pain_2D_3layer.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n",
      "/opt/miniconda3/envs/trial_env/lib/python3.7/site-packages/tensorflow/python/framework/dtypes.py:526: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  _np_qint8 = np.dtype([(\"qint8\", np.int8, 1)])\n",
      "/opt/miniconda3/envs/trial_env/lib/python3.7/site-packages/tensorflow/python/framework/dtypes.py:527: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  _np_quint8 = np.dtype([(\"quint8\", np.uint8, 1)])\n",
      "/opt/miniconda3/envs/trial_env/lib/python3.7/site-packages/tensorflow/python/framework/dtypes.py:528: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  _np_qint16 = np.dtype([(\"qint16\", np.int16, 1)])\n",
      "/opt/miniconda3/envs/trial_env/lib/python3.7/site-packages/tensorflow/python/framework/dtypes.py:529: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  _np_quint16 = np.dtype([(\"quint16\", np.uint16, 1)])\n",
      "/opt/miniconda3/envs/trial_env/lib/python3.7/site-packages/tensorflow/python/framework/dtypes.py:530: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  _np_qint32 = np.dtype([(\"qint32\", np.int32, 1)])\n",
      "/opt/miniconda3/envs/trial_env/lib/python3.7/site-packages/tensorflow/python/framework/dtypes.py:535: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  np_resource = np.dtype([(\"resource\", np.ubyte, 1)])\n"
     ]
    }
   ],
   "source": [
    "import os\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import st_remux as st\n",
    "\n",
    "from sklearn.model_selection import train_test_split\n",
    "from keras.utils import to_categorical\n",
    "\n",
    "\n",
    "\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Config\n",
    "\n",
    "highpain = 'Pain_data/final_data/high_pain'\n",
    "lowpain= 'Pain_data/final_data/low_pain'\n",
    "nopain= 'Pain_data/final_data/no_pain'\n",
    "\n",
    "fs = 4000\n",
    "win_len = 40\n",
    "\n",
    "use_filter = 0   # weird results if we use python filtering\n",
    "\n",
    "use_matlab_backend = 0\n",
    "\n",
    "#high = np.genfromtxt(highpain, delimiter=',')\n",
    "#no = np.genfromtxt(nopain, delimiter=',')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Using native backend\n",
      "['.DS_Store', 'pain_large.csv']\n",
      "ydim=16 xdim=40\n",
      "['pain_low.csv']\n",
      "ydim=16 xdim=40\n",
      "['.DS_Store', 'nopain.csv']\n",
      "ydim=16 xdim=40\n"
     ]
    }
   ],
   "source": [
    "if use_matlab_backend:\n",
    "    print('Using MATLAB signal processing backend')\n",
    "    # high_dir = 'datasets/high/'\n",
    "    # low_dir = 'datasets/low/'\n",
    "\n",
    "    high_dir = 'EMG_data/power_spectrum_for_cnn_input/high/'\n",
    "    low_dir = 'EMG_data/power_spectrum_for_cnn_input/low/'\n",
    "\n",
    "    def generate_3darray(directory):\n",
    "        files = os.listdir(directory)\n",
    "        dims = np.genfromtxt(directory + files[0], delimiter=',').shape\n",
    "        length = len(files)\n",
    "        arr = np.zeros((length, dims[0], dims[1]))\n",
    "\n",
    "        for idx, val in enumerate(files):\n",
    "            arr[idx] = np.genfromtxt(directory + val, delimiter=',')\n",
    "\n",
    "        return arr\n",
    "\n",
    "    hi = generate_3darray(high_dir)\n",
    "    lo = generate_3darray(low_dir)\n",
    "    \n",
    "else: \n",
    "    print('Using native backend')\n",
    "    hi = st.raw_to_arr(highpain, fs, win_len, 0)\n",
    "    low= st.raw_to_arr(lowpain, fs, win_len, 0)\n",
    "    no = st.raw_to_arr(nopain, fs, win_len, 0)\n",
    "\n",
    "X = np.concatenate((hi, no, low))\n",
    "\n",
    "hi_Y = np.ones(len(hi), dtype=int)*2\n",
    "low_Y= np.ones(len(low), dtype=int)*1\n",
    "no_Y = np.zeros(len(no), dtype=int)\n",
    "\n",
    "Y = np.concatenate((hi_Y, no_Y, low_Y))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_X, test_X, train_Y, test_Y = train_test_split(X, Y, test_size=0.1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training data shape:  (2572, 16, 40) (2572,)\n",
      "Testing data shape:  (286, 16, 40) (286,)\n"
     ]
    }
   ],
   "source": [
    "print('Training data shape: ', train_X.shape, train_Y.shape)\n",
    "print('Testing data shape: ', test_X.shape, test_Y.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total number of outputs:  3\n",
      "Output classes:  [0 1 2]\n"
     ]
    }
   ],
   "source": [
    "classes = np.unique(train_Y)\n",
    "nClasses = len(classes)\n",
    "print('Total number of outputs: ', nClasses)\n",
    "print('Output classes: ', classes)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5,1,'Ground Truth: 2')"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x144 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=[10,2])\n",
    "\n",
    "# Display first image in training set\n",
    "plt.subplot(121)\n",
    "plt.pcolormesh(train_X[0,:,:], cmap='jet')\n",
    "plt.title(\"Ground Truth: {}\".format(train_Y[0]))\n",
    "\n",
    "# Display second image in testing set\n",
    "plt.subplot(122)\n",
    "plt.pcolormesh(test_X[1], cmap='jet')\n",
    "plt.title(\"Ground Truth: {}\".format(test_Y[1]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((2572, 16, 40, 1), (286, 16, 40, 1))"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ydim = train_X[0].shape[0]\n",
    "xdim = train_X[0].shape[1]\n",
    "train_X = train_X.reshape(-1, ydim, xdim, 1)\n",
    "test_X = test_X.reshape(-1, ydim, xdim, 1)\n",
    "train_X.shape, test_X.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_X = train_X.astype('float32')\n",
    "test_X = test_X.astype('float32')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Original label:  1\n",
      "After conversion to one-hot:  [0. 1. 0.]\n"
     ]
    }
   ],
   "source": [
    "# Change labels from categorical to one-hot encoding\n",
    "train_Y_one_hot = to_categorical(train_Y)\n",
    "test_Y_one_hot = to_categorical(test_Y)\n",
    "\n",
    "# See the difference?\n",
    "print('Original label: ', train_Y[0])\n",
    "print('After conversion to one-hot: ', train_Y_one_hot[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(286, 3)"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_Y_one_hot.shape\n",
    "test_Y_one_hot.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((2057, 16, 40, 1), (515, 16, 40, 1), (2057, 3), (515, 3))"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Partition data for 80% training and 20% validation\n",
    "# should reduce overfitting and boost test performance\n",
    "\n",
    "train_X, valid_X, train_label, valid_label = train_test_split(train_X, train_Y_one_hot, test_size=0.2)\n",
    "\n",
    "train_X.shape, valid_X.shape, train_label.shape, valid_label.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "# let's get down to business\n",
    "import keras\n",
    "from keras.models import Sequential, Input, Model\n",
    "from keras.layers import Dense, Dropout, Flatten\n",
    "from keras.layers import Conv2D, MaxPooling2D\n",
    "from keras.layers.normalization import BatchNormalization\n",
    "from keras.layers.advanced_activations import LeakyReLU"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Parameters\n",
    "batch_size = 32\n",
    "epochs = 50\n",
