9TYVmOGmfx5vZfmbWMVo06LvR9qdFCw49LWmUmXWKfr+1me0W2jcAAOWg8QQArMkSFT6C4xkz+1iFRuUlSasX07lQ0o4qLFDzZxUWzinXX1U46zdV0pXu/mDjAe7+jqRBKixStFCFs3fDVNr/a6Mk/Z+kh81siaQnVTgmqfCy2HtUuB1ekjRFhTOOUmEBomPN7N9mNiph262jVWE/lvSCpG9KGrT6Y1yi92jeoML7U+dJ6hVdXu0vkt6StMDM5kZnaE+QdGU0159L+uMajm2JpAskzVXhJdC/kHScu0+Pfn+ECosUvarCS39vl7RBaN9r2AcAAE2ywv9hAAC0LDPbQtKbktq6+4qWnQ0AAKgkzngCAAAAAFJF4wkAAAAASBUvtQUAAAAApIozngAAAACAVNF4AgAAAABSReMJAAAAAEgVjScAAAAAIFU0ngAAAACAVNF4AgAAAABSReMJAAAAAEgVjScAAAAAIFU0ngAAAACAVNF4AgAAAABSReMJAAAAAEgVjScAAAAAIFU0ngAAAACAVNF4AgAAAABSReMJAAAAAEgVjScAAAAAIFU0ngAAAACAVNF4AgAAAABSReMJAAAAAEgVjScAAAAAIFU0ngAAAACAVNF4AgAAAABSReMJAAAAAEgVjScAAAAAIFU0ngAAAACAVNF4AgAAAABSReMJAAAAAEgVjScAAAAAIFU0ngAAAACAVNF4AgAAAABSReMJAAAAAEgVjScAAAAAIFU0ngAAAACAVNF4AgAAAABSReMJAAAAAEgVjScAAAAAIFU0ngAAAACAVNF4AgAAAABSReMJAAAAAEgVjScAAAAAIFU0ngAAAACAVNF4AgAAAABSReMJAAAAAEgVjScAAAAAIFU0ngAAAACAVNF4AgAAAABSReMJAAAAAEgVjScAAAAAIFU0ngAAAACAVNF4AgAAAABSReMJAAAAAEgVjScAAAAAIFU0ngAAAACAVNF4AgAAAABSRePZgszsUTM7vtrXBaqFjCPPyDfyjowj78h4ddF4VoCZvWVme7f0PNbEzE4zs/fN7CMzG2Nm7Vp6TsiOWs+4mQ02s5lmttjM5prZKDNr09LzQjZkIN9DzGylmS1t8LVXS88L2UHGkXcZyHhfM3vAzD4wM2/p+bQUGs86YGbfkjRc0jclbSHpS5IubMk5ARXWQdKpktaTtIsKWT+jRWcEVNZT7t6xwdejLT0hoMLIOPJsuaRJko5r6Ym0JBrPFJlZNzO7z8wWmtm/o8ubNhq2pZlNi85E3mNm3Rtcf6CZPWlmi8zshTKe/Rss6UZ3f9nd/y3pF5KGlLgt4Au1knF3/627P+7un7v7u5JukbRr6UcG1E6+gbSQceRdrWTc3We7+42SXi7jcDKPxjNdrSTdJGlzSZtJWibp143GHCvph5I2lrRC0rWSZGabSPqzpIsldVfh7M1kM+vReCdmtln0B7FZwjy2lfRCg59fkLSBma1b4nEBq9VKxhvbQ3V+546KqKV894teovWamZ3HS8lRIWQceVdLGa97NJ4pcvd/uftkd//E3ZdIukTSno2GTXD3l9z9Y0nnSTrMzFpLOlrSFHef4u6r3P0hSTMkHRDYzz/dvau7/zNhKh0lfdTg59WXO5VxeEAtZfwLZvYDSQMkXVnm4aHO1VC+H5PUV9L6kg6RdISkYRU5SNQ1Mo68q6GMQzSeqTKzDmb2OzN728wWq3DH2jUK82rvNLj8tqS2KrxPbXNJ34+ePVlkZosk7SZpoxKmslRS5wY/r768pIRtAV+ooYyvns9Bki6XtL+7f1DqdgCpdvLt7m+4+5vRA58XJV0k6dBSjwtYjYwj72ol4yjgZQzpOl3S/0jaxd3fN7MdJD0nyRqM6dng8mYqvPn4AxX+CCa4+48qMI+XJW2vwpuaFV2e7+7/qsC2Ud9qJeMys/0k/V7St6MHLkC5aibfjXijOQClIuPIu1rNeF3ijGfltDWz9g2+2qjwUtZlkhZFb1S+IHC9o81sGzProMIzfHe4+0pJN0s60My+ZWato23uFXhDdDHGSzou2k83SedKGlvKQaKu1WzGzewbKiwodIi7Tyv5CFHPajnf+5vZBtHlPiq8FOyeEo8T9YuMI+9qOeNmZu0lrRX93N7q8KMNaTwrZ4oKwV79NULS1ZLWVuFZk6cl/SVwvQkqNIHvS2ov6RRJcvd3JA2SdLakhSo86zJMgX8zK7yheaklvKHZ3f8iaZSkR1R4CcHbCv/hAWtSsxlX4UFKF0lT7D+fAXd/SUeJelXL+f6mpFlm9nE0zzslXVrCMaK+kXHkXS1nfPNoTqsXPlwmaXYzjy/zzL1uP8MUAAAAAFAFnPEEAAAAAKSqrMbTzPYzs9lmNsfMhldqUkCtIOPIOzKOPCPfyDsyjiwp+aW20TLEr0naR9JcSdMlHeHuf6/c9ICWQ8aRd2QceUa+kXdkHFlTzhnPnSXNiT576XNJt6nwBlwgL8g48o6MI8/IN/KOjCNTymk8N9F/f+Dq3KgG5AUZR96RceQZ+UbekXFkSpsyrhv6YN/Y63bNbKikoZK0zjrr9O/Tp08ZuwSaNnPmzA/cvUcFNkXGUZOqmXHyjWrjPhx5R8aRd0kZL6fxnCupZ4OfN5X0XuNB7j5a0mhJGjBggM+YMaOMXQJNM7O3K7QpMo6aVM2Mk29UG/fhyDsyjrxLyng5L7WdLmlrM+tlZmtJOlzSvWVsD6g1ZBx5R8aRZ+QbeUfGkSkln/F09xVmdrKkByS1ljTG3V+u2MyAFkbGkXdkHHlGvpF3ZBxZU85LbeXuUyRNqdBcgJpDxpF3ZBx5Rr6Rd2QcWVLOS20BAAAAAGgSjScAAAAAIFU0ngAAAACAVNF4AgAAAABSReMJAAAAAEgVjScAAAAAIFU0ngAAAACAVNF4AgAAAABS1aalJwCguj777LNg/emnnw7Wf/rTn8ZqL774YtH7u/baa4P1oUOHxmpt27YtersAAADIDs54AgAAAABSReMJAAAAAEgVjScAAAAAIFU0ngAAAACAVLG4UAq6du0arK+77rqx2j/+8Y+0p4M69sYbb8RqQ4YMCY7929/+FqybWVG1JKecckqw3r9//1ht4MCBRW8XAAAg5Oijjw7WJ06cGKutXLky7ekgwhlPAAAAAECqaDwBAAAAAKmi8QQAAAAApIrGEwAAAACQKhpPAAAAAECqylrV1szekrRE0kpJK9x9QCUmlSVLliyJ1VatWhUc+95778VqN954Y3Ds/vvvH6uFVsWVpHbt2q1piihD1jM+f/78WO3JJ58se7tbbLFFsL5gwYJY7eOPPw6OvfLKK2O1CRMmBMeuvfbaxU8OzZLljE+ePDlW++ijj4JjL7744mD97bffLmsOSatEh+YRmq8ktWpV/HPAAwbE/3mmTp0aHNuxY8eit5tnWc54c9x+++3B+l133RWrTZo0KTjW3WO1Y445Jjh2++23b8bsihd6THP88ccHx7Zv3z6VOWRJveS7uZJW3w/VzzzzzODY0P8bbdu2LW9ida4SH6fydXf/oALbAWoVGUfekXHkHRlHnpFvZAIvtQUAAAAApKrcxtMlPWhmM81saCUmBNQYMo68I+PIOzKOPCPfyIxyX2q7q7u/Z2brS3rIzF5198caDoj+CIZK0mabbVbm7oCqI+PIuzVmnHwjB8g48ozHKcgMC72RvKQNmY2QtNTd4yuGRAYMGOAzZsyoyP6qLWmBlNAiQE888UQqc0h6g/+pp54aq/Xr1y+VOWSBmc1M4831tZzx0CJXkrTPPvvEaknz69SpU7D+1FNPxWqbb755cOzs2bNjtf79+wfHht7gn7