    "num_classes = nClasses"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From /opt/miniconda3/envs/trial_env/lib/python3.7/site-packages/tensorflow/python/ops/resource_variable_ops.py:435: colocate_with (from tensorflow.python.framework.ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Colocations handled automatically by placer.\n"
     ]
    }
   ],
   "source": [
    "pain_model = Sequential()\n",
    "pain_model.add(\n",
    "    Conv2D(\n",
    "        32, \n",
    "        kernel_size=(3, 3), \n",
    "        activation='linear', \n",
    "        input_shape=(ydim, xdim, 1), \n",
    "        padding='same'))\n",
    "pain_model.add(LeakyReLU(alpha=0.1))\n",
    "pain_model.add(MaxPooling2D(pool_size=(2, 2), padding='same'))\n",
    "pain_model.add(Dropout(0.25))\n",
    "pain_model.add(\n",
    "    Conv2D(\n",
    "        64,\n",
    "        (3, 3),\n",
    "        activation='linear',\n",
    "        padding='same'))\n",
    "pain_model.add(LeakyReLU(alpha=0.1))\n",
    "pain_model.add(MaxPooling2D(pool_size=(2, 2), padding='same'))\n",
    "pain_model.add(Dropout(0.25))\n",
    "pain_model.add(\n",
    "    Conv2D(\n",
    "        128,\n",
    "        (3, 3),\n",
    "        activation='linear',\n",
    "        padding='same'))\n",
    "pain_model.add(LeakyReLU(alpha=0.1))\n",
    "pain_model.add(MaxPooling2D(pool_size=(2, 2), padding='same'))\n",
    "pain_model.add(Dropout(0.4))\n",
    "pain_model.add(Flatten())\n",
    "pain_model.add(Dense(128, activation='linear'))\n",
    "pain_model.add(LeakyReLU(alpha=0.1))\n",
    "pain_model.add(Dropout(0.3))\n",
    "pain_model.add(Dense(num_classes, activation='softmax'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "pain_model.compile(\n",
    "    loss=keras.losses.categorical_crossentropy,\n",
    "    optimizer=keras.optimizers.Adam(),\n",
    "    metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_1\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d_1 (Conv2D)            (None, 16, 40, 32)        320       \n",
      "_________________________________________________________________\n",
      "leaky_re_lu_1 (LeakyReLU)    (None, 16, 40, 32)        0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_1 (MaxPooling2 (None, 8, 20, 32)         0         \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 8, 20, 32)         0         \n",
      "_________________________________________________________________\n",
      "conv2d_2 (Conv2D)            (None, 8, 20, 64)         18496     \n",
      "_________________________________________________________________\n",
      "leaky_re_lu_2 (LeakyReLU)    (None, 8, 20, 64)         0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_2 (MaxPooling2 (None, 4, 10, 64)         0         \n",
      "_________________________________________________________________\n",
      "dropout_2 (Dropout)          (None, 4, 10, 64)         0         \n",
      "_________________________________________________________________\n",
      "conv2d_3 (Conv2D)            (None, 4, 10, 128)        73856     \n",
      "_________________________________________________________________\n",
      "leaky_re_lu_3 (LeakyReLU)    (None, 4, 10, 128)        0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_3 (MaxPooling2 (None, 2, 5, 128)         0         \n",
      "_________________________________________________________________\n",
      "dropout_3 (Dropout)          (None, 2, 5, 128)         0         \n",
      "_________________________________________________________________\n",
      "flatten_1 (Flatten)          (None, 1280)              0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 128)               163968    \n",
      "_________________________________________________________________\n",
      "leaky_re_lu_4 (LeakyReLU)    (None, 128)               0         \n",
      "_________________________________________________________________\n",
      "dropout_4 (Dropout)          (None, 128)               0         \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 3)                 387       \n",
      "=================================================================\n",
      "Total params: 257,027\n",
      "Trainable params: 257,027\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "pain_model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From /opt/miniconda3/envs/trial_env/lib/python3.7/site-packages/tensorflow/python/ops/math_ops.py:3066: to_int32 (from tensorflow.python.ops.math_ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Use tf.cast instead.\n",
      "Train on 2057 samples, validate on 515 samples\n",
      "Epoch 1/50\n",
      "2057/2057 [==============================] - 4s 2ms/step - loss: 1.0979 - accuracy: 0.3495 - val_loss: 1.0967 - val_accuracy: 0.3340\n",
      "Epoch 2/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 1.0842 - accuracy: 0.3913 - val_loss: 1.0659 - val_accuracy: 0.4175\n",
      "Epoch 3/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 1.0782 - accuracy: 0.3967 - val_loss: 1.0604 - val_accuracy: 0.4583\n",
      "Epoch 4/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 1.0629 - accuracy: 0.4152 - val_loss: 1.0469 - val_accuracy: 0.4311\n",
      "Epoch 5/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 1.0492 - accuracy: 0.4409 - val_loss: 1.0304 - val_accuracy: 0.4544\n",
      "Epoch 6/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 1.0402 - accuracy: 0.4443 - val_loss: 1.0265 - val_accuracy: 0.4136\n",
      "Epoch 7/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 1.0358 - accuracy: 0.4453 - val_loss: 1.0251 - val_accuracy: 0.4466\n",
      "Epoch 8/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 1.0234 - accuracy: 0.4521 - val_loss: 1.0268 - val_accuracy: 0.4602\n",
      "Epoch 9/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 1.0228 - accuracy: 0.4507 - val_loss: 1.0135 - val_accuracy: 0.4388\n",
      "Epoch 10/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 1.0197 - accuracy: 0.4589 - val_loss: 1.0220 - val_accuracy: 0.4699\n",
      "Epoch 11/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 1.0290 - accuracy: 0.4589 - val_loss: 1.0169 - val_accuracy: 0.4466\n",
      "Epoch 12/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 1.0148 - accuracy: 0.4643 - val_loss: 1.0119 - val_accuracy: 0.4544\n",
      "Epoch 13/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 1.0202 - accuracy: 0.4487 - val_loss: 1.0170 - val_accuracy: 0.4466\n",
      "Epoch 14/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 1.0180 - accuracy: 0.4633 - val_loss: 1.0300 - val_accuracy: 0.4563\n",
      "Epoch 15/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 1.0139 - accuracy: 0.4716 - val_loss: 1.0131 - val_accuracy: 0.4621\n",
      "Epoch 16/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 1.0061 - accuracy: 0.4740 - val_loss: 1.0461 - val_accuracy: 0.4427\n",
      "Epoch 17/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 1.0091 - accuracy: 0.4604 - val_loss: 1.0209 - val_accuracy: 0.4680\n",
      "Epoch 18/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 1.0039 - accuracy: 0.4696 - val_loss: 1.0199 - val_accuracy: 0.4660\n",
      "Epoch 19/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 0.9978 - accuracy: 0.4696 - val_loss: 1.0126 - val_accuracy: 0.4641\n",
      "Epoch 20/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 0.9977 - accuracy: 0.4793 - val_loss: 1.0313 - val_accuracy: 0.4583\n",