T4yqBBg4L1etVSGW+pfA8bNixYv/rqq2O1Sv2floakuSUtglGspEWEpk2bFqv17t27rH1VQ1r5jrY9QjWY8UrYbbfdgvXQfXjW9OzZM1gP/Z+ThQUX6/FxSrUlPWaeOHFirJZ03zxnzpxYrVevXuVNrE4kZbzkl9qa2Tpm1mn1ZUn7Snqp9CkCtYWMI+/IOPKOjCPPyDeyppyX2m4g6a7omdo2km51979UZFZAbSDjyDsyjrwj48gz8o1MKbnxdPc3JKXzIU5ADSDjyDsyjrwj48gz8o2s4eNUAAAAAACpovEEAAAAAKSq3I9TqRtPP/10sJ7WCrYhEyZMCNZvvfXWWO2rX/1qcOyf/vSnWK1Lly7lTQxVtWrVqljt5JNPDo6dPn160dsN5UiS+vTpU/Q2dthhh1jtu9/9bnDsvffeG6s9+OCDwbGsals/5s2bF6vdcMMNwbG1vIJtNS1dujRYD91XIPs+/fTTYH3x4sVVnkn1vPPOO8F66P+4pNV9geYaMWJErDZu3LjqTyRHOOMJAAAAAEgVjScAAAAAIFU0ngAAAACAVNF4AgAAAABSxeJCRbr22mtbegqJVq5cGaslLXp00EEHxWr33XdfcOw666xT3sSQitBCIrfccktwbPSh0v+lZ8+ewbH7779/eRNLcMUVVwTroQW7HnjggeDYZcuWxWprr712eRNDTbr//vtjtdC/f3N973vfC9a7du1a9raLlbQYUujv9M033wyOfeSRR2K1pGNj4bh8eu2114L1l19+ucozKc9WW20VrO+0006xWlKW+/XrV9E5AUgXZzwBAAAAAKmi8QQAAAAApIrGEwAAAACQKhpPAAAAAECqaDwBAAAAAKliVds689e//jVW+81vfhMcO2zYsLSngxLcc889ZV3/uuuuq9BMipO0cuHBBx8cq40ePTo4dvny5bEaq9oiZMMNNwzWr7/++mC9R48eaU6nZJ988kmwvnDhwlhto402Co5da621Kjon5MfZZ58drP/kJz+p2hw6derUrDqA7OOMJwAAAAAgVTSeAAAAAIBU0XgCAAAAAFJF4wkAAAAASFWTiwuZ2RhJ35G0wN37RrXukm6XtIWktyQd5u7/Tm+a2bLFFlsE60cffXSsds455wTHnn/++bHaggULgmPHjRtX/OSK3JckHXfccbFa9+7dy9pXLcpaxgcNGhSrfeUrXwmO3WGHHWK1Aw44oOJzStvNN98cq5144oktMJNsylrGy3HfffcF67WwiND48eOD9SlTphS9jb59+8ZqxxxzTHBsaKGldu3aFb2vLKmnjFdC0iJcG2+8cVnbfeaZZ4L12267LVY76aSTgmNZXCiOfDff+uuvX/Y2Pv7441jt888/D45lMbfiFHPGc6yk/RrVhkua6u5bS5oa/Qxk1ViRceTbWJFx5NtYkXHk11iRb+RAk42nuz8m6cNG5UGSVp9mGyfpoArPC6gaMo68I+PIOzKOPCPfyItS3+O5gbvPk6Toe/nns4HaQsaRd2QceUfGkWfkG5mT+uJCZjbUzGaY2YzQB18DWUfGkWfkG3lHxpF3ZBy1otTGc76ZbSRJ0ffwqjeS3H20uw9w9wG1sLgDUCQyjrwrKuPkGxlGxpFnPE5B5jS5qm2CeyUNlnR59P2eis2ohb388svB+t/+9reitxFavVaSLrrooqK3MXLkyFhtxYoVwbFbbrllrHbhhRcGx65cuTJW++yzz4JjFy1aFKvlcVXbBDWb8c6dO8dqzz33XAvMpDxnnHFGrPa73/0uOHbJkiVpT6dJL730UqzWp0+f4Ng2bUq9a62qmsx4aLXNVq3Cz5GuWrUqVjOz4Njly5cH623bti16bqEzBWeffXbR1x8zZkywnjTnkDvuuCNWGzFiRHDsX//611ht1113LXpfOVCTGS9X0t9DUo7cPVYLPRZIEvo7k6RbbrklVvvhD39Y9DZ+85vfBMe+8MILsVrSfW2dy2W+KyXpvvmaa64peht33313rDZ79uzg2KRPF8B/a/KMp5lNlPSUpP8xs7lmdpwKId/HzF6XtE/0M5BJZBx5R8aRd2QceUa+kRdNPi3v7kck/OqbFZ4L0CLIOPKOjCPvyDjyjHwjL1JfXAgAAAAAUN9oPAEAAAAAqaLxBAAAAACkKhNLL1bTggXh1ag//PDDordxzjnnVGo6/yVppcxzzz236G2cf/75RY89/fTTY7W77rqr6OsDa9KhQ4dYrUuXLsGxzVn1My2hFRiTlqU/77zzYrWMrHTb4g444IBYrWPHjsGxixcvjtX69+8fHHvUUUcF68cdd1ysds894cUhH3zwwVjtlVdeCY6tBYcddlisNnXq1OBYVg3Njr59+wbrAwcODNafeuqpWC30/7skffzxx7HanXfeGRz77LPPJk2xKEkr9ffr1y9WmzZtWnAsK4kiydprrx2sh/5Okj65IrSCdGiVaBSPM54AAAAAgFTReAIAAAAAUkXjCQAAAABIFY0nAAAAACBVrHaRU2eddVawPmfOnFht/PjxaU8HiNlwww1jtaTFMWrVJZdcEqyHFu7o3Llz2tPJrfbt2wfrocWFktxyyy3Nqre0pLyEFmT55JNPgmPnz58fqx155JHBsdOnT4/VWrduvaYpIsNWrVoVrDdnscK0fP7557Ha7rvvHhz73HPPxWq9evWq+JyQPaEFDCVpv/32i9VCC3BJ4YUN//73vwfHbrfdds2YXf3ijCcAAAAAIFU0ngAAAACAVNF4AgAAAABSReMJAAAAAEgVjScAAAAAIFWsaptTSasRtmvXruhtPPnkk7Haq6++Ghzbp0+foreL7Pvggw+C9U6dOsVqzclckmnTppW9jXKdf/75sdqUKVNaYCb155lnngnWL7/88ljtd7/7XdrTKdnBBx8crB944IGx2hFHHBEcG1rJ95e//GVw7BVXXBGrzZo1Kzj2jDPOiNWuuuqq4FjUpi233DJYT1qxs5o22mijWK1///7BsX/+859jtSVLlgTHhu6Xf//73wfHJq2Ojfqy/vrrl3X9kSNHBuuHH354WdutF5zxBAAAAACkisYTAAAAAJAqGk8AAAAAQKpoPAEAAAAAqWpycSEzGyPpO5IWuHvfqDZC0o8kLYyGne3urLKRMwsXLozV5s+fHxyb5cWFyPiajR07NlY77rjjgmP33HPPWG3UqFHBsaEFgx544IHg2EsvvXQNM6yODTfcMFabOHFiC8yk+bKe8Z49ewbrw4cPj9W23nrr4NgHH3wwWP/HP/4Rq82ZM6fouSUtnvX888/Har179y56u0nWW2+9WO2yyy4Ljn3jjTditTvuuCM4dvz48bHaz372s+DYpH+PlpL1fFfKueeeG6xPnjw5Vlu2bFnZ++vRo0esdtpppwXHHn/88bHauuuuGxx74403xmpDhw4Njr311ltjtTPPPDM4tm/fvsF6FpDxygll6Sc/+UkLzKQ+FXPGc6yk/QL1q9x9h+iLoCPLxoqMI9/Giowjv8aKfCPfxoqMIweabDzd/TFJH1ZhLkCLIOPIOzKOPCPfyDsyjrwo5z2eJ5vZLDMbY2bdKjYjoHaQceQdGUeekW/kHRlHppTaeP5W0paSdpA0T1L406slmdlQM5thZjNC7xkEahQZR94VlXHyjYziPhx5R8aROSU1nu4+391XuvsqSb+XtPMaxo529wHuPiD0RnSgFpFx5F2xGSffyCLuw5F3ZBxZ1OSqtiFmtpG7z4t+PFjSS5WbEmrF3nvvHat99atfbYGZVF89ZjxplcNf/OIXsZqZBcc+9thjsdouu+wSHBvaRtJ2k+ot7d133w3Wt9122yrPpPnykPHQ6qqnnnpqcOzJJ58crLt7rLZq1aqi55CUzbXWWqvobaTl+uuvj9X+/ve/B8e+8sorsdqzzz4bHFtrq9qG5CHfzdWcFZ3POuus4NgnnngiVjvwwAODY0P1pBXPm+OYY46J1ZJWtQ35zW9+06x6VtVjxtOSdJ/fqlX8/NysWbOCY++8885Y7Xvf+155E8uhYj5OZaKkvSStZ2ZzJV0gaS8z20GSS3pL0o9TnCOQKjKOvCPjyDPyjbwj48iLJhtPdz8iUI5/yBKQUWQceUfGkWfkG3lHxpEX5axqCwAAAABAk2g8AQAAAACpKmlxIdSH0MpntbBgBtJxxx13BOtvv/12lWcSF1oUYp999gmO7devXypzuPbaa2O1ESNGBMe+9dZbqcwBpWvTpv7+u1tvvfVitYEDBwbHhhYXQj587Wtfi9Ueeuih4Njly5fHakn/77dt27a8iaXkgw8+aOkpIGNCiwhJtbuwYZZxxhMAAAAAkCoaTwAAAABAqmg8AQAAAACpovEEAAAAAKSKxhMAAAAAkKr6W+avCVtttVWwvvnmmwfroRU/x40bFxx72GGHxWpdu3ZtxuyK96c//SlYnzVrVtHbuOaaayo1HWTAu+++W/TYadOmBeuhv5+RI0cGxybVQ+bOnRur7bLLLsGxe+21V9H7as4KuK+++mqstnjx4qKvD1TbE088EatNnjy56Os///zzwfqgQYNKnhNqQ9JKtbWwcv3KlStbegqoM+eee26wfskllxS9jffff79S08k1zngCAAAAAFJF4wkAAAAASBWNJwAAAAAgVTSeAAAAAIBUsbhQIz179gzWt9tuu2A9tLjQCSecEBx72WWXxWonnXRScOx3v/vdpCnG3HPPPbHaOeecExy7YsWKoreL+tKpU6dg3d1jtaeffjo4dptttonVBg4cWPR2m2P58uXB+tSpU2O1AQMGBMfuueeesdr9999f9BzKPQa0vBEjRsRqb775ZnDseeedF6slLUhXTcuWLQvWQ/NtzoJY1113XbB+wQUXFL0NVN4f//jHYP3ll18O1k877bRYrUuXLhWdUymSsnjooYdWeSaodz/+8Y+D9T/84Q+x2vz584NjTznllFjtxBNPLG9iOcQZTwAAAABAqmg8AQAAAACpovEEAAAAAKSKxhMAAAAAkKomG08z62lmj5jZK2b2spn9NKp3N7OHzOz16Hu39KcLVB4ZR96RceQZ+UbekXHkRTGr2q6QdLq7P2tmnSTNNLOHJA2RNNXdLzez4ZKGSzozvam2rHbt2gXrrVu3jtVWrlwZHE6/26oAACAASURBVBtaAXfYsGHBsT//+c+bMTuUiYxLGjJkSLB+8cUXx2qh1dsk6fe//32sNmvWrOBYM4vVfvSjHwXHDh48OFa76KKLgmOfe+65WG3hwoXBsY899lisdswxxwTH9urVK1b72c9+Fhy79tprB+stiIwnePXVV2O1O+64Izh28uTJsdp3vvOd4Nhvf/vbsdoee+wRHBtaTb1Vq/Dzwu+9916sdu655wbHPv7448F6sZL+f6pBuc33hx9+GKuFVqmVpHnz5gXrd999d6wWWs1Zkg466KBY7aOPPgqOXbp0abAeEloddNSoUcGxn376adHb7dChQ6yW9P9IxuU247Vg4403Dta7d+8eq73//vtFb/e1114L1nv37l30NvKmyTOe7j7P3Z+NLi+R9IqkTSQNkjQuGjZOUvzeCsgAMo68I+PIM/KNvCPjyItmvcfTzLaQ1E/SM5I2cPd5UuEPQtL6lZ4cUG1kHHlHxpFn5Bt5R8aRZUU3nmbWUdJkSae6e9GfQG1mQ81shpnNSHq5G1ALyDjyrpSMk29kBffhyDsyjqwrqvE0s7YqBP0Wd78zKs83s42i328kaUHouu4+2t0HuPuAHj16VGLOQMWRceRdqRkn38gC7sORd2QcedDk4kJWWAHkRkmvuPuvGvzqXkmDJV0efb8nlRnWiEmTJgXr06ZNi9V233334Njly5fHaqEFVqotab41uEBKKsh4wTrrrBOsH3zwwbHa6NGjg2NffPHFove3/vrxVwSFFjKSpHXXXTdWmzJlSnBs6M38e+65Z3Bs6JnfmTNnBseOGTMmVuvYsWNwbK0h48lCOUwSWvQkaSGiUN3dg2OPPfbYWK1Nm/B/z6EFjhYvLvrER6K2bdvGal/72tfK3m415DnfocUKP/vss2ZtI3S/fOihhwbH/u///m+s9uSTTwbH/vOf/2zWPNJw2GGHxWr77LNPC8wkXXnOeC277rrrYrW999676Osn9Q5JC8LVg2JWtd1V0jGSXjSz56Pa2SqEfJKZHSfpn5K+n84UgdSRceQdGUeekW/kHRlHLjTZeLr7E5KSTst9s7LTAaqPjCPvyDjyjHwj78g48qJZq9oCAAAAANBcNJ4AAAAAgFTReAIAAAAAUlXM4kJYg5133jlW22mnnYJjk1aGq6ZddtklVvvzn/8cHJu0yinqyy9/+ctYrVu3bsGxI0eOjNU6d+4cHPv444/HaqHVa5urd+/esdqcOXOCY8eNGxerJa0mmpUVbNE8Z555Zqx2zz3hhSHffffdVOYwYcKEVLYbElq9VpJGjRoVqyWteI7qCX30xW233RYcu++++xa93aQVlpO23dI6dOgQrJ911llVngnqSdJjnWIl3befeuqpsVq9PMbgjCcAAAAAIFU0ngAAAACAVNF4AgAAAABSReMJAAAAAEgViwul4E9/+lOwfu+998ZqQ4YMCY41S/qc4PJcddVVsVq9vKEZpVl77bVjtUsuuSQ4Nqne0pIWyjrxxBOrPBPUmk022SRWmz17dnDsAw88EKtNmjQpOPb2228vb2IV0Lp161htt912C479f//v/6U9HVTIHnvsEazPmDEjWD///PNjtSlTplR0TpX0jW98I1a79NJLg2O32mqrtKeDOrbFFlvEatdee21w7CmnnBKr9e3bNzg2aRHDesAZTwAAAABAqmg8AQAAAACpovEEAAAAAKSKxhMAAAAAkCoaTwAAAABAqup3WaUUdevWLVgfPHhwUTUAQMtp3759sD5o0KBY7dvf/nZw7L777hur/fCHPyxvYglCq5ZK0u677x6rff3rX09lDqietm3bBuv9+vUL1u+6665Y7YUXXgiOHTVqVKz29ttvB8dOnz49Vtt2222DY0Mr+O+6667BsTvuuGOslnTMQJq6dOkSqyWths8q+cXhjCcAAAAAIFU0ngAAAACAVNF4AgAAAABSReMJAAAAAEhVk4sLmVlPSeMlbShplaTR7n6NmY2Q9CNJC6OhZ7v7lLQmCqSFjCPPyHe62rQJ/zfKYnLVQ8bXLJTR/v37B8fefvvtaU8HJSDjyItiVrVdIel0d3/WzDpJmmlmD0W/u8rdr0xvekBVkHHkGflG3pFx5B0ZRy402Xi6+zxJ86LLS8zsFUmbpD0xoFrIOPKMfCPvyDjyjowjL5r1Hk8z20JSP0nPRKWTzWyWmY0xs/CHVwIZQsaRZ+QbeUfGkXdkHFlWdONpZh0lTZZ0qrsvlvRbSVtK2kGFZ2F+mXC9oWY2w8xmLFy4MDQEqAlkHHlGvpF3ZBx5R8aRdUU1nmbWVoWg3+Lud0qSu89395XuvkrS7yXtHLquu4929wHuPqBHjx6VmjdQUWQceUa+kXdkHHlHxpEHTTaeZmaSbpT0irv/qkF9owbDDpb0UuWnB6SPjCPPyDfyjowj78g48qKYVW13lXSMpBfN7PmodrakI8xsB0ku6S1JP05lhkD6yDjyjHwj78g48o6MIxeKWdX2CUkW+BWfE4RcIOPIM/KNvCPjyDsyjrxo1qq2AAAAAAA0F40nAAAAACBVNJ4AAAAAgFTReAIAAAAAUkXjCQAAAABIFY0nAAAAACBVNJ4AAAAAgFTReAIAAAAAUkXjCQAAAABIFY0nAAAAACBVNJ4AAAAAgFSZu1dvZ2ZLJM2u2g6raz1JH7T0JFKStWPb3N17tMSOyXhmZe3YWiTj5DuzsnZs3IenI2s5aI6sHRsZT0fWctAcWTu2YMbbVHkSs919QJX3WRVmNoNjg8h4JuX52CqMfGdQno8tBWQ8g/J8bCkg4xmUl2PjpbYAAAAAgFTReAIAAAAAUlXtxnN0lfdXTRwbpHzfVhwb8nw7cWyQ8n1bcWyQ8n1bcWw1rqqLCwEAAAAA6g8vtQUAAAAApKoqjaeZ7Wdms81sjpkNr8Y+02RmY8xsgZm91KDW3cweMrPXo+/dWnKOpTCznmb2iJm9YmYvm9lPo3rmjy1tecp4XvMtkfFykPFsIOOlI+O1j3yXLk/5lsh4Fo9NqkLjaWatJV0vaX9J20g6wsy2SXu/KRsrab9GteGSprr71pKmRj9nzQpJp7v7lyUNlHRS9G+Vh2NLTQ4zPlb5zLdExktCxjOFjJeAjGcG+S5BDvMtkfEsHltVznjuLGmOu7/h7p9Luk3SoCrsNzXu/pikDxuVB0kaF10eJ+mgqk6qAtx9nrs/G11eIukVSZsoB8eWslxlPK/5lsh4Gch4RpDxkpHxDCDfJctVviUyrgwem1SdxnMTSe80+HluVMubDdx9nlQIjaT1W3g+ZTGzLST1k/SMcnZsKaiHjOcuA2S8Wch4BpHxZiHjGUO+m6Ue8i3lLAd5zHg1Gk8L1FhKt4aZWUdJkyWd6u6LW3o+GUDGM4aMNxsZzxgy3mxkPEPId7OR74zJa8ar0XjOldSzwc+bSnqvCvuttvlmtpEkRd8XtPB8SmJmbVUI+i3ufmdUzsWxpageMp6bDJDxkpDxDCHjJSHjGUG+S1IP+ZZykoM8Z7wajed0SVubWS8zW0vS4ZLurcJ+q+1eSYOjy4Ml3dOCcymJmZmkGyW94u6/avCrzB9byuoh47nIABkvGRnPCDJeMjKeAeS7ZPWQbykHOch7xs09/TPtZnaApKsltZY0xt0vSX2nKTKziZL2krSepPmSLpB0t6RJkjaT9E9J33f3xm96rmlmtpukxyW9KGlVVD5bhdeWZ/rY0panjOc13xIZLwcZzwYyXjoyXvvId+nylG+JjCuDxyZVqfEEAAAAANSvarzUFgAAAABQx2g8AQAAAACpovEEAAAAAKSKxhMAAAAAkCoaTwAAAABAqlqk8TSzlWb2fIOv4Snua4iZ/Tq6fIKZHZvCPsaa2ZtmdkL08x5m9qyZrTCzQ9dwvf3MbLaZzUm6Dcysu5k9ZGavR9+7JYw7K9rObDP7VsKY/mb2YjTu2uizghqPaWdmt0djnjGzLRK2NTia0+tmNjhhTK9oG69H21wrMMaiucwxs1lmtmPCtoK3lZndYmYfrul2bgl1kPFK5uTkaDtuZuutYQ5kvIaQ8S+uVxM5aTTm+2b2spmtMrMBazjmmpg7GSfjjcaR8Sog319c7y9mtsjM7lvDtnksrgrl292r/iVpaRX3NUTSr1Pex1hJhzb4eQtJ20ka37De6DqtJf1D0pckrSXpBUnbBMaNkjQ8ujxc0sjAmG2i67eT1CvabuvAuGmSvirJJN0vaf/AmBMl3RBdPlzS7YEx3SW9EX3vFl3uFhg3SdLh0eUbJP0kMOaAaC4maaCkZ5p7WzW+/Wvhqw4yXsmc9Iv+Zt6StF7C/sk4Gc9yxquSkwbjvizpfyQ9KmlAwvHW1NzJOBkn4+S72vmOfvdNSQdKum8N2+axeIXyXVMvtTWzt8zsQiucLXzRzPpE9Y5mdlNUm2Vmh0T1I6LaS2Y2ssF2fmBmr5nZXyXt2qA+wszOiC4/amYjzWxaNHb3qN7BzCZF+7k9epYg8dm0EHd/y91n6T8f/Bqys6Q57v6Gu38u6TZJgwLjBkkaF10eJ+mghDG3uftn7v6mpDnR9r9gZhtJ6uzuT3khIePXsK3V+7tD0jcDz8Z8S9JD7v6hu/9b0kOS9mu0P5P0jWgbTc19vBc8LalrNNeGir2tal5eMq4K5USS3P05d3+riP2R8Qwg4y2fE3d/xd1nN3F8NTn3LCDjLZ8TMp6eOsu33H2qpCXN2BaPxcvId0s1nmvbf5/e/98Gv/vA3XeU9FtJZ0S18yR95O5fcfftJD1sZhtLGqnCDbqDpJ3M7KDohrpQhZDvo8IzEEnauPvOkk6VdEFUO1HSv6P9/EJS/9WDzewPJQQ/ySaS3mnw89yo1tgG7j5PkqLv65e4rU2ielP7+2Jb7r5C0keS1i1hf+tKWhRto6j9NTH3Ym6rWpL3jFcqJ8Ui47WHjGc7J1mee7WQ8WznJMtzrwbyXTwei1co322ae4UKWebuOyT87s7o+0xJ34su763CqWZJkrv/28z2kPSouy+UCq8vlrRHNKRh/XZJvYvY1xbR5d0kXRPt5yUzm9Vgv8cXdXTFiT3rIslT3Fax+6vUtqq9v1qT94xX+9+tFjNHxsl4lnOS5blXCxnPdk6yPPdqIN+VVYt5q7l819RLbSOfRd9X6j+Nsam4G2C1Ym+IpH1Vw1xJPRv8vKmk9wLj5q8+3R19X1DituZG9ab298W2zKyNpC6SPixhfx+ocKq+zRrGNGfuxdxWWZGHjFcqJ83e3xq2RcZrBxn/j1rNSZbnXgvI+H/Uak6yPPeWVi/5LhaPxSuU71psPEMelHTy6h+ssJrUM5L2NLP1zKy1pCMk/TWq72Vm65pZW0nfb+a+npB0WLSfbSR9pQLzD5kuaWsrrDa1lgrPIt0bGHevpMHR5cGS7kkYc7gVVsHqJWlrFd68/IXopQFLzGxg9JrvY9ewrdX7O1TSw9Hr0Bt6QNK+ZtYt+rfYN6o13J9LeiTaRlNzP9YKBqrwMo55jcYUe1tlWdYyXpGcNHN/ZDzb6jLjNZyTLM+9VpHxglrJSZbnXovymO9StsVj8XLy7VVa0arhlwrPajzf4OvyqP6WolUtJQ1Q4TS9JHVU4Q2xL6mwitL3ovqRkl6M6qMabP8Hkl5TIfzXKFpJS9IISWdElx9VtBKapPUkvRVdXkeFN+HOivY5U9LW0e/+oMDqaYqvpLWTCs8MfCzpX5JeTrgdDojm+Q9J5ySMWVfSVEmvR9+7J4w7J9rObAVWyGpwm74Ujfu1JAuMaS/pjyq8KXqapC8lbOuH0Zg5kn6QMOZL0TbmRNtsFxhjkq6P5vRi6PZt6rZqfPvXwlcdZLySOTkl+ntZocKzZ38g42Q8ZxmvSk4ajDlYhb+pzyTNl/RArc+djJNxMk6+Wyjfj0taKGlZlKlvBcbwWLxC+bZoMCLRMzZt3f1TM9tShYD19sIKTknXGavCMsx3JI1Berj9m4eMZw+3f/OQ8ezh9m8eMp493P7FI9/ZU+zt31KLC9WyDpIeiV4aYCp83k1i0CMfSfqFma3n7jekPkN8wQpvZP+a/rNUNJpGxjOEjJeEjGcIGS8JGc8QMt5s5DtDmpNvzngCAAAAAFKVlcWFAAAAAAAZReMJAAAAAEgVjScAAAAAIFU0ngAAAACAVNF4AgAAAABSReMJAAAAAEgVjScAAAAAIFU0ngAAAACAVNF4AgAAAABSReMJAAAAAEgVjScAAAAAIFU0ngAAAACAVNF4AgAAAABSReMJAAAAAEgVjScAAAAAIFU0ngAAAACAVNF4AgAAAABSReMJAAAAAEgVjScAAAAAIFU0ngAAAACAVNF4AgAAAABSReMJAAAAAEgVjScAAAAAIFU0ngAAAACAVNF4AgAAAABSReMJAAAAAEgVjWcLMrNHzez4al8XqBYyjjwj38g7Mo68I+PVReNZAWb2lpnt3dLzSGJmh5vZbDP7yMwWmNk4M+vc0vNCdpBx5FkG8t3OzK4ys/fM7N9m9hsza9vS80J2ZCDjN5jZ0gZfn5nZkpaeF7IjAxk3M7vYzN6NHqs8ambbtvS8qo3Gsz78TdKu7t5F0pcktZF0cctOCagoMo48Gy5pgKS+knpL2lHSuS06I6CC3P0Ed++4+kvSREl/bOl5ARX0fUk/lLS7pO6SnpI0oUVn1AJoPFNkZt3M7D4zWxg9S32fmW3aaNiWZjYtevbjHjPr3uD6A83sSTNbZGYvmNlepczD3d9x9w8alFZK2qqUbQENkXHkWa3kW9KBkq519w/dfaGka1V4AAOUpYYy3nBO60g6RNK4crcF1FDGe0l6wt3fcPeVkm6WtE2J28osGs90tZJ0k6TNJW0maZmkXzcac6wKDyA2lrRChQcUMrNNJP1ZhbM23SWdIWmymfVovBMz2yz6g9gsaSJmtpuZfSRpiQp36FeXd2iAJDKOfKuVfFv01fDnTc2sS4nHBaxWKxlv6BBJCyU9VsoBAY3USsZvk7SVmfWO3ioxWNJfyjy2zKHxTJG7/8vdJ7v7J+6+RNIlkvZsNGyCu7/k7h9LOk/SYWbWWtLRkqa4+xR3X+XuD0maIemAwH7+6e5d3f2fa5jLE9HLEDeVdIWktypykKhrZBx5VkP5vl/ST82sh5ltKOmUqN6hAoeJOlZDGW9osKTx7u5lHRygmsr4PEmPS5qtQvP7fUmnVeQgM4TGM0Vm1sHMfmdmb5vZYhWevesahXm1dxpcfltSW0nrqfDMzPejZ08WmdkiSbtJ2qicObn7uyo8w3JbOdsBJDKOfKuhfF8i6TlJz0t6UtLdkpZLWlDCtoAv1FDGV8+npwpNwfhStwE0VEMZv0DSTpJ6Smov6UJJD5tZXT2BSOOZrtMl/Y+kXdy9s6Q9onrDl0z1bHB5MxUeTHygwh/BhOjZk9Vf67j75RWYVxtJW1ZgOwAZR57VRL7dfZm7n+zum7j7lyT9S9LM6H1CQDlqIuMNHCvpSXd/o4xtAA3VSsa3l3S7u8919xXuPlZSN9XZ+zxpPCunrZm1b/DVRlInFU6nL4reqHxB4HpHm9k20TMeF0m6o8Gbjg80s2+ZWetom3sF3hDdJDM7KnrtuZnZ5io8ez615CNFvSLjyLNazvcmZrZxlO+BKrwULDQXYE1qNuMNHCtpbBnXR32r5YxPV+Hs6QZm1srMjlHhzOqcko40o2g8K2eKCsFe/TVChcVN1lbhWZOnFX4T8QQV7mTfV+HU+ylSYZVOSYMkna3Cm+zfkTRMgX+z6AH3Ukt+Q/M2Krw8a6kKHzsxW9KPmn+IqHNkHHlWy/neUoV8f6zCSp/D3f3BEo4R9a2WMy4z+6oK79HnY1RQqlrO+EhJL6jwlolFKry/8xB3X9T8w8wu473bAAAAAIA0ccYTAAAAAJCqshpPM9vPzGab2RwzG16pSQG1gowj78g48ox8I+/IOLKk5JfaRssQvyZpH0lzVXjT7BHu/vfKTQ9oOWQceUfGkWfkG3lHxpE15Zzx3FnSHHd/w90/V+Ez8wZVZlpATSDjyDsyjjwj38g7Mo5MKafx3ET//YGrc6MakBdkHHlHxpFn5Bt5R8aRKW3KuK4FarHX7ZrZUElDJWmdddbp36dPnzJ2CTRt5syZH7h7jwpsioyjJlUz4+Qb1cZ9OPKOjCPvkjJeTuM5V1LPBj9vKum9xoPcfbSk0ZI0YMAAnzFjRhm7BJpmZm9XaFNkHDWpmhkn36g27sORd2QceZeU8XJeajtd0tZm1svM1pJ0uKR7y9geUGvIOPKOjCPPyDfyjowjU0o+4+nuK8zsZEkPSGotaYy7v1yxmQEtjIwj78g48ox8I+/IOLKmnJfayt2nSJpSobkANYeMI+/IOPKMfCPvyDiypJyX2gIAAAAA0CQaTwAAAABAqmg8AQAAAACpovEEAAAAAKSKxhMAAAAAkCoaTwAAAABAqmg8AQAAAACpovEEAAAAAKSKxhMAAAAAkCoaTwAAAABAqmg8AQAAAACpovEEAAAAAKSKxhMAAAAAkKo2LT0BALVt6dKlsdrrr78eHHvWWWfFag888EDR+5oxY0aw3r9//6K3AQAA6tvixYuD9QkTJsRqN910U3DszJkzY7XLLrssOHb48OHNmF394ownAAAAACBVNJ4AAAAAgFTReAIAAAAAUkXjCQAAAABIFY0nAAAAACBVZa1qa2ZvSVoiaaWkFe4+oBKTqkUrVqwI1ufMmROrnXPOOcGxd955Z6zWqlXxvX/Silnnn39+rNauXbuit4tk9ZTxVatWBevHH398rPbHP/6x6O02J+NDhgwJ1idPnhyr9e7du+jtIlk9ZRz1iYwjz8i39NRTT8Vqxx13XHDsq6++WvR2zSxWmzRpUnDsKaecEqt16NCh6H3Vi0p8nMrX3f2DCmwHqFVkHHlHxpF3ZBx5Rr6RCbzUFgAAAACQqnIbT5f0oJnNNLOhoQFmNtTMZpjZjIULF5a5O6DqyDjybo0ZJ9/IATKOPONxCjKj3MZzV3ffUdL+kk4ysz0aD3D30e4+wN0H9OjRo8zdAVVHxpF3a8w4+UYOkHHkGY9TkBllvcfT3d+Lvi8ws7sk7SzpsUpMrKUkLSJ00UUXBeuXXnpp0dsOLbISeuNykpEjRwbrXbt2jdWGDRtW9HaRLI8Zd/dg/Vvf+law/vDDD5e1v2OOOSZY32abbWK1ESNGBMeG/v4mTJgQHNucvynkM+NAQ3nL+De+8Y1g/dFHHw3WQwvEHXXUUcGxX/rSl2K1ddddNziWhVNqQ97yvSaLFi0K1kN/E5999lkqc3j++eeD9dBiRhMnTkxlDllW8hlPM1vHzDqtvixpX0kvVWpiQEsj48g7Mo68I+PIM/KNrCnnjOcGku6Kzi60kXSru/+lIrMCagMZR96RceQdGUeekW9kSsmNp7u/IWn7Cs4FqClkHHlHxpF3ZBx5Rr6RNXycCgAAAAAgVTSeAAAAAIBUlbWqbR6dd955wfoVV1xR5ZkU74ILLojVDjzwwODYPn36pD0dtIBly5YF66FVmn/84x8Hx5a7eq0k/fznP4/VLrvssqKv//rrrwfrY8aMidXOOeec4Ngvf/nLRe8P+bN48eJgffz48cH6zTffHKtNnz697HnsvPPOsdqRRx4ZHDt48OBYrXPnzmXPAfUlaUXv0P3njTfeWPQ2QivdStKZZ54Zq4VW9gSa69NPPw3WBwwYEKw3ZwXbHXfcMVY79NBDi77+1VdfHaxPmjQpVktaEfrXv/510fvLG854AgAAAABSReMJAAAAAEgVjScAAAAAIFU0ngAAAACAVNX14kKhhYRGjRoVHJv0pv3mmDZtWqy29dZbF3397bbbLlh/5513YrWLLrooOPbWW28ten+oTUuXLo3Vvvvd7wbHtmkT/xOfOnVq2XM45JBDgvWLL764rO1uv33xH0c2duzYYH3kyJFlzQHZEVqAImlxqfnz5wfr7h6rVeL+PrRAUej/ACmc2Tlz5gTHtm/fvryJIfOSHgu88sorwfqCBQtitVDukyRlcejQobHa9ddfHxz7yCOPxGpdunQpeg6oL6GFESXpjTfeKHobSYu5hRbbWmuttYre7l577RWs77bbbrHauHHjgmNZXAgAAAAAgJTQeAIAAAAAUkXjCQAAAABIFY0nAAAAACBVNJ4AAAAAgFTVxaq2y5cvD9bHjx8fq61atSo4tlWrcI/+ne98J1abNGlScGy7du2SpliUyy67LFg/6qijYrVFixaVtS/Urk8++SRWe+GFF4JjK5GDn//857Ha+eefHxzbunXrsvaV9LcKhLz66quxWtLqtUlCmV1//fWDY/fee+9YbeDAgcGxo0ePjtWS/k7ff//9WO20004Ljv3tb38brKN+XH311cH6hRdeGKwvXrw4VgvlUwqv+BnKpxRe/Tkp40OGDInV7rrrruBYoBKuvPLKYL05K9iGJN3n/+QnP4nVbrjhhuDY119/PVZrzqdcZBlnPAEAAAAAqaLxBAAAAACkisYTAAAAAJAqGk8AAAAAQKqaXFzIzMZI+o6kBe7eN6p1l3S7pC0kvSXpMHf/d3rTjFuxYkWwPnPmzFgt6Q337777bqyWtIjQ9ttvH6zfdtttsVq5iwglmTt3brAeeoN/0nEgrlYzniS08MmOO+4YHPvwww/Hat26dQuOvemmm4L1Aw88sBmzK88ll1xStX3Vk6xlvFhPPfVU0WMPP/zwYH348OGxWt++fUue02onnHBCrNacxbcmTJgQrIcW++rVq1fxE8upvGa8Obp06VJ0/Re/+EVwbKgeWtBOkq655ppY7ZxzzgmOvffee4N1FId8177QY/+VK1cGx4Ye64wdO7bSU6pJxXQnYyXt16g2XNJUd99a0tToZyCrxoqMI9/Giowj38aKjCO/xop8IweabDzd/TFJHzYqD5I0Lro8TtJBFZ4XUDVkHHlHxpF3ZBx5Rr6Rf7gFhQAAG+1JREFUF6W+HnMDd58nSdH38IeeAdlFxpF3ZBx5R8aRZ+QbmZP6GwHNbKiZzTCzGQsXLkx7d0DVkXHkGflG3pFx5B0ZR60otfGcb2YbSVL0fUHSQHcf7e4D3H1Ajx49StwdUHVkHHlXVMbJNzKMjCPPeJyCzGlyVdsE90oaLOny6Ps9FZtRkSZOnBis/+AHPyhrux06dAjWb7jhhmB97bXXLmt/SRYvXhyr/epXv0plXwhq8Yw3x+TJk4P1UGZOOumk4NhK/Gc0ffr0WO3zzz8Pjh03blysFsp9kiOPPLL4iSEkUxkv19Zbbx2sV2IF25DHH3+8rOt/+umnwfr8+fNjNVa1TVRXGU9L0uOis846K1Z78803g2PHjBkTqz3zzDPBsbvsskszZlfXcpvvpMfW5557brB+8cUXx2pJKyyHPumie/fuRc/t7rvvDtavvfbaoreRtNptPWjyjKeZTZT0lKT/MbO5ZnacCiHfx8xel7RP9DOQSWQceUfGkXdkHHlGvpEXTZ7xdPcjEn71zQrPBWgRZBx5R8aRd2QceUa+kRepLy4EAAAAAKhvNJ4AAAAAgFTReAIAAAAAUlXqqrYt7t13301lu0krEe6www6p7O+dd94J1nfeeedYjc9eQpLOnTsH6yNGjCh725988kmsdsEFFwTHXnXVVbGau5c9hzPOOCNW23bbbcveLrIttBJzUt4uuuiiYL3cfM6dOzdYv+mmm2K1VatWBce2ahV/Drh169bBsR07dmzG7IDqmjFjRrBeif8HUD+S7v+GDRsWrL/22mux2v9v7/6jpa7rPI6/3vFDRNBAEA1EUlHX0mAlQtHi1G5p2fFHspt7FDPTzKW2jnuK9BCaeRSyLM2j6eLiD1q0xTaO6S7olgqHVVETKEL8wVlAkutiqKvlXnnvH/M1r3w/c+935s5n7nw/83ycc8/Mfd/3fL+f7/Ca4X7mO/O5oefg7urN1M4rkXPGEwAAAAAQFRNPAAAAAEBUTDwBAAAAAFEx8QQAAAAARFXaxYWGDh0arIc+wD548OBg79atW3O1Zi/ccP/99wfr27ZtK7yN0DGHFoSRpNdffz1X23333QvvC+0n9Pj53e9+F+yNtYDEVVddlauNGDEi2Hv++efnatWeL1BuJ554Yq5WbdGGjRs3BuuXXXZZrmZmvRpXtW2EFhGq1lttwbCxY8f2bmBARE8++WSwHsp4Oy+wgvpU+798/vz5udrKlSuDvdUW9Yxhjz32CNarLdDYDjjjCQAAAACIioknAAAAACAqJp4AAAAAgKiYeAIAAAAAomLiCQAAAACIqrSr2p533nnB+lFHHZWrjRo1Ktjb7BVsQw444IBgvberKj744IPB+hFHHJGrzZ49O9h76qmn5mqsDgpJet/73hes33PPPU0bw6xZs4L1O+64I1ertnr0Xnvt1dAxobkGDRqUq912223B3ltvvTVYX7JkSa720ksvBXvfeOONwmMLrRZ+xhlnBHtXrVqVqz3xxBPB3t/+9re52pQpUwqPC2iUDRs25GqxVjYHuhPKXStnsV+/fn09hD7DGU8AAAAAQFRMPAEAAAAAUTHxBAAAAABExcQTAAAAABBVj4sLmdnNkk6UtM3d35/VLpF0rqSOrO0id2/eqiKSBgwYEKyXbZGF0GI/kvTFL36x8DZWrFiRq61duzbYu3HjxlztnHPOCfauXr06V5szZ06wd8899+xmhK2tVTPeyj71qU8F62PGjMnVrrjiimBvKOPvfve7g71z587N1V588cVgb2hRlmnTpgV7b7/99lyt2sJJZdZOGT/66KNrql9//fW52lNPPRXs3b59e+FxhB4LoZokrV+/PlebOHFi4X21u3bKdysJLS5UbWHEUP25554L9u6zzz69G1iCyHj3vve97+VqmzdvLnz7yZMnB+unnHJKrlZtscL77ruv8P7aWZEzngskHR+oX+3uE7Kvtgw6krFAZBxpWyAyjnQtEPlG2haIjCMBPU483f1BScVf5gVKhowjdWQcKSPfSB0ZRyp68xnPmWa22sxuNrNh1ZrM7DwzW2Vmqzo6Oqq1Aa2IjCN1PWacfKPEeA5H6sg4SqXeief1kg6SNEHSVkn5N1dn3P1Gd5/k7pNGjhxZ5+6ApiPjSF2hjJNvlBTP4UgdGUfp1DXxdPcX3P1Nd98p6SZJ4U/lAiVFxpE6Mo6UkW+kjoyjjHpc1TbEzPZz963Zt6dICi+hih6NGDEiWL/uuusKb+OPf/xjrlZtda2ZM2fmatVW/vrBD36Qq23dujXQKf3kJz/pboilQ8a7d9xxxxWuhzJXq6985Su5WrWVm6dPn56rhVZolqR77smvxTB27Nhg79ChQ7sbYumQ8eoOOeSQpu7v0EMPzdWGDQu/a+7SSy/N1e66665g7+677967gZUY+e4b7l6490Mf+lDEkaSvHTMe+n1XkubNm1d4G6HV8xctWhTsHTduXK527rnnBntD9bvvvjvYu2XLllxt9OjRwd7UFPlzKv8iaZqkEWa2WdIcSdPMbIIkl7RRUvG//QG0GDKO1JFxpIx8I3VkHKnoceLp7qcHyvMjjAXoE2QcqSPjSBn5RurIOFLRm1VtAQAAAADoERNPAAAAAEBUdS0uhNYyaNCgXO3EE08M9obqs2fPDvZefvnludrvf//7YG9nZ2eu1r8/8SqT7373u8H63nvvnat9/vOfjz2cHr3//e8P1h966KFc7Zhjjgn2zpo1K1fbvj38N7qvuOKKGkYH9E61P3mwdOnSXC20UIUkHXzwwQ0dE9ATM6upDtTiscceC9Zff/31XK1a5kKLsYUWEapm+PDhwXrod4Sf/exnwd7QYkihBT2l9B47nPEEAAAAAETFxBMAAAAAEBUTTwAAAABAVEw8AQAAAABRMfEEAAAAAETFsqPQZZddFqxv2rQpV7v99tuDvTfccEOuNnPmzN4NDFFUW7X1yiuvDNZ37NiRq33gAx8I9k6cODFXe9e7mvv61ogRI3K1j3zkI8HeZ555Jlfr6Oho+JiAWv3iF78I1g844IBcbeHChcHeOXPmNHRMQE/cva+HAEiSvvCFLwTr06ZNi7K/Pffcs3Dvtddem6uF/pKEJA0ZMqTuMbUizngCAAAAAKJi4gkAAAAAiIqJJwAAAAAgKiaeAAAAAICoWFwIVR122GGFe0Mfij777LODvXvssUfdY0LvDR8+PFj/zGc+E6zPnz8/V5s8eXKw98c//nGuVu0D/rF0dnbmatu2bSt8+xUrVjRyOEBdhg0bFqwPGjQoV7vqqquCvSwuhGYzs5rqQC3GjBkTrPfvn5/OhBYajOm1115r6v7KijOeAAAAAIComHgCAAAAAKJi4gkAAAAAiIqJJwAAAAAgqh4nnma2v5n90szWmdlvzOwfsvpwM1tmZhuyy/BKCECLI+NIHRlHysg3UkfGkYoiq9p2SrrQ3R83s6GSHjOzZZI+J+l+d7/SzGZJmiXpG/GG2pqeeOKJXO20004L9n76058u3Hvsscf2bmANMHfu3MK9HR0dudqbb77ZyOHERMYVXpFWktw9V7v55puDvV/60pdytaVLlwZ7zzzzzFzthBNOCPauX78+V3v22WeDvd/5zndytVWrVgV7Q6ZMmVK4t0TIeMkMHjw4WJ8xY0auduONNwZ716xZk6sdccQRvRtYayLfLSL0/wUagoxLGjlyZLA+cODAXK3aat8vv/xyrlbtd48DDzwwV7vvvvuCvdX2h3fq8Yynu29198ez669IWidptKSTJN2Std0i6eRYgwRiIuNIHRlHysg3UkfGkYqaPuNpZuMkTZT0sKRR7r5VqjwgJO3T6MEBzUbGkToyjpSRb6SOjKPMCk88zWyIpMWSvuru+fPU1W93npmtMrNVobdjAq2CjCN19WScfKMseA5H6sg4yq7QxNPMBqgS9IXufldWfsHM9st+vp+kbaHbuvuN7j7J3SdVe2820NfIOFJXb8bJN8qA53CkjowjBT0uLmRmJmm+pHXu/v0uP1oi6SxJV2aXP48ywhLatGlTsH7NNdfkatdee22wN7QAROj2UngRikMOOSTY+/zzz+dqF198cbB3x44duVolDnlTp07N1Xbbbbdgb6sh4xXV/m2vvvrqXG3dunXB3pUrV+ZqixcvDvaG6v37h5+SQgtVxVrE4pJLLomy3b5ExtvTT3/601wtxcWFyHfrqPb/SLU6iiHjFdUWXVu+fHmuNm3atGDvddddV6gW06hRo3K1ar//pKbIUU6VdKakNWb266x2kSohv9PMzpH035KmxxkiEB0ZR+rIOFJGvpE6Mo4k9DjxdPflkqq9VPWxxg4HaD4yjtSRcaSMfCN1ZBypqGlVWwAAAAAAasXEEwAAAAAQFRNPAAAAAEBU7bGEUkQTJ07M1R555JFg78knn5yrbd68Odi7du3aXO2jH/1osDe0umcjVpALbaPadr/97W/namVZ1RbdGzJkSK527733Bntvu+22XG3evHnB3tDqz52dnTWOrphjjz02WL/00ktztbFjx0YZA9AIoef7ais8hx6noedqoB47d+4sVJOkd72L8xyIZ8KECbnaTTfdFOw955xzcrVXX3214WOSpAEDBgTrd955Z642aNCgKGNoNTwTAAAAAACiYuIJAAAAAIiKiScAAAAAIComngAAAACAqFhcKILQh5wl6cknn8zV1q9fH+w96aSTcrWOjo7eDawBzj333GD96KOPbvJI0JeGDh0arF9wwQW5WmhRLUmaP39+rvbggw8Gezds2JCrhRYIqDa20LgkaeDAgcE60KpOP/30XK3aIhqh/1+2bNkS7B09enTvBoa2c+ihh+Zq1RYRasSCh0Atpk+fHqwfd9xxudqyZcuCvQsXLszVli5dGuz91re+laudccYZwd6DDz44WG8HnPEEAAAAAETFxBMAAAAAEBUTTwAAAABAVEw8AQAAAABRMfEEAAAAAETFqrZNtNdee+VqkydPDvYuX748V1u8eHGw191ztUWLFgV716xZk6t985vfDPaGViM98sgjg70DBgwI1oH3vOc9wfrs2bObPBKg/KZOnZqrHXPMMcHeFStW5GozZswI9t577725Gqs+ozvjx4/P1Xbu3BnsDa12+9prrwV7Bw8e3LuBAd3Yd999c7Uzzzwz2FutjvpxxhMAAAAAEBUTTwAAAABAVEw8AQAAAABRMfEEAAAAAETV4+JCZra/pFsl7Stpp6Qb3f2HZnaJpHMldWStF7n7PbEG2m4OOuigXO3rX/964dt/4xvfaORwkkbGkTLynZZ+/frlahdeeGGwN7S40AMPPBDs7ezszNXKsrgQGW8d11xzTbD+ta99LVfbsmVLsDe0aFG7I+NIRZFVbTslXejuj5vZUEmPmdmy7GdXu/tV8YYHNAUZR8rIN1JHxpE6Mo4k9DjxdPetkrZm118xs3WSRsceGNAsZBwpI99IHRlH6sg4UlHTZzzNbJykiZIezkozzWy1md1sZsOq3OY8M1tlZqs6OjpCLUDLIONIGflG6sg4UkfGUWaFJ55mNkTSYklfdfeXJV0v6SBJE1R5FeZ7odu5+43uPsndJ40cObIBQwbiIONIGflG6sg4UkfGUXaFJp5mNkCVoC9097skyd1fcPc33X2npJskTY43TCAuMo6UkW+kjowjdWQcKSiyqq1Jmi9pnbt/v0t9v+w955J0iqS1cYYIxEXGkTLynb7DDz88WJ88Of876KOPPhp7OE1HxlvHl7/85ZrqKIaMIxVFVrWdKulMSWvM7NdZ7SJJp5vZBEkuaaOkL0YZIRAfGUfKyDdSR8aROjKOJBRZ1Xa5JAv8iL8ThCSQcaSMfCN1ZBypI+NIRU2r2gIAAAAAUCsmngAAAACAqIp8xhMAALSo8ePHB+srV65s8kgAAKiOM54AAAAAgKiYeAIAAAAAomLiCQAAAACIioknAAAAACAqJp4AAAAAgKiYeAIAAAAAomLiCQAAAACIioknAAAAACAqc/fm7czsFUnrm7bD5hoh6cW+HkQkZTu2A9x9ZF/smIyXVtmOrU8yTr5Lq2zHxnN4HGXLQS3KdmxkPI6y5aAWZTu2YMb7N3kQ6919UpP32RRmtopjg8h4KaV8bA1Gvkso5WOLgIyXUMrHFgEZL6FUjo232gIAAAAAomLiCQAAAACIqtkTzxubvL9m4tggpX1fcWxI+X7i2CClfV9xbJDSvq84thbX1MWFAAAAAADth7faAgAAAACiasrE08yON7P1Zva0mc1qxj5jMrObzWybma3tUhtuZsvMbEN2Oawvx1gPM9vfzH5pZuvM7Ddm9g9ZvfTHFltKGU813xIZ7w0yXg5kvH5kvPWR7/qllG+JjJfx2KQmTDzNrJ+k6ySdIOlwSaeb2eGx9xvZAknH71KbJel+dx8v6f7s+7LplHShu/+FpCmS/j77t0rh2KJJMOMLlGa+JTJeFzJeKmS8DmS8NMh3HRLMt0TGy3hsTTnjOVnS0+7+rLu/IWmRpJOasN9o3P1BSdt3KZ8k6Zbs+i2STm7qoBrA3be6++PZ9VckrZM0WgkcW2RJZTzVfEtkvBfIeEmQ8bqR8RIg33VLKt8SGVcJj01qzsRztKRNXb7fnNVSM8rdt0qV0Ejap4/H0ytmNk7SREkPK7Fji6AdMp5cBsh4Tch4CZHxmpDxkiHfNWmHfEuJ5SDFjDdj4mmBGkvptjAzGyJpsaSvuvvLfT2eEiDjJUPGa0bGS4aM14yMlwj5rhn5LplUM96MiedmSft3+X6MpOebsN9me8HM9pOk7HJbH4+nLmY2QJWgL3T3u7JyEscWUTtkPJkMkPG6kPESIeN1IeMlQb7r0g75lhLJQcoZb8bE81FJ483svWY2UNJnJS1pwn6bbYmks7LrZ0n6eR+OpS5mZpLmS1rn7t/v8qPSH1tk7ZDxJDJAxutGxkuCjNeNjJcA+a5bO+RbSiAHqWfc3OOfaTezT0r6gaR+km5298uj7zQiM/sXSdMkjZD0gqQ5kv5N0p2Sxkr6b0nT3X3XDz23NDM7VtJDktZI2pmVL1LlveWlPrbYUsp4qvmWyHhvkPFyIOP1I+Otj3zXL6V8S2RcJTw2qUkTTwAAAABA+2rGW20BAAAAAG2MiScAAAAAIComngAAAACAqJh4AgAAAACiYuIJAAAAAIiqTyaeZvammf26y9esiPv6nJn9KLt+vpnNiLCPBWb2nJmdn32/m5ndYWZPm9nDZjauyu3OMrMN2ddZVXrem21jQ7bNgYGew8xspZn9ycz+sZtxHmVma7JxXZP9raBde5o9dsvG8rSZrTazv6yyrePNbH3WN6tLfaGZbTez06odd18g43++HRkn443YVztkvFc52aVnuJkty/a3zMyGVen7Zrad9Wb2iSo9jXxM/buZ/cHM7t6lTsbJeNe+Zmf8cjPbZGavhn7eiLG3YsbJ959vl3q+izyHf9jMHjezzu4y2pB8u3vTvyS92sR9fU7SjyLvY4Gk07p8f4GkG7Lrn5V0R+A2wyU9m10Oy64PC/TdKemz2fUbJH0p0LOPpA9KulzSP3YzzkckHS3JJN0r6YRAT7PH/slsLCZpiqSHAz39JD0j6UBJAyU9Kenwavd/K3yRcTLepYeM935f7ZDxXuekS988SbOy67MkzQ30HJ7dfjdJ78222y/Q15DHVPazj0n6tKS7e7r/W+GLjLdNxqdI2q+7f+9GjL3VMk6+2ybfRZ7Dx0k6UtKt1TLaqHy31FttzWyjmV2azbrXmNlhWX2Imf1zVlttZp/J6qdntbVmNrfLds42s6fM7AFJU7vUL7HsbImZ/crM5prZI1nvcVl9sJndme3njuwVjkk1HspJkm7Jrv+rpI8FXmH4hKRl7r7d3V+StEzS8bvcHybpo9k2lG3z5F135u7b3P1RSf9XbUBmtp+kPd19pVcScmtoW80ee7a/W73ivyS9OxtrV5MlPe3uz7r7G5IWZbcrHTJOxkXGyXj8nHQde3f7W+Tuf3L35yQ9nW2/69gb+ZiSu98v6ZXA7UuFjKeTcUly9/9y962B28cae0sj3+nku+hzuLtvdPfVknYG9lPr2LvVVxPP3e2dp/f/tsvPXnT3v5R0vaS33lI3W9IOdz/C3Y+U9J9m9h5Jc1UJwwRJHzSzk7M7+VJVQv7XqrwiUE1/d58s6auS5mS1CyS9lO3nMklHvdVsZv9UMPijJW2SJHfvlLRD0t7VejKbs1pXe0v6Q7aNaj1Fjc5u393+3jGuJo29yLaK9LQaMk7Ga9kWGSfjjczJqLd+kc4u9+nF/hr1mCojMp5+xotK8XmcfKef76LP4UU0JN/969x5b73u7hOq/Oyu7PIxSadm1/9KldPkkiR3f8nMPizpV+7eIVXeXyzpw1lL1/odkg4psK9x2fVjJf0w289aM1vdZb9fKHR0ldPZu/JIPUUV3Vazx97s+6FZyDgZj7GtVkLGy52TZo+9jMh4+hlv5LbK9jgg3+nnu+UeTy31VtvMn7LLN/X2xNhU7M58S9E7otq+emuzpP0lycz6S9pL0vZqPZkxkp7fpedFVU7V9++mp5Yxjelhf+8YV5PGXmRbRXrKhIy/jYwX7ykTMv62vsjJC2+9xSu73NaL/TXqMZUaMv62Mme8qHZ7Hiffbytzvos+hxfRkHy34sQzZKmkmW99Y5XVnR6W9BEzG2Fm/SSdLumBrD7NzPY2swGSpte4r+WS/ibbz+GSjqhjvEsknZVdP03Sf2bvre7qPyR93MyGZcfz8az2Z9ltfpltQ9k2f17HeN46Vf+KmU3J3q8+o8q2mj32JZJmWMUUVd7GsetnLR6VNN4qq4oNVOUVtyU9H3WpkPEKMk7G2zXjjcxJ17F3t7/PWmXlx/dKGq/KIhRdx97Ix1Q7IOMVpcl4DXgeJ9+ly3cNz+FFNCbf3qQVrbp+qfKqxq+7fF2Z1TdKGpFdn6TKaXpJGqLKh2vXqrKK0qlZ/e8krcnq87ps/2xJT6kS/h8qW0lL0iXKVsSU9CtJk7LrIyRtzK7vocoHiFdn+3xM0vjsZ//01m12OZ4FeudKWoMk/VSVD/o+IunAKvfD57OepyWdXaXnwGwbT2fb3C3Qs68qr0S8LOkP2fU9A32TsvvqGUk/kmSBnmaP3SRdl41pTej+zfo+mf2bPiPp4u7u/1b4IuNknIyT8WbnpEvP3pLul7Qhuxxepe/ibDvrFVjpMMJj6iFJHZJeV+Ux/AkyTsZbIOPzsjzuzC4viTH2Vss4+W6bfBd5Dv9glv3/lfQ/kn4TK9+WNSOTvWIzwN3/aGYHqfIPfohXVnCqdpsFqiwP/6/VehAP939tyHj5cP/XhoyXD/d/bch4+XD/F0e+y6fo/d9Xiwu1ssGSfpm9NcBU+Vs9VYOe2SHpMjMb4e43RB8h/swqH2Q/Rm8vc42ekfESIeN1IeMlQsbrQsZLhIzXjHyXSC355ownAAAAACCqsiwuBAAAAAAoKSaeAAAAAIComHgCAAAAAKJi4gkAAAAAiIqJJwAAAAAgKiaeAAAAAICo/h9VcrKcJyHcGgAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 1152x576 with 10 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "visualize_processed_sample(test_x_orig, test_y_orig, test_y, encoding = True, dataset=\"test\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "### Final Data forms\n",
    "\n",
    "**Raw data:** <br>train_x_orig, train_y_orig<br>dev_x_orig, dev_y_orig <br> test_x_orig, test_y_orig <br><br>\n",
    "**Processed data:**<br> train_x , train_y <br>  dev_x, dev_y <br>  test_x, test_y\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Loading Dataset form the image file"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from PIL import Image\n",
    "path = \"/home/befrenz/Sameer Kattel/Education/Course/CSIT/7th Semester/Project Work/Dataset/\" \\\n",
    "    \"mnist_png(white on black)/training/1/\"\n",
    "\n",
    "image = np.array(Image.open(path + '3.png' ))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "plt.imshow(image.squeeze(), cmap = \"Greys\",interpolation='nearest')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "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": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}