      "Epoch 21/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 1.0019 - accuracy: 0.4750 - val_loss: 1.0164 - val_accuracy: 0.4660\n",
      "Epoch 22/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 0.9942 - accuracy: 0.4745 - val_loss: 1.0183 - val_accuracy: 0.4738\n",
      "Epoch 23/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 0.9977 - accuracy: 0.4696 - val_loss: 1.0353 - val_accuracy: 0.4602\n",
      "Epoch 24/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 0.9928 - accuracy: 0.4769 - val_loss: 1.0266 - val_accuracy: 0.4563\n",
      "Epoch 25/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 0.9874 - accuracy: 0.4852 - val_loss: 1.0266 - val_accuracy: 0.4699\n",
      "Epoch 26/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 0.9840 - accuracy: 0.4895 - val_loss: 1.0148 - val_accuracy: 0.4563\n",
      "Epoch 27/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 0.9793 - accuracy: 0.4895 - val_loss: 1.0270 - val_accuracy: 0.4718\n",
      "Epoch 28/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 0.9862 - accuracy: 0.4910 - val_loss: 1.0300 - val_accuracy: 0.4447\n",
      "Epoch 29/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 0.9664 - accuracy: 0.5012 - val_loss: 1.0339 - val_accuracy: 0.4369\n",
      "Epoch 30/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 0.9742 - accuracy: 0.5007 - val_loss: 1.0573 - val_accuracy: 0.4330\n",
      "Epoch 31/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 0.9528 - accuracy: 0.5080 - val_loss: 1.0556 - val_accuracy: 0.4544\n",
      "Epoch 32/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 0.9625 - accuracy: 0.5066 - val_loss: 1.0478 - val_accuracy: 0.4019\n",
      "Epoch 33/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 0.9634 - accuracy: 0.5134 - val_loss: 1.0775 - val_accuracy: 0.4408\n",
      "Epoch 34/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 0.9552 - accuracy: 0.5182 - val_loss: 1.0567 - val_accuracy: 0.4447\n",
      "Epoch 35/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 0.9551 - accuracy: 0.5100 - val_loss: 1.0488 - val_accuracy: 0.4505\n",
      "Epoch 36/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 0.9466 - accuracy: 0.5124 - val_loss: 1.0539 - val_accuracy: 0.3806\n",
      "Epoch 37/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 0.9439 - accuracy: 0.5241 - val_loss: 1.0622 - val_accuracy: 0.4563\n",
      "Epoch 38/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 0.9515 - accuracy: 0.5177 - val_loss: 1.0501 - val_accuracy: 0.4369\n",
      "Epoch 39/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 0.9452 - accuracy: 0.5177 - val_loss: 1.0576 - val_accuracy: 0.4194\n",
      "Epoch 40/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 0.9114 - accuracy: 0.5348 - val_loss: 1.0600 - val_accuracy: 0.4233\n",
      "Epoch 41/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 0.9241 - accuracy: 0.5416 - val_loss: 1.0573 - val_accuracy: 0.4136\n",
      "Epoch 42/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 0.9253 - accuracy: 0.5382 - val_loss: 1.0755 - val_accuracy: 0.4233\n",
      "Epoch 43/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 0.9216 - accuracy: 0.5479 - val_loss: 1.0630 - val_accuracy: 0.4427\n",
      "Epoch 44/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 0.9096 - accuracy: 0.5586 - val_loss: 1.0892 - val_accuracy: 0.4233\n",
      "Epoch 45/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 0.8833 - accuracy: 0.5664 - val_loss: 1.1105 - val_accuracy: 0.4330\n",
      "Epoch 46/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 0.9184 - accuracy: 0.5430 - val_loss: 1.0697 - val_accuracy: 0.4272\n",
      "Epoch 47/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 0.8825 - accuracy: 0.5790 - val_loss: 1.1132 - val_accuracy: 0.4194\n",
      "Epoch 48/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 0.8830 - accuracy: 0.5751 - val_loss: 1.0944 - val_accuracy: 0.4272\n",
      "Epoch 49/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 0.8707 - accuracy: 0.5819 - val_loss: 1.1112 - val_accuracy: 0.4252\n",
      "Epoch 50/50\n",
      "2057/2057 [==============================] - 3s 1ms/step - loss: 0.8703 - accuracy: 0.5819 - val_loss: 1.0900 - val_accuracy: 0.4311\n"
     ]
    }
   ],
   "source": [
    "pain_train = pain_model.fit(train_X,train_label,\n",
    "    batch_size=batch_size,\n",
    "    epochs=epochs,\n",
    "    verbose=1,\n",
    "    validation_data=(valid_X, valid_label))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_eval = pain_model.evaluate(test_X, test_Y_one_hot, verbose=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test loss:  1.0799371237521405\n",
      "Test accuracy:  0.4720279574394226\n"
     ]
    }
   ],
   "source": [
    "print('Test loss: ', test_eval[0])\n",
    "print('Test accuracy: ', test_eval[1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "accuracy = pain_train.history['accuracy']\n",
    "val_accuracy = pain_train.history['val_accuracy']\n",
    "loss = pain_train.history['loss']\n",
    "val_loss = pain_train.history['val_loss']\n",
    "epoch = range(len(accuracy))\n",
    "plt.plot(epoch, accuracy, 'bo', label='Training accuracy')\n",
    "plt.plot(epoch, val_accuracy, 'b', label='Validation accuracy')\n",
    "plt.title('Training and validation accuracy')\n",
    "plt.legend()\n",
    "plt.figure()\n",
    "plt.plot(epoch, loss, 'bo', label='Training loss')\n",
    "plt.plot(epoch, val_loss, 'b', label='Validation loss')\n",
    "plt.title('Training and validation loss')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "predicted_classes = pain_model.predict(test_X)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "predicted_classes = np.argmax(np.round(predicted_classes), axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((286,), (286,))"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "predicted_classes.shape, test_Y.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 141 correct labels\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 9 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "correct = np.where(predicted_classes==test_Y)[0]\n",
    "print(\"Found %d correct labels\" % len(correct))\n",
    "for i, correct, in enumerate(correct[:9]):\n",
    "    plt.subplot(3, 3, i+1)\n",
    "    plt.pcolormesh(test_X[correct].reshape(ydim, xdim), cmap='jet')\n",
    "    plt.title(\"Predicted {}, Class {}\".format(predicted_classes[correct], test_Y[correct]))\n",
    "    plt.tight_layout()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 145 incorrect labels\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAagAAAEYCAYAAAAJeGK1AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvIxREBQAAIABJREFUeJzsfXt4XlWV/ruaNGnTNE3TtOm9X2lpS4Fa2kJRsBapClUuoyKg+IMZERh1vOEo8zgz4HhDR1FnRh0Z1CIwiCIichEoUEG5ptKBll5obaChofe0SZsmTbt/f+y1zl5fspN8aW4n6XqfJ88+WWefs/c55z3nfPs9a61NzjkYDAaDwZA2DOrrDhgMBoPBEIO9oAwGg8GQStgLymAwGAyphL2gDAaDwZBK2AvKYDAYDKmEvaAMBoPBkEr0mxcUEWWIyBFRPv//EBFd3gvt3kBEt/d0O220vYKIruyLtvs7jC+GzuAY5csyIvpaX7SdK7r1BUVEVUTUQET1RLSNiH5ORMXd2YbAOXeuc+7WHPu0pCf6wPs/m4jWEdEBInqCiKZ0YtsCJuirRLSf+/ozIsr0VH/b6cvpRPQoEe0moh1E9GsiGtfDbRpf+i9fCojobu6DI6LFvdDmMcWXrp5j8vg0Ea1mvlTzfX1yT/Q3h/7MJaKVzP2VRDS3o216YgR1nnOuGMA8AKcC+OdIR4mI+s3orS0QUTmAewD8C4AyAJUA7urELu4GcD6ADwMYAeAtAFYCOLt7e5oTRgK4GUAGwBQAdQB+3gvtGl9yR5r4AgB/AnAZgDd7sc1jhi+MrpzjHwD4DIBPw/NtBoB7Aby323qXI4ioAMDvANwO/6y5FcDv2N42nHPd9gegCsAS9f+/A7ifl1cA+DqAPwNoADAd/ib7KYAaAG8A+BqAPK6fB+A7AHYC+CuATwJwAPLV/q5UbX0cwFr4B+sr8AS+DcARbq8ewBe57ukAngZQC+D/ACxW+5kK4I+8n0cB/BeA29s43qsAPK3+H8ZtzcrhXC3hupPaqZMcI4BpAB4HsIvPyR0ASlXdL/E5rAOwHsDZbD8N/kG4D8A2ADfleC3nAajrTn4YXwYmXwBU63NifOkevnTlHAM4HsBhAKe1U2cZgK/x8kgA9wPYAWAPL09Uda/g81QHYDOAj7B9Oh/PXj6Xd7XR1rv5GpCyvQ7gnHaPo6cIBGASgDUAvqou+OsATgSQD2Aw/Nv8J/A36hgAzwO4mutfA2Ad76cMwBNtEQjARXzwpwIgPmlT2iD1BPibdin8CPJd/P9oXv8MgJsAFAJYxBekrQfODwD8uIVtNYAP5HCubgTwxw7q6GOczn0tBDAawJMAvs/rZgLYAmA8/58BME0dz0d5uRjA6Tley88CeLY7+WF8GZh8QR+8oI4FvnTlHPPxvdZBnWUIL6hRAD4AoAjAcAC/BnAvrxsG/4NlJv8/DsCJvHwngC/zsQ4BcGYbbX0OwEMtbPcDuLa9Puaj+3EvETXDv1EfAPANtW6Zc24NABBRBYBz4X/VNQDYT0Tfg/+V+RMAH4K/obZw/W8CWNxGm1cC+LZz7gX+f2M7/bsMwIPOuQf5/0eJqBLAUiJ6Ap6ES5xzjQCeJKLft7OvYvhfHBp74S9wRxgF/8suJzjnNiIc1w4iugnA9fz/YXjCzyaiHc65KrXpIQDTiajcObcTwLMdtUVEcwD8K4ALcu1fF2B86ed86WUcS3zpCjrLl10AfiP/E9HX4V/agiMATiKi151zNWrfh+A/CYx3zlXDS5IxFMNfM40Oud8TOu2FzrlS59wU59wnmByCLWp5CvyvnBoiqiWiWnjijOH141vUf62dNicB2JRj/6YAuEja5HbPhP9VMB7AHufc/hzbrQdQ0sJWAv+rqCPs4jZzAhGNIaJfEtEbRLQPXsstB5KH0WcB3ABgO9cbz5t+DF57XkdELxDR+zpoZzqAhwB8xjn3VK796wKML/2YL32AY4kvXUFn+VJERD8hoteYL08CKCWiPO7vxfCjshoieoCIZvGmX4QfVT5PRGuI6O/aaOKouN/bHxKdWt4CoBFAOROu1DlX4pw7kdfXwBNDMLmd/W6B19w7alPq3qbaLHXODXPO3chtjiSiYTm2uwb+QzUAgLebxvaOsBzAaUQ0MYe6APBN+GOZ45wrgf+lRrLSOfe/zrkz4W8QB+BbbH/VOXcp/I35LQB3tzi+BOQ9ypbDyya35divnoTxJSB1fEkhBhpfuoLHAEwkogU51r8WXvpdyHxZxHYCAOfcw865d8G/9NYB+B+2v+mc+7hzbjyAqwH8iH/ktsQaAHOIiJRtDjrgfp95uvAw8REA3yWiEiIaRETTiOgdXOVXAD5NRBOJaCSA69rZ3S0AvkBE89mDZzoF991tAI5TdW8HcB4RvYeI8ohoCBEtJqKJzrnX4D8Qf4W8i+eZAM5rp93fwg97P0BEQ+BlsZecc+uAJMZhRRvHvxz+I+lvud/5RDSciK5p41fIcPhfIbVENAHAP8oKIppJRO8kokIAB+E/2h7mdZcR0Wjn3BH4j7aQdRq8z8cB/NA599/tHHOfwPiSLr5w3UI+DgAo4HNDsbq9jQHCl3bPMRFdQURVbRz/qwB+BOBObl+2vYSIYsc6HJ4HtURUhiAHg4gqiOh8frE2wvNK+HKR+tG0B/6FHePLCrZ/mo/pU2x/vL3j77GPmJF1K6C8Ytg2AsCP4T8A7gXwIoBLeF0+gO/BD1U3o2Mvm2vgvZHq4T88n8L2C+A/ntYC+ALbFsJ7nuyG/ybwAIDJvO44AE/xfjr0soH3rlrHF3cFgIxa91MAX29n2wIAX4HXtPfDD/dvUX1JjhH+4+9K7tcq+F881bxuDvwH4Do+pvsRPoDfDmA7b7cGXiKJ9eV6Pr/1+q87+WF8GTh8UdfPtfjLtFXf+HJUfGnzHMOHK9zRzrYE72a+BsABeEePuxAcHJYhOEmM52OuB7ABfjTk+DyNQ/DUq+V6s3m7b/N+6+Fl0Kva6c8pzMkGAH+Rc9jeH/GGhh4AEa2Cd9/d1dd9MaQfxhdDZ0BEj8B/K17b133pKdgLymAwGAypxECJtjYYDAbDAIO9oAwGg8GQStgLymAwGAypRE9kkmgTROUOmALksydqs/7+JfFaB5WtObYXLgcrGwcjDy8MtaYFT8cJeW8AAIYixPTtwUgAwM6tYxIbanbygo4nO8DlkGAaopYFSbf3RbbNU7YRvigLORJLpnpv3uP3hVjA+ldb7208R0/smxUSOL/aMNMv6Bht7j4VhHPgtnEf3qxVFTftdM6Nbn0w6QCVlDuMzmDEyD3BxmEntQfKQkUJdTyguSN80hyKfW+VW6BQ2fgEFkQ8ppt0jCX/vhs2NJh0hAvTckhp2GYc5/wsa1bXQWiiu18fsfH+9kwakZj+uoFDTtTtMHvqat/uqsbEVsNUqNCHNM//s2r/vGBbJ53R8ZMj1fLLqeZMEZErBTCe6bFtauhq9VYOOapRCRYm+FjWmWODn0HxBj4HTaGam+3LvxyYH4zr5aYrUj2Qe13fuXLSw3WDUEYPEY5w2Ryxxajr2lhOcEgty3XVcc2lXEaeZ9HnmCZPBQBg8PzAsTn7PO92vxpqbQa6xJdefUH5mMBngVK+m3bqE/hHLrVDyu7IPqTLFcrGoQ0Lj08sQ34THmqfK/EJj0/GS4ntLlwCAPjpv34y7Oart/CCzp7/Fy5PCKapalmQdHt5ZFudzeNcX7w3k1jO+MU9AIAHH/lAYvvze1rv7XpOkv+HZ0Ls3bkvcTaSh1TFd/licCaQrOk7/ND95j2q4gd6Koq9ezA6A3yjEosu/nViyuMQi3v/ckmodzWXlZo7K7jUHIr94JEXnY4t5Os/YXDLysDm59Q//JSZMyeYTlerz+Q9v/+FxHSdj4fFR3b8JtR7kUt1Yye3wzple6cvfv39RYnpQ++6zy+onAG/+oU/lhml4QfPN/hZ+jn1Hj5c6R9MZc9Vqv4LZ1eoht+vlqemmjOl8HS4nvN13/SLi5J11/7rD/3CV78eNviHLwMAbv7SwsS0aMnzfuH1UK2RE/gMWaXO1RmSpeg01QP5catvSOYJnRtMkodBvxvkR0ltxBajrrYditm3KSMfU9bzVUKwIs+z6HNMvy4+DwAor9ycWCof8eFgd7wn1Lqsi5kyTOIzGAwGQyphLyiDwWAwpBK9LPERgME+WUabXcmP2GLjW6X7y7BaKWmHm8N3n10YBQDYrmTBJvA3oIzep8yWXaVsUmFCMOnvAq2g+y/y0fDWtvJgKRSxW0nU8+R7k/7kwZnDDmd902LoeiwRNO1U39LksxUpuSbtIXCDAAwH6mIJj7WkMbzVArL50RIxjkXqN8eWtRwS2SbyHalBfaOoZd3/jdGjEtuE4yNxuVtbmyRrW6G+gf6Fy7HBVMTfDAZPDbbTVvmyRNlEBC84XknBxN+j3J9DxfxMWI7diinCMLDgxp+Ws86VZOE74cvB9lZf5OnsPHKOlBzanOd/yxfP3ZnY6keyRLZHf68RqOeF8ETftrFnSEOLUtc7hNbQCYU6/AYl94aWsmOPf9km8syC+u7Lt8F4TVSWK9UXzS7DRlAGg8FgSCXsBWUwGAyGVKJ3Jb5h8Gkq3+T/67SX1NAWJRB8Z5s7qMdSVsxbEkEi0lLRUHGdXKL2fSmfjj9k1Ma8rGdWEXVGy2ot+wIgDIlLWq8vRWuMD4tD2VHv3NVqPTspbtMejOJs81+qnuQWzqhx/wJ2Ef2YqncL0o0hAGa5LPnlQJZLr6oHIPs8Cz80x5ojNlmO8E7fHSLJDlNtyCaad1VqmbffdObsxPTicacAAEqVq1YphxlICQBnnLLSLzys9sftzMWqxPTNRZ8DAAxXbuFT17EbtaLJEpGPlUJZ0Ojlr/llKxPbM5ec5RfuvCZUVM5n6Knp9boJxQScUYjkXtKhJUPP9aJmw8jgNj9/kZcys6QqCRVQkvuw7d7f+6xJYQ6/33+BPQRvV5wQT8zm4BWYQKtmIt3VK5uoh1rii2p7HSHG84oWJRAeZBqyTSZiOzmYLvTFbLzSqtkTVOQOtnfU1/ZhIyiDwWAwpBL2gjIYDAZDKtG7Et8IAOfAz4YCZE+43CzDZD0sjXlYyfqInKOi+JsPBZcZkYUaEbI3VPDY87TJTye256/hAMiM2nV15DhkBF2lbBJk7fQplbF6Set6ykNI+vfGJOXZ9Q727NKj8ONlr+p8iCq0R3ld7TkDADBiYgjUy894mWxXsfIuSrnEN6RwP6YfV4lRCJ5TjUG7CUjk0oiUGvXm09coJhm33C/CfKpaphEeqMQEWfGtz3I5K0TgP/WJtwMAtirNeAu7Z2pvv4tH3wUAuPqynyS2yRs9ZyfvCLrJFaN/7rvVqDI/SOxkiFvHYLk35gZb/mEvWy1ECD5+5gaW+Bao+1Btk3aJL68IKDkZiZRZqy5iXj5LxbOCrC8S1ZhGHdQaAZ/TsyeFANYHPrYUAHCkXj14VnCpnxvSnKadSHw7dD3xBnxDGbP0vhY7in0iadmQoCxSr6TtTZuVFJjPy2eqehwnfzJeDjZxmFS8M4nPYDAYDAMS9oIyGAwGQyrR+xLf+xDyUG5S6zbHPN4kj5oe5orGosal7cRa+q39isPqcMWLagFCbq3Di7ws+PJJwVulaR33R8uR0m8dbCdKwh497JYhuwpwi3gaSiDxy8pLZsK7V/gFLfGxujUD6xPTlH/zydpem35GqLfEj7UXFATvLPHyWv+2mYkt7dNwlqAOZ2EFClSwZeKJWaw8FIexhKZzWTrhkaa4eERposiy4p1QbKKqJrKFCohN5FUt5zSos9rA7a0Oufo2/MUvb4DK3/djLlcE0zfu+jcAQN684MF43nSfd68cIbBXZOuGwiAPZhZ6zZF04lrZZHIw7Szy5NIyzUdn/A8AYMvnJyU27XF4L1KOYgBnAI6d6J7G25JV9TeyK6bynNt2k/ZqawHtPMxOfnPUuVpY4fPbPXP6WaGeluxa2iLB9GjWzzbhzkZlk/UxHU57tOr1QuCS9m1yv2gpu7xFCYRPHipF6UlzXuBVVcEony10LLv68nA0sBGUwWAwGFKJDl9QRPQzItpORKuVrYyIHiWiV7kc2d4+DMcWjDOGzsD4YmgLuUh8y+DDQH+hbNcBeMw5dyMRXcf/f6mjHRUWNWDivDXYtI6TYq1SKzfLUFvJYUm6eB2sVtK6ngxRlWfckcbwj8hC2qNHZIsxys1kPst9Y8qCR8+Wt3mpY9O+kMOq4Vm+V8KMHmFUvkePb+X0Kp1OJCKVR2sra3fPIQT3zTzDy3hTi5WLGEsTS3b/KTH9ouxyAMDL/08F0TG0NCOegpOUVtmDEt8ydANnhmE/3oanEwkUAGr4XA0adiCxHSllLUtLFXtikrHIJZr2ET7FYhrl8mvZb2eLEgC2qOsvqouWSyRIvUrZHuNys5oa5E++Pw/PC3MXSP5IHbgs3NH5GceXeT3qhEUhiFK8VnU9uR/qEOYXE7lvvpK+e0HiW4ZuesaIxPdKmXe7XI8Z2a0AWV8Mtt/ko0q1RDpsGN9oWo5nuU8H/ibB0bqeXBodgCunT0t8yW609+DuFqVejkl8bXnxCZfrIja1jeNtdP+F89oT7yQuwyw/GJ/lusqQHIZ6ypkuegp3OIJyzj2J1hMzXQDgVl6+FUlcscFgnDF0DsYXQ1s42m9QFc65GgDgckwH9Q0G44yhMzC+GHrei4+IrgJwFQAUTR6F+ajEptM5N9mzyu1KHM72aB7KjyodHSlyj5JkIrMxoD4cmgRC6nxlxVnDXw/x8qtQsp9MhVFYEuZ/fvEtPnLxSLVykxLPPu3RtUePkxmisChZqOYlPzZeMWdxYpO+LnlLCAycs2UDAICUNLpor/ckWpT/fDBKt5SyuO8kLw+NygsN/2fr3vU5NF/GTC5EKWqzAqxFYhk+Mly/veV8wNrjMZH4Yl5aWg4RvikpUGJo1dQUicQ3PTJXzElKW36fsgsFtcRXylrRWHXrfUcWFKdP9+3o/JEPwAeGrn79lFDvbt6P1mtP5fKc4IY2cXIVgOzpJ5qZjPq+mAnPsbnJNL/ZsnAaoTkzeTyAuUHKXJLop0Dt616aL0KQh6+AD3Qu36g0OZHpVC4+oYmWm5Nga50vU5IQbNSfJeT8xoL4tRefyG/6gSb70S6FrZ9d2ZyWbfS+97VYByScr1PbHo5UE29l9WxbP8NLp6tUFPfMqf6zxJx3boj07+hwtCOobUQ0DgC4bDNe2Dl3s3NugXNuQeHo2JvEcIwgJ85ovowYHZly3XCs4KieMaPL2qpl6I842hfUfQAu5+XLAfyue7pjGMAwzhg6A+OLoWOJj4juBLAYQDkRVQO4HsCNAH5FRB8D8DqAi3JpbCgacDJextPHeT2sepaa3THD5R4tyazjUv94iuTsE9lMzzCphqOvjPWS4raysG+RAA4fCV5NQwf5IXGF8qwZHhlOi7yUSEtAmLE2a8bRv3CpZtzcuNiXlaoaS0ArGs5JTFsXeu8skScB4G8m/RYAsGi/kvNEXtASj3gLqek7Ss7wEuXC89S2PYTu4kwTCrAFk7JkLvFky89XF1tWZ01hIjzRWoVch0g+Mu3EnExXEkyDZvqTOrMiyBeSIzB/fOhL85zWsx0nMyZDeeCpgFmZgjRPkUfkZi0prdzKblTL1G17vSx8O9hu+aIvl4R61ZP4XtN0Fporz8TVl3l9sGppJrE1ZAWEPoTuRnc+Yw4V5uHNqcXJfXsx7krWnQ8f6Kw9d0/cyNr8g2one7nUXxuO80WVIkXVAdaAtTfyRpHV7lBGuQ5a8hfSNkfqxQLJ6yL1YnlLgSDt6W1i089w9PYuRYAqLmtVtTVcBudhvFY1CwBw199drFrw0mnjVD1HjNY/O48OX1DOuUvbWHV2l1o2DFgYZwydgfHF0BYsk4TBYDAYUolezcWXj2aUY1cye2V1JiLxrVLeVG4KL8SC0NTXUBm16mGpGlk2ven3WVOv9i3qYZXahhXA6iWqX2/3Xk+jxgbvt4b6IjmggNIWJQDUiUYQGX6vjkyZESbrxIa5nLPt0pCzbdM8H3x4YNaPEts5WfM7MP7KpVab2Hlr+OGYB1A6sR/DUIn5yFfarcwmfKBeyU4SaJgl8cn5nYDW0F/Sef3E1iZtq6jwspCeeVWWhyqvsBi0xCceZDrgUzzrGlWk+XbWl3RgbXGpv3b105XX4Ee5XP3FYJNLXKU6kXjJKrkZv+IySMs45A965cSQ23HonPaPL02ow3CswOIkf6POvymyX5ZsLzkKdV5NUe6VxOeYMnpG57x8lue0l+ZI5t0elbgumT5DPweyvgUw5EGmJdWYt9/RzLIbm1aG++DU/jZLHzRP5FOFkgc3en5sGntiYnp46XvQGl2T+GwEZTAYDIZUwl5QBoPBYEglelXia0Y+dupoSi3JZLicpWxrRd7Sw1vuciw1lZb4dMZ68XDT03skctqPlfEK3o/euZdTdp2spKIhPNVDsaom/daj3DWsEVUrrUiUQp3Hb8+rvnxZzU65/LTsfQD4/fXekSlvYZC86mZ5b6CzZgV9MAk6VCpCI3uNvZI3WzWs3Y/ShyYUoCorWjYERzbUqpg6OU7NJ3F+3KKmtUi8qNT1laDcDFrb1P7E+05LPJIHTwcS63yPTcwdPV2IBIFrqVBy3RVEgmh1UOn4Ir/NhjNVx4qZ/Pq+EflaB43LIY9SMvfQK305U9WTmSNKA3lqs7XTVGMnynELrsR6PqjqV5Rcz/ftu48L0wJ/cuEPAQDnj3kk1JPbQjnp1o3w11hfj4oCfy33zlJzsMjMy5UZ1St5QMVmx9WQelrCk/b0Mym2n4hnalYeSpG1x0RssXhDvT+We0eqepKXT+XxE4/jFVDTj+C7kX3nDhtBGQwGgyGVsBeUwWAwGFKJXpX49qEEj2EJXt7HU0PoFF8ifWXNSMnSV9a0jCx5NagZZEWS0544GqJQzFc2SSGV9/fBJtKOUgVEOikoD14tzYe8/JJM8wCEPHCj1bbi3aNT78uy7qsT6UkF8smsDbovrAC9qPJfiSykZ+MdNZ0DSJX3mwS76sDftEt8zchHLUpxQMkNNUc4+rhaUVfk0jDRrJK3nlNG0XrnBVOtlgBbQEnG1a9nAABbC8cltiPb+PrroOsVallOr5KCVy/mBeXkNXSBP4AxJSFAPD8iKW7bxvJMjTp26aPmmChA2lFMHP+0t6Jwa66ysXQzaqKeBqL/oO7gCDy24b3Af7Phe9obzUtej/zy/MRScbE/zjFTQ/Du6fl84dT0GE15XuLTOTxFpt0wUXEoUZ71+RNJUSchyHCpPUpjnnYi0+njyFXi0/uOTD8jNNIKrjy/Jig5Tx5LEZ6MOOnNxCTeqlVZennXYCMog8FgMKQS9oIyGAwGQyrRqxJffV0JVjx+TpBftKedyBEZvQWPLWuV9CUj2YrW1XCisukJZtlLqeD4MEwex7OOlitdKOZNJYF+Oh+cyGTVe5X+JnKA9pySGLXNyiYeOvmRIbT2xDopYuOr9drrod2aYi95FQwJwaCHm70E2dgQvMuO7GepqF5NcYJrkWYcwmBsxXjUNQWNbG81X3jtsSlKpnKmSuSILWGWYjTycsx7VEtfMa/QKm880qxumSoutcSnVVO5/nrGUoHad8MqnwjwteLIrOba21NkcK0eSR/0vVQjPF+O1jgvLL6VOagDTTPc7I5wkgoqmtBfMGbIm/jwjBtx+CZ/DxTcFO5lmTbkFHWRFh7wuSkLn1U7EQdLdanLG72GesZbnk5s2yVofGmQYZ8vXuQXpqkH1Ar+jFCl2pDnnb4dW6dxDNx2kXVtgSI24aB+4sttpa+/3Af6WSqzu5weOjHjOO9xPE0RT2bZ1YHQ38+xy23BRlAGg8FgSCXsBWUwGAyGVKJXJT68DuBTiEseAu15JA5T2vFsVMQmitdJyjY37GjaZD/T4zQVqZth3U3PnlsUyakm0t5WNXeFTItQl1HS00bWl/RwWQfyJuB+He5gMj6RgLR0I8sHw2WTPINN2iNSUn9pj0iRCqJ9SicONRai+q/TgFqlWYiEWqUqiiOR9mQTjo1TNvFk056WGS6nKNvEFiUAjPVSkeTDA4C8M/21PHxJuB4H1Oykiayq+y/XNTg/BW/WF5RNqFqlbLKNjuMUGul7ahJ7apW+P9iElxlVT+6baco2rEXfAdRsVfO2pByTat/A9377T0Fq1ang2Cm44a/BVMPSfGSeWVToKVHYa236O4KG/4/n+Tmp3z71qcS2fJFPwL5iUQhWrWzy7sN7K5UGHcsFKhJuR16/sqzTaiqPwyjHGsTzTzVYx1JwjZIjqzK+1HzizyVDy4MuPRP+mXq2kpHfBi9/nloTTrpJfAaDwWAYkLAXlMFgMBhSid6V+A4eBNauVc1qPW83l9oN7jUudbQqB+iWq2Hp27nUw9LycGi1E71HUt2g4IlXz9JdkQp6k3xrhcqLT6ZS0LnTRAqUXFwAsHWpl0G2Lw25rrYd8H2sf1PlH9zBEtDeYEo8tfQwXYb2O5VNlvUpEtnvVWVrFi+uKmVkPWCPDuhLORoArKbsw5Dj1TMIy4yfNVqokbyG2uWNeUTKK/R0LrMkstY2mW5l3KDAA/EALShQXm5qpgQJMN6ldN+tTV5z3LtOyT0i2bwWTMm1zvIAfQmt0MxBonpWEfFgPEHZxBtUy5bSreLI1A9KRsaWwtbr04oGeFlPJD7Fk918LqvU4e5Ga0hI61A1kXfJ67yg5EGRD2dOXZ+YJEhV55DcUuAJtXeIuubSBy3niSSvnwNCaT1buGyrvxJo/kr+AH2tG/i+H3xC63r6c4l8JlF5EAre5Z8nC0tC0PtcPsGz8Upim93IyyqlaFdhIyiDwWAwpBL2gjIYDAZDKkHOdSYCrIuNjVjgcEZlkDRik0pq0VGGrSrlXRJkqT2xZOSsg3fHtV4/dGyIehxe0np2zQK0DkiUGU11TjQJHN27UwVW7mQZRHvOyLIO+Ix55cSQq/ja3KLUbeiUXSIb6HZ/SyudcwuQUtDkBQ5fqMw+pyKD6HPaiNaI5RmTZc0TkTe0HDLR3xPFY0MQd0WpcNd5AAAgAElEQVTRNt5FaFi4o3lTGOlMs4rAlCk49BQdDcytuC3Ia8LFw5GIzrwsDai1LbZtY5Nvr25PkL6P7OWbTZ/zGrV8Sbo5M2HBWPf3lZcl0r2+XuWsm45RnrsVLAGPOhyuddF+fz3z1D11YJj/Lb+9MJBnI7s/voyQi+85+GlynlYaWfWTLC3/QXX0US51kHeiZWtZOpaZIDJNhlbu5XmpZqRJuK+D2UUW1l9QIkkDRkz3ZJhUEFyFJSi3XH2DkHOtn6nfohu6xBcbQRkMBoMhlbAXlMFgMBhSiS5JfER0DoAfwGeRusU5d2O79ccucPhoZZBkYhJUTA7TNtlGKykyFNeyWCznlA5SHRKxyTZaQRGVJBYop/u1s0Wp+5rlK6TT5gsiqfBleK5H9iJr6mBgGbrr45AhvvbyiTlOfrP35ZrOcIZKFjgsqMyelkWCkLNmHJB/9EXng9cyh5zLcRFbbJoULS2LhNJekDmQfX6FJ4diFRViwbYxzsryiA76lR+5pw9Sdp+AwF8d0C1Kl/YU1cu39S5nOv2MoZMdcA+CS6SKwB7J+tXZaoN3cXlmME2Z7aNopyr3UfHi1ZKhSKhaNpXAfj0L8TYmmZ7qJgl+Xqc8JGMeubGAbVnWeRqzKsSmSpFnjNKyhzJ5tAeo3BuxhAP69hIeae7IvblO2VzX+HLUIygiygPwQwDnApgN4FIimt3+VoZjGcYZQ2dgfDF0ReI7DcBG59xfnXNNAH4J4ILu6ZZhgMI4Y+gMjC/HOLoSqDsB2eJLNYCFLSsR0VUAruJ/G/EdWt2yTj9EObLFvO5HXYsSyM7L1z2Y2e17bB8dcqYVX57oIl/qIsvdfx5zQc9zpnfQm5w5umcMZsQ5I5LY3cp2d+tqr7Uo+wg9zxdRxmP5PrsPXeJLV15QsVlHWonfzrmbAdwMAERUmWYX1VwxkI6jt5uM2LI4MxD5AgycY+llztgzpp+jq3zpisRXjewkGRMBlQ/IYGgN44yhMzC+HOPoygvqBQDHE9FUIioAcAmA+7qnW4YBCuOMoTMwvhzjOGqJzznXTESfAvAwvAvoz5xzazrY7OajbS9lsOM4ChwFZwbKeQYGzrH02nHYM2ZAoEvH0aupjgwGg8FgyBWWScJgMBgMqYS9oAwGg8GQSvTKC4qIziGi9US0kYiu6402uwtENImIniCitUS0hog+w/YyInqUiF7lcmRH+0oDiCiPiF4kovv5/6lE9Bwfx138MbrP0V85Y3zpGxhf0oHu5kuPv6AGQLqSZgDXOudOgJ9/9ZPc/+sAPOacOx7AY/x/f8BnkD0f7bcAfI+PYw+Aj/VJrxT6OWeML70M40uq0K186Y0RVL9OV+Kcq3HO/YWX6+BP/gT4Y7iVq90K4MK+6WHuIKKJAN4L4Bb+nwC8EyGePi3H0W85Y3zpExhfUoCe4EtvvKBi6UomtFE31SCiDIBTADwHoMI5VwN4kgEY03c9yxnfB/BFAEf4/1EAap1zkoM7LddmQHDG+NJrML6kA93Ol954QeWUriTtIKJiAL8B8FnnXGzOjFSDiN4HYLtzbqU2R6qm4dqktV85w/jSq0hrv3KG8SWOruTiyxX9Pl0JEQ2GJ88dzrl72LyNiMY552qIaByg5pFOJ84AcD4RLYWfPagE/hdPKRHl86+ctFybfs0Z40uvw/jS9+gRvvTGCKpfpythHfWnANY6525Sq+4DcDkvXw7gd73dt87AOfdPzrmJzrkM/DV43Dn3EQBPAPggV0vLcfRbzhhf+gTGlz5Gj/HFOdfjfwCWAtgAPz/kl3ujzW7s+5nww9KXAKziv6Xw+upj8PNfPgagrK/72oljWgzgfl4+DsDz8In2fw2gsK/71585Y3wxvhhfuo8vlurIYDAYDKmEZZIwGAwGQyphLyiDwWAwpBL2gjIYDAZDKmEvKIPBYDCkEvaCMhgMBkMqYS8og8FgMKQS9oIyGAwGQyphLyiDwWAwpBL2gjIYDAZDKmEvKIPBYDCkEvaCMhgMBkMqYS8og8FgMKQS/eIFRUQZInJElM//P0REl3e0XTe0ewMR3d7T7bTR9jIi+lpftD0QYJwxdAbGl3Si215QRFRFRA1EVE9E24jo5zxLZLfDOXeuc+7WHPu0pCf6QEQFRHQ3t+GIaHEntyci+jQRrSai/URUTUS/JqKTe6K/HfRlBhH9joh2ENFuInqYiGb2QrvGmc5tf0xz5hjky+lE9Cif3x18rcd1YvvU8IX7czMRrSeiI0R0RS7bdPcI6jznXDGAeQBOBfDPLSvwSesXI7cc8CcAlwF48yi2/QGAzwD4NIAyADMA3Avgvd3Wu9xRCj9B2kwAFfDzt/TWBGnGmdxhnDm2+DISwM0AMgCmAKgD8PNObJ8mvgDA/wH4BIC/5LxFN05SVQVgifr/3xEmrVoB4OsA/gygAcB0ACPgZ5KsAfAGgK8ByOP6eQC+A2AngL8C+CT8pF75an9XqrY+DmAt/AV8BZ68twE4wu3VA/gi1z0dwNMAavmELVb7mQrgj7yfRwH8F4Dbczj2ar2fHOofD+AwgNPaqbMMwNd4eSSA+wHsALCHlyequlfweaoDsBnAR9g+nY9nL5/Lu3LsXxmf71E9PLGZccY4Y3zJgS+87TwAdf2dL/A/0q7I6Th6gjwAJgFYA+Cr6mK/DuBEAPkABsO/yX8CYBiAMfC/wK7m+tcAWMf7KYOfNjhKHgAXMflOBUB8wqa0QegJAHbBz1g5CMC7+P/RvP4ZADcBKASwiC9GTzxsrgHwWgd1NHlGAfgAgCIAw+FnpryX1w0DsA/ATP5/HIATeflOAF/mYx0C4Mwc+3chgJru4oZxxjhjfOkaX3jbzwJ4tr/zBX34gqqH/9XwGoAfARiqLva/qboVABplPdsuBfAELz8O4Bq17t3tkOdhAJ/piND8/5cA3NaizsMALgcwGUAzgGFq3f/mQh50/mHz5Y6IpskTWTcXwB5Fnlom19AW9X4BLxFM7ETfJsLfjJd2FzeMM8YZ40uX+TIHwG4Abx8AfMn5BdXdOu2FzrlS59wU59wnnHMNat0WtTwF/hdODRHVElEt/C+dMbx+fIv6r7XT5iQAm3Ls3xQAF0mb3O6Z8L8IxsNfkP05ttsV7OI2cwIRFRHRT4joNSLaB+BJAKVElMf9vRj+F1MNET1ARLN40y/C/+J7nojWENHfddDOaACPAPiRc+7Ooziuo4FxJjcYZzyOOb4Q0XQAD8G/JJ/KsR+p5Etn0ZsfEp1a3gL/66acyVbqnCtxzp3I62vgSSGY3M5+twCYlkObUvc21Wapc26Yc+5GbnMkEQ3Lsd2u4DEAE4loQY71r4X/GL3QOVcCLw0Anhhwzj3snHsXPCHXAfgftr/pnPu4c248gKsB/IjJ3gpENBL+QXOfc+7rR3lc3Q3jTIBxpmMMOL4Q0RQAy+GlzNvaq9sCqePL0aBPPF2cczXwxP4uEZUQ0SAimkZE7+AqvwLwaSKayDfBde3s7hYAXyCi+ey9M50vKgBsA3Ccqns7gPOI6D1ElEdEQ4hoMRFNdM69BqASwFfIuwOfCeC89o6DiAqJaAj/W8D7I153BRFVtXH8r8LLE3dy+7LtJUQUO9bh8B9ia4moDMD1qg8VRHQ+k74RXgI5zOsuIqKJXHUP/M10OHIcJfAyxJ+dc+2d6z6DccY40xkMBL4Q0QR4KfKHzrn/jqzvN3zhugXMfQIwmPvT/jsoV90wB12xCkqLbbFuBZRHDNtGAPgxvBa/F8CLAC7hdfkAvgc/TN2Mjj1srgGwnk/cagCnsP0C+A+ntQC+wLaF8F4nu+E9Vh4AMJnXHQfgKd5Phx42fMyuxV+G1/0LgDva2ZbgXUDXADgAr+HfhfDxcRnCB8zxfMz1ADbA/1JxfJ7GIXjR1HK92bzdt3m/9fASxVVt9OVy3t9+rit/k7uLH8YZ44zxpXN8gX9JuBbnt16t7zd8Uee0JfcXt3fNiTc0dDOI6BF4zXhtX/fF0D9gnDF0BscCX+wFZTAYDIZUYiBEWxsMBoNhAMJeUAaDwWBIJewFZTAYDIZUIr83GxtJ5CYAGMKRCKuHzE7WNa4d6hcO1KstJJ5Ney3KO7VQ2Yp4VWErEwAfrgewR38LxGyxz3LadojLA8p2pJEXDkY2HhIWpY+6f3IV9GHK7poalfGIL6YOTSxjy7b6UuUezdvLnd0Ttjy4y5e71d5qgJ3OudGRDqcCRUSuFN7/VTCcHXo3jZya2GpfKfMLDeqCVPAJHqE2buJSn4R9cmH3KaNcw6i3bAfIU8ty3VUn+GzThLDvEXl7W+1l72G/jduq9rddFupUTeYECpRtKFpDSKsPXu6XpmAq93GsmSl/TUyj1gUirdyfbs6UELnRAEbO8P+vHq6eMa/weWnQ11rifHO91vqBIddmsLIVRGxyg0fGA5ouSRd0HK/0rzyYhFa6Cb1roXTsedKsrjVqudTPUuGO7lhe9irA+/wBmFj6emKqqNoBANi6K1Tr6jOmV19QE+ATPJ1wt/9/5qwQeL7h1Dl+ofLPaovnuNQ3pJylKco2zxfDjg+muWp1BZcxzsTOQHNk+ZCybeNylbLVvcoLGyM7VHFr0kfdP+GePkyJW9/8qjIyWb82J7Fc/uGvAAC+hG8ltpG/53r3hC3XLvPlr9Tebui5rAfdglJ4X9d3KNtiPsy/+eA3Etu9cy7xCy+rJMkfZU6cqzau5vKXyvaQXNg/KKOcc31BcoV+nZ6AVp34oC+G3Bge+meXPNhqLw/uWwoAaPjnkcH4n7KwQtWUB9gEZWN+aG43C2nvUMYMl28E04X/AAC44X8uTkyXvzWwhp5NN2dGA/gGgIt/7P+f+U71jJnD5+Xl5WqLl7jM9Vrrk1rC5Rhly3Cpr8coLtUTXnaj6SLvC/ecMr7M5ZXBJL/NdBP65SHPrFplE0rvrFbG37boMwDITBxlysbHOUuZbvDF587/ZGL6/N/+CADwlWVZ1brEF5P4DAaDwZBK9OoIashg4ITR8KkHARRpjSwZBEbe3FlDGvmpoIdDg1tX0xm6GluUQDhyPZIV6KFxc4tS71vbkv5kGVv3VVbr/smPN92/BJmwOJz3o0b7B1gr3CpjbgAjp/Lwa0aodwKvPmNrrI10YhiA0wDMUwqp0GO4/sW7mMud81rvRP+KrG+9OlyboRFb1hCknZ7qepqXvE9qXbWhPmi8u0pGoSX0+gSyH6fbaOeXf5SfJyibkPBvgulUXxRpggZqpR5FYD2FVatSTQBRUl7Wzxi57vqGbO9a5wq9DxmlK5408/XQXwSSzwi6f5G5BWV30ecFwvPrYMSW9W1BlB3NJ9l55BxoEx9Knjby+T2tjW4dDWwEZTAYDIZUwl5QBoPBYEglelXiQxGABUAzOzUVa3lirCxUqA1k+K27Kct6WMrQjgzaEUa+C8ZG7oURW2zorLtwsEWZ1fgutIbqmGyzP1JNS4sia52ijvN0Lk8KHRSJrxLzE9vWk3yW/QUnrUxsI/O8hLFET4ignTxSiJI8YEkxMFg7lPD36PFQWuWnuNQ5lEu5zFWli3IsV0RkPb2sJUpBbSDezvGtJb5kfbTP2gtNyK2/mDPfSPXLiWykJSNef0q45wZd4ImZdX61D0DKUTgYOH40knM+CjvDyiR3eVveBbkg5mmlbTENXzwndVt8PQ6qbWV3zcrZq+U6IDyfatrolkA/nxIeaflQ5F7dV8E2tcx8q84EE5/WXfp7AztvLMz5nusYNoIyGAwGQyphLyiDwWAwpBK9K/GVADgb2MZeS4U6QDCR+DrysIl5WDH0cHKHWhbpTAWuJsg1UFfHKxS2U68jyUC20f2rjdQTiUrHHpzpizHjw/B7K0+a+RsJsAGwkbWuDDYntq9/8Z8BAKeMUYmP/7b9rvY1aDgw+J0A1JRre07y5zdP6aHTZqwBAGzCiaGiKDv63LYrN0Skuah00xE0L4e2NslulPJUcyTiJifrtWwtUmGDvkdE7ot4EpYqUyIzTgw2Ubo+HkxnVawA0ELi66kpGHsCBfBSE9+jIzUBMrKgPyOIp7AOYD6a+DeBXDAtw+6O1OOT7zLBJLSLfXbQEG9UxaFofoCYh3KsD1n9k+UqZePnzZ4QG4ca39ldUPI0K5NlJ6lNu/gZwUZQBoPBYEgl7AVlMBgMhlSi9yW+dwPb2S1IyzSJxKeltDoZiushaDtefBramc6JRPiSMvIw3mmPHoFK/SIdqgvphVAfk/FEKjihnXVt9U8WlJQpbSh5SySbxqaQd+3lAt+v6ueUC9s/+WLtxFMSU/4v/Ln+8RV/H+r9bczjMEUohZ8Q++xgehA+BVClOjFb97FEFpM5tDIn67VslgS/xiS+mGdfJFoxJuvp5ZgHqJJndm2McFDWa89OkewaNMdE7lOylbQ3VlWTJiYpG3uFFl8WOrMAldxzFUSvZea0oxBekuRzMErfaHKLqOxR2CPnTXutxSLxY9c69gySbbRMKH2IPccyrU36GSi700HmyacK3efIvpu1d6iWhQWxIHW5OfRnlaGt6/GmjVqPFKVaOyGaxGcwGAyGgQh7QRkMBoMhlehQ4iOinwF4H4DtzrmT2FYG4C748WkVgA8552I+clk4NCQPb8wqxTaWI7LyOEm8l3awqRNdQnvEyDZ6WCqHobSbrHxlMsTW0p1ebgnl6ZZoI0qGcRPRGtKf2FA6Igk6rTNJBnSVyb3uQ76sVfIgqwZ73wzBcXubK1ptiqe4PDWY1mMmAGClCugFHon0tevoLs40jxqEnVcU4QGW9QDgJ7gGAPDMfWeFinK8+pKr+MEE0pqWzcTTqTmWi0/vMBbMKNDbxuZ5URCnMp1UOhbIK+tjsmVMMtY2yWupJT5RBXXQM3uFzi56JTFJ7rrt6kYc+Z7W04F0J7rzGYN8ZAUWl8a8+DLBhD1yX1cpY0OLEujwk0KCWC472U+O+4htmnXkQo51yqZnPRAuZJRNZn/Qsl9MthYu6235mablYV6dlU+Vc6x2Z+7GXEZQywCc08J2HYDHnHPHA3iM/zcYBMtgnDHkjmUwvhgi6PAF5Zx7Eq0d+S8AcCsv3wrgwm7ul6Efwzhj6AyML4a2cLTfoCqcczUAwGWb2bqI6CoiqiSiyl07opGthmMDOXHG+GJgHNUzZkd7Sqyh36HH3cydczcDuBkATl5Q4BowFAdY54xmkhinNt4oWmpsimbti9kRYi7gMfdygZ7RpKRF2RakX7Eo9MjcVVmQy6DaJW6vWFWTbyf16rIVs368RNn+5IsxC8N0zO/BwwBaaPIphObLxAUV7ue4GPfhvGT9M3fxt6d/VBttkRlIFwbb27nMqHrtXaJowteY+23M9TiWQFQt6+9IcvqrlC2WpKK6RQkAb8qC/kYq94YKg4glqRUeaRu3q7MByLdKPd/WK0Wz1Ub3Rjrbt9CcWTCOXPItBNnJYgtm+XPVNEvdyy9meEGfUxnMxd52bV3rlusjiWGzbBm0QvI9VNmSpNL6m7Vcm7bexrlms5DvpXrfsk9dn/uvP72zy/4Y7eou/R7RRreOAkc7gtpGROMAgMvt3dclwwCFccbQGRhfDEf9groPwOW8fDmA33VPdwwDGMYZQ2dgfDHk5GZ+J/yk2uVEVA3gegA3AvgVEX0MwOsALsqlsTwcRilqk+mkh8fmg8qoDZ6KRMknQ8qYbBaZWwUABvOQ/qAa2uf6eUMyDejEiyK1ZbmKS7+0VCDQ0iL3gdRwv5RDr7VbsETvv0XZxHW6OGgA4yZvAQBM4hIAZmI9AGAhnktsp/HyzMMbIv3rXnQXZ/ZiBO7DeXh629uCkWezz2ZuJHtHFZcx+UyrnInLeSwbROz20Nc8x+wCWuITxalK2WKSo9BJS3wJ36oiFdU9Up/xpT5O2aS+tW3TxpBkd+ti7yO8vmRmYqvIyljQ/RJfdz5jMAhexuQ5kyrUwGtSmb9HNs1SSYUlq8QeLfmLvKWvtSB2rWNZSHS4SVlrm8zVpRP6yu50swl3YkSOyYhtIeY633IdEI5du63P84XKKDJmjv98MAnhuRPNW9xFdLgr59ylbaw6uw274RiHccbQGRhfDG3BMkkYDAaDIZXo1WSx+c1HUL6jHqNGe51DzzkzapbP7LBrrhpqi1pWuVjtRXumCFg20459WhUUu/aISzqllmOj6FiyRpFparScI/JhZPrurKE9D8t1Ak5JYplRNlnWyoP0tTZ0uq7cH1xBUZgGXobdJ+PlxDbnwGoAQGFw7Es99jcPw/O7F+JIlXLLEklksao4kc+9vkaC2LTX+5UtkXpjknFMukGkXlueXc0t2kCYpltXizlWxmS/fG6n+VxlFD0oIm9rebCKSz0PmZwPlWe44UKveT1zbsjUUTA3ds+lFA7+uJgLWp6cxvrwprlK4pO5i57SMnHM401ksJgUrLkhDxstufHDSM89JzyOZRHRmU4S7sQ87fR16ciHRPoYe8jphyV/bhi+OJjew+XVwbQEywEAp+hssPJs0fdXF2EjKIPBYDCkEvaCMhgMBkMq0bvzQfHwW7z49Fwt4wZ5uW9XRmlaErSbpZpEksXK0FknCNWJDcdG1scCFwVaFhLZSMswItPo/dVkfLk/E2zSb+2pI6NpVS3payzB6Sa1LE55ai6h+gq/0Z/+5l2h2bd5jWCa2nhakV+eUJjyOaAUXEMemlaVqABVBNlK2+R8aIlPzn3M+zKagDUWqKuDs5sj9aSRmPwHRAMmd/I+G1U14WJMbtbKTlSlGdz2tjXaFpsTjRMmbwyB0FjJ+1NcbDrYUZB6inAEXmJilW6Mkr6mc1LmR+aqEziXT9xqdV2TOaJiyaV1Qgs5Lx0FeTM0F6ULMS5GA8ljAeL6uhyK1NWJi2P94shbPT+WSJ6Lle2Dvnj3nPsS04XszTlni/IK3sxlN+YWthGUwWAwGFIJe0EZDAaDIZXoVYnvSD6wvyy8E/VcIuUi92mZS2YTzpI0VnCpvGTcGVxPDbW1dCcSW3nEFpP4dHsy3NYegNKvmNeglm5a1tftlkbq6TlfRFLUEt8TsnCXMrK0tz+cjxcX+El/FhaEQN3Z8HP+DJ+s9QOVCzGNaIb3OtuobOI09AdlczIZlvZqEq+sTDAJ27WXVIJYUGZsYh4dtBqT9rQt5g3G+6xT9epit6G0rYMoIx57MekmNnX4EF4/TOUrlHmjlBdfsqxzYsYcv9KKI/D3IHuSaU9hkbxPmBy8W9eecYpf0Bx7lLmTNUdYhkt9PdqTeCNzSen9RWW8GGK8lP1ouTE2T7zmlciBkb7qBAESn70gmKbNWcOmysQmz5NE1gPCtHtRCf3oYCMog8FgMKQS9oIyGAwGQyrRqxLfoMNA0f4jGF7ox7c6j9Nc1m52LgqBrquLec7yt6qdrH6/L7UXV0z5ULGdyZBTeb8lQ2y9jexHy3TiGaa3leXqiC2W6kqPqmU4rVPXizw4WtlkWXsjypRtEy8Otrk+km/+cWHO98WsBb49mQs9nOvhe1Mu62kUIokbTHCwRQkAL7PEuxOtQRFbFB3lNZP1FZF1bUHIoIMoY7Jfy/pAIKj2upT9aDlHTpDS6ZozvtQy8jQudTyqBItriU/4WdqfdD2FPPh7n7tfujuc00yZ16MSeQrA2ree7BfWq0ehPFuSqTiAcG3eiNhiUrC+zsKdmAdoR9PAy0Mpxg0tae+LrNfyoEh8kWmG3lSclk8Lqrlmdj88rNwQm1DgF/RzVh7d5sVnMBgMhoEOe0EZDAaDIZXo3UDdJoA2A+PzvVSxqyRIfBl2B5mu3Na2nOT1rb3Vys1EZDWV5T0Zdetg2pjzS0ej6di2IvfpINAkP5bW+N5oUaoGG6YE02bWWOoindESn0iApwfTuHn+HGlvmvlY2comcumEzUoeEmemmJdhSlFctA9z5z2Khnkh4HDnxV5HqNk9PrE1vcryRZXaWJZfU7bYLLXCo52xoFwtm8itosnRUY46ucaxKRo6IqNso7VgkfNi7qNKphFvVe2dleFS54DkoMyCk8JxyJQUOohee9uu6KDXfY7B8LI4z+pKymNTPIUzyvVsxmQv9204Xc1InHBC7XeLSGP6msu93p5s2xZiEl+MLzEIPw9EbHr7WN7ASHt5rU0xejYqd+Ra1o/d1LCe5Bmj99dF2AjKYDAYDKmEvaAMBoPBkEr0rsSXB2AEUFfih5t1Sqpo4LxRB1T+qL07OUlUldqHxNg9pmxO5uXQs9lqr5dcvWjaS5+vbbGpNbQUkwN0SnqRnPTVkABiJdPsyvj2DpSFc3Q4Mp5ObDoIWZZ794p3CcNRhyVZFxpo4OuwsyxEXW9d6KNKqxYGvaFqX8bXX6USja1uUQLB000Ham4RL75YfjMtpcQ8rGIeXR0hJu0I32L51GKyUKSa9rASGmvPPubWmLLgZZjhm0172I5SWteK1q2lC4Ph1U45duXtOYaDrGer58Qu1kMPvzvcR5tG8HQcSr7Cs3zPr1KBzskpikn92nNTgrtjQdexKV207Wi+Swi0bJ3hUrlxjuN2zlDVlvhi0BnhASWfXfTUJQUc5N+knjGFEjfcjQ6gNoIyGAwGQyphLyiDwWAwpBK9Kvi4wUDjeKCRg7xKldvdDKxvVb90vE9O9/Kng4fN2vf5PHN4VkVgVvKwdZUavoZ0W8ob51VlFPeu2PQTWlYRbUQPlyUQrr0cakB85ktZpYJBxbNP5xIU9VAFJDe96fuwtSx4sFXx0F2fyzxONndgXJAKKsZ5yaHoQK6yUzohx1ao3BFl+pbhKsHZ8BK/3FCuJD6RS7V3lnhnxvKHVSvPOCfXS+fiE+8tnVhNX+uGFqVejs2Qiogttu9YfcWnmMQypEUJAEN8RX0u8+KJCvsPBsHLeyLxqVtUeKKlKpEy9fOn9lSvg+7aq4JaY+dU9l2l5H0nN65+rsR4kCtylX/1rN1cduTFyY9SKLHTL7sAACAASURBVNVy3MK2PYX1DN2S47AwFpQ7LGI7StgIymAwGAypRJdeUER0DhGtJ6KNRHRdd3XKMHBhnDF0BsaXYxtHLfERUR6AH8LP91AN4AUius8590pb2xwalI9tRWUhj5PCSJaoMsplTzxFKvSsmMd5d6vNx2USW9WF3t2mfp2aTyPsJixXq8RuO3k5JvfokXh7ikfMSy42zYf2nIoFUU5sUQIYOt3Lm+NLwnQBMQ+rJMeekoLk/NaqqTIP86UuKtLBfXrK1Z5HZznTjHxsx5hogOAu5UG5jYNUtxwIiQsTLmhVV1Scdcomy1XKlkx7oo3inaUlPpHcYkGSQHxm3rIW69qqJ9KykhnzWe7JqGqyrHMWxnLszfLR5WOOa82dMeqY5D4sULJfQ5aU1Hs4mmdMkotPLoO6NHKP6OOdyaTQz6TDg/zGL7875K2sGcEuffpejt3f6/haNmgPX/Hyi0m9sU8CsWBwHQws2+i8eupiO/4kckhxTGQ37WzMPCk+OTwEp7E7q34Oi5ynPyMMj80XIuc8NoXRUaIrI6jTAGx0zv3VOdcE4JcALuiebhkGKIwzhs7A+HKMoytOEhOQnXCoGlmf2zyI6CoAV/G/jVNo++qWdfohyhHPnd1tkN9Ser7CTbGKXcPMjqt0KzrkTEu+/IiuHQh8AbqDM/LDWcdsyfLy3HaxvY3lTqA3OXNUzxh6B9rgjNxVerUs/6Yr/ewJdJ0veuvlLUoFncXtTy3KbkCX+NKVF1RsIgPXyuDczQBuBgAiqnTOLWi1VT/DQDqO3m4yYsvizEDkCzBwjqWXOWPPmH6OrvKlKxJfNbJnK5oIqPmVDYbWMM4YOgPjyzGOrrygXgBwPBFNJaICAJcAuK97umUYoDDOGDoD48sxjqOW+JxzzUT0KQAPw/vO/Mw5t6aDzW4+2vZSBjuOo8BRcGagnGdg4BxLrx2HPWMGBLp0HORcK0nXYDAYDIY+h2WSMBgMBkMqYS8og8FgMKQSvfKC6s/pSohoEhE9QURriWgNEX2G7WVE9CgRvcrlyI72lQYQUR4RvUhE9/P/U4noOT6Ou/hjdJ+jv3LG+NI3ML6kA93Nlx5/Qal0JecCmA3gUiKa3dPtdiOaAVzrnDsBwOkAPsn9vw7AY8654+GnT+wvN8VnkD2z47cAfI+PYw+Aj/VJrxT6OWeML70M40uq0K186Y0RVL9OV+Kcq3HO/YWX6+BP/gT4Y7iVq90K4MK+6WHuIKKJAN4L4Bb+nwC8E8DdXCUtx9FvOWN86RMYX1KAnuBLb7ygYulKJrRRN9UgogyAUwA8B6DCOVcDeJIBGNP2lqnB9wF8EcAR/n8UgFrnnCTRScu1GRCcMb70Gowv6UC386U3XlA5pStJO4ioGD5h12edc5EZCNMNInofgO3OuZXaHKmahmuT1n7lDONLryKt/coZxpc4emNG3X6froSIBsOT5w7n3D1s3kZE45xzNUQ0Dkede7PXcAaA84loKXxC/BL4XzylRJTPv3LScm36NWeML70O40vfo0f40hsjqH6droR11J8CWOucu0mtug/A5bx8OYDf9XbfOgPn3D855yY65zLw1+Bx59xHADwB4INcLS3H0W85Y3zpExhf+hg9xhfnXI//AVgKYAP8jBFf7o02u7HvZ8IPS18CsIr/lsLrq4/BT4n3GICyvu5rJ45pMYD7efk4AM/DT9zwawCFfd2//swZ44vxxfjSfXyxVEcGg8FgSCUsk4TBYDAYUgl7QRkMBoMhlbAXlMFgMBhSCXtBGQwGgyGVsBeUwWAwGFIJe0EZDAaDIZWwF5TBYDAYUgl7QRkMBoMhlbAXlMFgMBhSCXtBGQwGgyGVsBeUwWAwGFIJe0EZDAaDIZXoNy8oIsoQkSOifP7/ISK6vKPtuqHdG4jo9p5up422VxDRlX3Rdn+H8cXQGRhf0olufUERURURNRBRPRFtI6Kf80yR3Q7n3LnOuVtz7NOSnugD7/9sIlpHRAeI6AkimtKJbQuYoK8S0X7u68946udeBRGdTkSPEtFuItpBRL/midJ6sk3ji/GlM20aX44xvvTECOo851wxgHkATgXwz5HOEhH1m9FbWyCicgD3APgXAGUAKgHc1Yld3A3gfAAfBjACwFsArARwdvf2NCeMBHAzgAyAKQDqAPy8F9o1vuQO44vx5djiSzdPVFUFYIn6/98RJq5aAeDrAP4MoAHAdPiT9lMANQDeAPA1AHlcPw/AdwDsBPBXAJ+En9grX+3vStXWxwGs5QN/BZ7AtwE4wu3VA/gi1z0dwNMAagH8H4DFaj9TAfyR9/MogP8CcHsbx3sVgKfV/8O4rVk5nKslXHdSO3WSYwQwDcDjAHbxObkDQKmq+yU+h3UA1gM4m+2nwRN7H4BtAG7K8VrOA1DXwxObGV+ML8YX40vb9XqKQAAmAVgD4KvqZLwO4EQA+QAGA7gXwE/4xI+Bn3nxaq5/DYB1vJ8y+KmDowQCcBGfvFMBEJNzShuknsAXYSn8CPJd/P9oXv8MgJsAFAJYxBekLQL9AMCPW9hWA/hADufqRgB/7KCOPsbp3NdCAKMBPAng+7xuJoAtAMbz/xkA09TxfJSXiwGcnuO1/CyAZ7uTH8YX44vxxfjSGb7ko/txLxE1A9gL4AEA31Drljnn1gAAEVUAOBf+Ld0AYD8RfQ/+V8NPAHyIT9AWrv9N+KmEY7gSwLedcy/w/xvb6d9lAB50zj3I/z9KRJUAlhLRE/AkXOKcawTwJBH9vp19FQPY0cK2F8DwdrYRjIL/ZZcTnHMbEY5rBxHdBOB6/v8wPLFmE9EO51yV2vQQgOlEVO6c2wng2Y7aIqI5AP4VwAW59q8LML4YXzoD48sxxJee0GkvdM6VOuemOOc+weQQbFHLU+B/5dQQUS0R1cITZwyvH9+i/mvttDkJwKYc+zcFwEXSJrd7JoBx3OYe59z+HNutB1DSwlYC/6uoI+ziNnMCEY0hol8S0RtEtA/A7QDKgYRcnwVwA4DtXG88b/oxADMArCOiF4jofR20Mx3AQwA+45x7Ktf+dQHGF+NLZ2B8OYb40tsfEp1a3gKgEUA5E67UOVfinDuR19fAE0MwuZ39boHXUDtqU+reptosdc4Nc87dyG2OJKJhOba7Bv7DIwCAt5vG9o6wHMBpRDQxh7oA8E34Y5njnCuB/6VGstI597/OuTPhbxAH4Ftsf9U5dyn8jfktAHe3OL4E7CG0HF42uS3HfvUkjC8BxpeOYXwJGBB86TNPF+dcDYBHAHyXiEqIaBARTSOid3CVXwH4NBFNJKKRAK5rZ3e3APgCEc1nD57pFNwxtwE4TtW9HcB5RPQeIsojoiFEtJiIJjrnXoP/4PcV8i6aZwI4r512fwvgJCL6ABENgR+2vuScWwckMQ4r2jj+5fAfSX/L/c4nouFEdA0R/V1kk+Hwv6hqiWgCgH+UFUQ0k4jeSUSFAA7Cfxw9zOsuI6LRzrkj8B9tIes0eJ+PA/ihc+6/2znmPoHxxfjSGRhfBghfOvpI1Zk/tPhg2GLdCiivGLaNAPBjANXw2uqLAC7hdfkAvgc/VN2Mjr1sroH3LqmH/5B4CtsvgP94WgvgC2xbCO9Jsxte430AwGRedxyAp3g/7XrZcP0l8B9bG7hPGbXupwC+3s62BQC+Aq/97ocf7t+i+pIcI/zH35Xcr1UArgVQzevmwH8AruNjuh/hg+btALbzdmvgJZJYX67n81uv/7qTH8YX44vxxfjSGb4Qb2zoARDRKnh3zF193RdD+mF8MXQGxwJf7AVlMBgMhlSi30dbGwwGg2Fgwl5QBoPBYEgl7AVlMBgMhlSiJzJJtIm88pEuPzMBjt3rDx0sDCv3cLlVeyg2+2Kkqicx1EODKb+4ye9feTc2blMVqncCAOgtIxPT7PxXAABDqpsSW8O27K4APgcKAAx5S7Ct3DXfL7wRbEPmHWjVh/1rubPqZ0DpzN2+Pg4mth1uNADg8LrBoeKBrb4sHp+YCmf6mMQMqhJbcbNvtzY/xPNtOni8X1ij4gGHcmhCRh3c2pU7nePGUwgqKHcYkkHejEOJrZQ9Wfep+MVDqwpbbZs/11/XYtQnttp6vprrdWynnHNlG+6vW2bGXxPTqDeYFUdCtYZJvt1X6k8KxvXbwvLYCt/nCbsTUwn2AQAOoCix7dzPsaObVbcaa9EKBaW+VNE7haW+340HFd/lup8YwlHm718JANhWFaoJU8eXB9uWKRMAANs3jg3GvfqeXJVqzowsz3PjM4PxysoZAIDC+aHvJ9X4e/7g1lB/CN+i+2aEpOivbp7pF3YH3uFEz5MhQ8I9dXAfn9/XVQca+ZmlH61yGTRN5XKNCj4AJYP3tjqeffWlrdsQqo4IJpoajjOTVwUAKNseOHSQQ5J1hO/oAl8mkWEA1gya7Q9js+KT3ELqqo8b6x9+4197M7Ft9Y/ZrDQXG4Au8aVXX1D5mQkYV3k3mvhK1WyYGlb+ksvr96kt+GY/+/hgOotL9UwoW+Sv3nB1+jfdpM76tbcAAAY//qHEdHfZHADArC+EQO6XvuvL36oeyBazHg+2Qcsq/YLKozy18kUA4QEKAM+cyp1V1/rsJ+8AAMzE+sT2w6ZPAAD2nqEeCpVf8eUp1yemKU++BAC4GR9PbIt2PA8AuG/06Yntgg0P+4WZz4X9TV/oy2Xq4OZTe1HsfY8hGWBhJUofDr8EzhvkM8MsR5jhoLpseqtNyyo9J96GpxPbvU9f4hfOeEnVlDhGZVu4GABww6MXJ6bLv/wrv6De+S9934fCvOXpymA846aw/LefBwAs/sYvE9N74K9NJRYktp8+90m/cJnq1sZ7Wh0TJrzfl99WvX+/j9nc9Iri+4l83e9ZmJgqn/Y/Cm/621BN7pbrLwy2T//PZwAA/3n+Pwbj7/U9OSLVnBmfGYw7K6fgLXQfAGBiZbhglV/1D421/xrqn3CaL/+wPFyPcz/8hF+4U/3YuMf/2Jg+44XEtPqRU/3CJ1UHNsqPkbJgm8Olpqlcro82JqbTxz/c6ngeefJ8v/ApZXyZy0XBNOT28LP6GyU+zOmS/7g3sa31lxUr1G7+nh83jX9SXS26EwCw4cNzglGSF10dTFd+6csAgH/7eMg09RX/mMU7QjWc1X6mjA5hEp/BYDAYUgl7QRkMBoMhlehVie+wy0NtUyma5NvTnkilcpUbcVfLPIkIuoSS6Gt3s06rRtVQahk+6Gc1Hle2Tu2GlVKVCWsOf+55Q2nUJ0hqSSXDF1/jxdb6Q0G8P4w8AECD1vMulQ2CqQle+N2JUWHbZr4MGdXnSpb2lCwg37dqUZrY3hg9Kvt4AAwd609sw4Ig8ST7Dp++0o88AMOBRvWtcleRP1453wCAD3CprtHhI3597aBwrjCEy0lKvtgigv7iYLvIF4UI3yeTZDZK4mvka1kwK0hgTW//fKjA105/b9oGLxXVaVIISvU/J7ReH9kkue76Tv4Hf93fOuOJYPNKMBQjkPR6TLAlMnlWBjf9zS7dKEATJmELsD4DADgbPwgr+flwgs5+x9/zChGktkRqPVgRbPxpaRfUBzu5l9SnY+BAdmNA+IZTr6oJj2oVt8ePQis0tzbFcLg53A8JJ9S3SjnmQ/pbFn85qSsKxMprnaUoPHPVcSb1FHdOk7Zy63JOsBGUwWAwGFIJe0EZDAaDIZXoXYmvaTD2VlcAtZzFXWeQEnlDeRQlw2AtfYhDlxqJNmW8FDi0LGhzEz8c5hTL+7AfJ78XDya2ehkG61zCvPm59ymbjNRfDKab3u1lnOVfOjuxVWA7WmLM571NywcnswuOln1Ki7xeWX+Wkg9EsTszmIpYPtijTsgulgr10HxhiffiWvGDc8LG7AIak4lSCwKQBxQOCeevgGW32XglsdX9oPX8bScP8ud5lCLZxHmeE9XXKd10OUuyyiv0pKu8p9bJibuU7weALEmjiKWvpWWBV/f+8m+S5VFj/UnX3qUiz9bF5pzTs/cQCyU6E5koTkOCSfY9Tcl55f/h2/0cvhcqvuqLE2J3/Ci9yOerXFe4JbJROpFX4zDyqw3Y8S/+/JbXKF1N3Pj1JBt8TmdgQ2L60NJbAQArlp6V2MQ7N0talnOppTss5/LcYKrhRrRSKvLg3GDaNtvXK0pkQlWvWrexjPd7RWJpqg6fQ9aXeRf75vB4Qj7Tcs4qtRt2t9uiTkjy+UDX28kepdXvT0zJp4wQBZNIe90pCNsIymAwGAyphL2gDAaDwZBK9KrEh0YAG8nPbgJke6iIavV2ZZOh80Zli814P8sX5XOCnLMQIUh1Es/srINoX4GPmN48NQQLL73xAQDA2KkqonstlyrK/+Nb/GSQSyYtT2zi0aUhAclaFhBpRwdqlrP+Vp1R0pN0VXsjMuqVPCRDci3xify1821Bu9m0z++74c2QTSP1KAFwLpAZVJWY5FrqQOeZRX5ZnwORqvIUyUbxeX7+E8GXbcs1Xt44YVCQDK/BTwAAJ25Us3zLrlX0vki3Wgo8MD5ItwIt8R1gaSRL4pN960wDSklKUNqiRMgqIgHAennO40G2Ev6Wqdj4RNpTxyTnF1cobbHyy2H5IRWdnkK4bcCh7wLllfzwUMeWfDJQspRIthN2hGfHR0b/LwBgppL9tnHFDZiZ2DaN5Whb5eyHLeLFq4Kbd3KFncq2kSW58BjA9sW+jTFl6nOByLnaq3KP9sWU/YXFx+b4IPY7SgIvL7+WA81V7L48N9erYzrQxP3PknhZrtTHKVC2DPPpz904+YeNoAwGg8GQStgLymAwGAypRO9KfHXwyaBW8P/D1DpJrXaysonH2Wple0Z8RFoPobXH1nysTJbHs3ueyHoAcBs+CgBY9UrIYXfe7F8DAL559T8lthOfZJlHD1vZI2rq6zXBJt3R9cT5TAcQc3PbxoWxcSJNrVP1/sClCqzdutRrE1uVRiHHrL0CZX9FMX+a6tamtKK4fB8W/N0fMB8h1910+OsxVHk6HWYaazlPZDUtry1gTmg5LH+QP1fTEOS8E9fxckxOVhLZRo7E1V5Q25Wbn1wH7cUptqyAT5FzVeo3NIu2rG7R/RxZGZH4dM7BOatZmtL3jUhdOlGa0Eh5tZ3MOQmvPe7riW3Fg8GbbSUh1Th0GHhjL5CRY1femRKYmhVJyhKbBLwDgVvTlG6mPw8INs7z17/6g1qa5/yNVapisySd1fkVOVi8el5iaar393BhmfJanesfLE2fUkkLnuUDUN6c+lqvKvYPmR+9OyQJbOLExgsnBY1PZOaX1UO3qMA/M/Zeofb9Xpb9VN7A5Fmrn23MozKT+AwGg8Ew0NHhC4qIfkZE24lotbKVEdGjRPQql/3oy7uhp2GcMXQGxhdDW8hF4lsG4L8A/ELZrgPwmHPuRiK6jv//Uod72gvgfgAv/5kNp4V1x3Oip1NUfVFBtCfbJB5uFqucdzxX03Q1JJ+kJlDJZ1llE6YltlU3s9Z29YrE9vsf+iRs4z4RAn4/suh/eX9bEptIAGNrlLefHJIOcJMuaO8XljXLx+1MTIkH4DOq3lM8bUN9yO1Wc5nXl9bPCF43sfx84iH4/Fbl7VNZ2Lp/PYdl6AbOVGBbdrApgEZ2ddPHK8sFSkqTPHql6rpNZd0lK3hTpnzSczHJ5dd3BwdUrpkVOHQ3PggAWPbKNaHecrUNKz8bl4Q5cyoKvIfWtiYV8asn6Ukg/FbSDh/yoFH7lWkPl0qCkn5reUuooPLQNbLs15wXfqfKfv4fbktsS/BYsqzCT7sTy9BNz5iCEUDmHQDOYIMKVn1zvj/gtUrqFw+2jerZUAv/LtTysJyX8Qiy/tv4pl/xpeDBuz3DJ1jPnPGH/9/euQdXcV93/PtDQkhEyCooCAnhXGzZlWlqSqMYP6jL2M6UOB7sZui0qV0/Jh5npumMO1Onec3E7Uwzddq0SZs07TBpHE+dNm2Sxo/EjlsTiM3AQIWhgMvDYESMEcg8JPAgUAS//vE7Z3/f1V2h1+VqJc5nxrPL7+69d3fv2V3/vvqec+Te1rUyjs0SbSxeypjTEjRe7vdWmB3W9zwSNzy0uLX4O56h9e+HxebHYj+OM/eHeLqVMoP13sF/+jh6VOKSXXxy/21qpiKlCnVC0nN9HTsnf1b8ltEw7AzKe/8KgBODhu8G8JSsP4V0/QfjMsdixhgNFi/GUIz1b1CN3vsuAJDl3KE2dM494pzrcM51YOCdMX6dMQUYUcxwvJx6pz9rE+PyYEz3GAuZqcUld/F571cDWA0ArrrdB2OZzL+5JpbKeJSQVluQthatlNS4Kkg8V7RG2eQjVaEWGruzFiUZtrG1QRPiFHXa3UEmuXBkefxs6eTOyXjfwe8DSEtKfeKY+9WmmAj30KonAQCtbJOj1gyDx1iSUWfazj/4YNxuukh7bfTe6uBSYxeftu/YR305dm0QnZRbA6tJLeftNjheFrXX+AV4K3W8exDqjG2mxhEsUSiaPLscsUaduhobzpLEpzmR3G1b3aXk9nr7tuDyUlkPAJ48+mBYeZjeu/H5uN4aCj32fjZq1L3LZL2SEmG1jQHnX9YXwpJrJ4rTtaEx2qTUwdhJvVrOt4Us3+Nt0Zm2ST58E32JSstLSPfVBHeWmVpTmfL5g2NmSfs0f/LZaryAOwEAz1GxTT32gz8n191uuQWyg1bviivib3T7VeEewwUA9LxxHc6O3w3y+sabqOLAcvnALrK8FcKi5sOx59AN08JnX0+J31p7spEsni8vDb9b949Jr93KFtB1YfEvsSP0zkK4t3Qti9dShThYu7fT56hU+JM4pG7BroFoYd2w8mYAwCvkCrz1UenpwnNhqjU4FsY6gzrqnGsCAFkWV0o1jDQWM8ZosHgxxvyAeg7AA7L+AIBnS7M7xhTGYsYYDRYvxvASn3Pu3xDajTY45w4BeBzAEwD+wzn3cQSv2u+M6NveA+CDgOTIpqZ/NcvDVPcDdTEpU11Xc2fG6Wtjc/gfKZYglkgvjNYDJK+9Qd9bHSSRT7X/TTK0tDFMTbd9IdoGtd5WF0lKL+G3AAAH/560tq+FxfMPx8Pe8emQ7PaXqyjJd7boatzFUpR0lvg0yfL0yihl7lkZZMZ+qvGn7+ExdSHt/T/qEvu0LP+RvlclgtasglqlpVQx8y5q8Qp+IyXhrRGda/+3fiVu+NXkDQn7PxFe3/Dpm5MxlW5vXxitdksWBpmm+Ux0Z52bEc7v/oro7FqLkKy6jjrvXnhbtEDaFfRQ/5ZC8X4lCnB1RsZrQ/FQqitHRl1GlfZepSKWA1LIj+W89S98KKw8VvwZL/5FbKNw00eDJHoH2RHZwQpy95WKUt5j3sRVuBd/hRc3yTF9iV5cJ0t2TSZJtKxpyUX62/H8rfl6iLsFzfFc6HXLSd5aB/PWK19Jxg7cXwAQ3YFAbKnB51YTg9kpqMnncxBdv5qU/8x9JM310XXdKdIeJyTLRx5/eX4cU3Wxg7b7ge40SdV6Plri+Xhpebgvzq2L9+YdC8I98MwCrkf5NYyHYR9Q3vuPDfHS7UOMG5c5FjPGaLB4MYbCKkkYhmEYuaS8tfjqAKwAaleF6erNM2P9MJXpuLOl1tDj6W2D1IBacCpOgyvV9MJJqCzxiUNrxvfi0Ifa1oflsvXJ2O7F7wMAPIPYFTWRSVim0RxKcumpi46lllm3BS2h+VT8+25PXa1sH+UjrYnF7SJUAqhI9SRJbw+Qu5BLhemvSrnMiQTAzsl8m7NwDA34Nh7Cgf5CMtb7E9G5OCF2B4oJnVPQXRNlkG+skE7I196RjGnbjqaZxUmILPXq77v/xNVF26XEJ+7QrLXS2ImnbTTYdKUOsvU0prHM9SrFuNpdHY9pw21hyfUAVQLa9XMqbKkyzi/iUBJudN1sbAtS5vFF0QE4J1VgsvQSXyk59W49Xtzw0ZD2CwA/ZEuZyk0kcyU6GJ2rmkJYFuLQtIri61Cv0QJleWs8cTsYdaEepYx9lem5pqRKe83kNtZkYZb91DV49bXxAt721/FPFSqJd71JB9ApknInHYCu88+r4d1CgazO6uVx6Fxf2P8tdbFfiEroh1P9TMYn8dkMyjAMw8gl9oAyDMMwcklZJb5Zc3pxw/0/RrvoDdchdjG9UtwstWSx0SREbk2gUsa+uii1LLolJOXOr6S5KnXjSDIoOJNCEzNj+Ta04SAA4J7FMcNV5Y0dn4sSwPHPBfmjhtpZ6DSf63fpVHtbXbQrZkmBL5wISYX936W6a1tl+V7aZ+0M0k5l7+cF+XPOjW8nY6fbggSYKtEviaFzr4quoe6ct07oO/0ebPvpjWn5UtcX01hBlizD6k/Dsq9IgXvfGx2Pe1tkPcMhhyO0rjIctytR6Y7yPrm2WqLosNSq+3+MxvSz19FY34uywnKU7Cu5/VTuO/3+KPvOqgsxeMW8+CW9D4uDbBW17dWk7Yy7QMoVmlkrMKcMIJxb/T2XUHLsjsfjNoqeNk4oVQU45mRjaWNIQmX5TZNo+6kVsna7PkB9WbZIBYCtVGhU72N8v9BWJ0uxORnTP33MpZuXJqHznz5YAlS377pCdPse3yZxxI49jTtO3tfY4rZHWtOR3KqV04M+nJbzAkk9vxJgMyjDMAwjl9gDyjAMw8glZZX45uEIHsOXkykoO9R0arwF0RWiMtgPT0VXXd/TIlWQ7DDtwWCn++TSbyRjDy19Mllfslvq8nGHUaUprnoxQp2n06JTenbY9YlmM5BYsiLpunFB72GHlXav3LL3lvgmTaz9ehzCSa1xRQXaHpJlbZQUqpqDzLCI5FIt0d84O1rFVKrkml73Fe19zjgM4M+QjtL6QUsgOuPYjaQGp5TD7zVZUvark34BbM5TSa6Txk6rLfRfaVDiciHJYSvoZZWRqb5k4uzj/b8xY7seaWzBx55Rr1Ivob4jMQlUHVYzamLl1JralkY5UwAACpFJREFU4BY7R260C71iEWQpU8/bLhrbj8nDNIRzrLIrX6JSa5MuZUDNipzUKr/H1YteT4ZY2lP0uubrWxOnNyAmiG/5L7nWP09v7tD+PPE+sOvPgwS47guxg/E9UlCznTqE6zXMXbSznL3H95E8vE6W/0T74L9ddEzAvWHRPr34JYrF/vPhzwdd8+haqpQT21O6x4rNoAzDMIxcYg8owzAMI5eUVeIbQCV6UJ9MR8+QvalbLE87yD6iSbJ966nbs8p0lHB4YVOQKl5eGRMwWZLb3hacML/UFu1gJ2UazK01tBPtpv7Y6bd3q+gqLB+pTMO10xqCs06lFACokCnv+YGoM6j8kkJbsd3BgyLtkRQ0rTFImQsai7vEsnSnXVbZITRD7IpZsmRuebcLePWLSPehkJO0kIayHHiqZKWSlX+9eLvq4qHElJVqFSBSYOXjcUjlIU5+ZvmNE3QHwxKf7gMfhzqr2HE2MOg1IDqxqF3EBcnu7RugLF99bzSeRgkyS+Jjt1ffdkwaTiPIWXo+WOLtlCWrV1pik9y8+tvsr4/2TJXSWUqrx0kMRuvt8XZ4v3z4F8lBeUSkPY4Rccvxdav3Ra5HqX9GOE/X8lFql6XjtS3R5ffuHXKzol3AtgfDku9tCl83ervhWFQ3awNZgSvlccJu2nFiMyjDMAwjl9gDyjAMw8glzns//FYlovIDi33txpfQWBWSztgZo+uckKaJaDNJl9DpbxXNyWemdIsAT3+PixZzgIprbZOkuVdPxITZJFH2aUQ2asbvZhp8nyyviUMq83A3C5VxuGVCliyU5c7SsYY4r64S+bCqurivdZaMeOEczef1dX7v/Oot3vtom8wZ7op2j1s60vXo9JyS+zKRWlk2U+mEJbzKUY4NJ4CfHbQcvN4zaAkA78iSE3VVne2iMZVdspKUs75vpJcxx2KWIzLrvDFrXb5jZla7x5KOKOdxp5DkRJPGVyOJvHxEvzZoCURXYCFej5oIXV8VfyS9P/E9Sf/cwK7lymTs/EW3myHJwFUpDbKYPnL0HZML4jBdJEkX4Z0U1B2DlkB0b75NYxpjHCd6r+L7ncYWS4HPjy9ebAZlGIZh5BJ7QBmGYRi5pKwuvvOvTUfvjHnovSnoV3vvogTHu8Li2uujYyirZp+61bgFwKyMYmFnUjYUHYvTYC2HXzE7TrH3/OG1AICDK6ig2jaR/XaSxU6dQbGifnQXvtVJg5oYyr0VNCOQHGU18h0F2kwlvvr4E/VXhO36ebavUhHXiEtkDXZf/UyWXNst57QAeAKoKUS3VHNdsRTcICeBuxRHKThKmjMyZBJ1NbJEoomOx8imqXLJW/1Rm+3dLT8Sy0gcE+ok66QxXWc5DycyNlSNhWNH5WbWULIuYZWwqA6dBtdpKhyoLVjqizdL6q8B6STmtRlflyfqEIyeKlWmTo8cL5++LJlWXWjc/kTbu5yNH9hbEX7/3hqyX+q5ZHdellya5ao8N+g1fi//RhqW7PrM+j4+Jr2EOmlM45ZjUcOuj4uZyvpJ1vMkxtgVqPtYQMmwGZRhGIaRS+wBZRiGYeSS8nbUbQbwCUQHCLcpqA/z2h6ay26XpF2ub6cyzkzqRMkyjqIdKwHgnMxDWfbTRLqe1Nw5cEVLlFV6K2U+PZ9Olbp7OMFR148V4liPrLMTS6f0rDbplJ67nWa1ZShu6hnhw0jaPJCE2iDrc2i7b17k83JAVc1ZNF2/O+miDERpl5MkZ4kmk5WYXJEqvBY4R7GRFQeaHMk11vYfVhcUaRraEuV/6cO5vUdSz45lus7kmyP6w3MGqVqiWKbLapVcl7GdjFF+e3LNFWiskDHWWjzGEmuxXzZnzAZwnwfOSgIpn2YNI5bD9fqiROdknesRJv9g2VylWT73eqK55UTxnxvijnHHX13nG4FK8uQYVudhG23GMmxBluwUzmqjcZMsH6YxDcFKbtUj6/y0kPY9Sf29ofjKxV8eDptBGYZhGLnEHlCGYRhGLhmXxOecWwHg7xCK2n/Te//ERd9wEsAziLNgrmU3J+xK96wrk6HuyitRhDpTshIYsxIdgSi/8XuyVBV1v7AMlpXMqOpLVnIny3A8U1d0Rsz1qrLkvESaYAlAtQnOouPXFZUFCjQm6w2YUEYTM/2HqnHwU204OJwcqrCSokocy1xZDqvajLGs31LXWTLSz2bHGzurVAp+h9xPPbLOsXixHEx2SenxZTm2so6pOmOMP09jn/clI3mzb4BPYnkZ9T3mFwAOuXicfH1nyZzLZDlczcMBcd+epb4c+jrrnucHvcZcTKIvJVnfo/cW/rNE1jWURSL70Vi1SKi1NJgVi+NkzDMo51wFgH8A8GEAiwB8zDm36OLvMi5nLGaM0WDxYoxH4rsBwD7v/Zve+34A3wVwd2l2y5iiWMwYo8Hi5TJnPBLffKRTFA8h3RcBAOCcewTAI/LPc9jqdibupzzA02GdqhdX0R9MA9KC3OQhvde/PMRWl4phY6YoXr7ssvogT0Ymb8ykKWfMjO0e85tTImYsXjC+B5TLGCsqWem9Xw1gNQA45zryXGhypEyl4yj3V2aMpWJmKsYLMHWOpcwxY/eYSc5442U8Et8hFLdqOzzEtoYBWMwYo8Pi5TJnPA+o/wFwjXNuoXOuCsDvAXiuNLtlTFEsZozRYPFymTNmic97P+Cc+yMALyFYQL/lvX99mLetHuv35Qw7jjEwhpiZKucZmDrHUrbjsHvMlGBcx1HWhoWGYRiGMVKskoRhGIaRS+wBZRiGYeSSsjygnHMrnHN7nHP7nHOfKcd3lgrn3ALn3Frn3C7n3OvOuUdlfLZz7r+dc2/IcuLqwYwC51yFc26rc+5H8u+FzrlNchz/Ln+MnnAma8xYvEwMFi/5oNTxcskfUFOgXMkAgD/x3l8H4EYAn5T9/wyANd77awCskX9PBh5FupHAlwB8RY7jJICPT8heEZM8ZixeyozFS64oabyUYwY1qcuVeO+7vPevyfpphJM/H+EYnpLNngJwz8Ts4chxzrUA+AikE5RzzgG4DcD3ZZO8HMekjRmLlwnB4iUHXIp4KccDKqtcyfwhts01zrkCgCUANgFo9N53ASHIkO5Qlle+CuBPAVyQf88B0OO914JPefltpkTMWLyUDYuXfFDyeCnHA2pE5UryjnOuFsAPAPyx9/7URO/PaHHO3QWg23u/hYczNs3Db5PX/RoxFi9lJa/7NWIsXrIpR8v3SV+uxDk3HSF4vuO9/08ZPuqca/LedznnmgB0T9wejohbAKx0zt2J0LmlDuH/eOqdc5Xyfzl5+W0mdcxYvJQdi5eJ55LESzlmUJO6XInoqP8MYJf3/m/ppecAPCDrDwB4ttz7Nhq895/13rd47wsIv8FPvff3AlgLYJVslpfjmLQxY/EyIVi8TDCXLF6895f8PwB3AtgLYD+Az5fjO0u478sQpqXbAWyT/+5E0FfXAHhDlrMnel9HcUzLAfxI1q8CsBnAPgDfAzBjovdvMseMxYvFi8VL6eLFSh0ZhmEYucQqSRiGYRi5xB5QhmEYRi6xB5RhGIaRS+wBZRiGYeQSe0AZhmEYucQeUIZhGEYusQeUYRiGkUv+Hxx9EByIB3sMAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 9 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "incorrect = np.where(predicted_classes!=test_Y)[0]\n",
    "print(\"Found %d incorrect labels\" % len(incorrect))\n",
    "for i, incorrect, in enumerate(incorrect[:9]):\n",
    "    plt.subplot(3, 3, i+1)\n",
    "    plt.pcolormesh(test_X[incorrect].reshape(ydim, xdim), cmap='jet')\n",
    "    plt.title(\"Predicted {}, Class {}\".format(predicted_classes[incorrect], test_Y[incorrect]))\n",
    "    plt.tight_layout()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "             precision    recall  f1-score   support\n",
      "\n",
      "    Class 0       0.47      0.94      0.62       109\n",
      "    Class 1       0.48      0.17      0.25        93\n",
      "    Class 2       0.68      0.27      0.39        84\n",
      "\n",
      "avg / total       0.53      0.49      0.43       286\n",
      "\n"
     ]
    }
   ],
   "source": [
    "from sklearn.metrics import classification_report\n",
    "target_names = [\"Class {}\".format(i) for i in range(num_classes)]\n",
    "print(classification_report(test_Y, predicted_classes, target_names=target_names))"
   ]
  },
  {
   "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": "3.7.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}