Classification_task.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "import warnings\n",
    "warnings.filterwarnings('ignore')\n",
    "\n",
    "from utils import *\n",
    "# data processing\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "\n",
    "# image processing\n",
    "from PIL import Image\n",
    "\n",
    "# data visualization\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sn\n",
    "\n",
    "# tf and keras\n",
    "import tensorflow as tf\n",
    "import keras\n",
    "from keras.models import Sequential, Model, load_model\n",
    "from keras.layers import Input, concatenate, Conv2D, MaxPooling2D, AveragePooling2D, Conv2DTranspose, Dropout, BatchNormalization\n",
    "from keras.layers.core import Dense, Activation, Flatten\n",
    "from keras.optimizers import SGD, Adam, RMSprop\n",
    "from keras.utils import plot_model\n",
    "from keras.callbacks import ModelCheckpoint\n",
    "from keras import backend as K\n",
    "\n",
    "# dataset processing, ml models and metrics\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.metrics import confusion_matrix\n",
    "from sklearn.model_selection import KFold\n",
    "from sklearn.model_selection import StratifiedKFold\n",
    "from sklearn.metrics import confusion_matrix\n",
    "from sklearn.metrics import classification_report\n",
    "import scipy\n",
    "import glob\n",
    "from pprint import pprint\n",
    "import datetime\n",
    "import os\n",
    "import json\n",
    "# import itertools"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "pd.set_option('display.max_rows',None)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "K.clear_session()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "dir_ = 'datasets/HASYv2/'\n",
    "model_dir = 'trained_models/'\n",
    "processed_data_dir ='processed_data/'\n",
    "plot_dir = 'plots/'\n",
    "model_logs = model_dir+'/logs/'\n",
    "K.set_image_data_format('channels_first')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Networks"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# CNN 1\n",
    "# adapted from large layer\n",
    "def get_layers1(input_shape, data_format, classes):\n",
    "    return [\n",
    "        Conv2D(4, (2,2), padding='same', data_format=data_format, input_shape=input_shape),\n",
    "        Activation('relu'),\n",
    "        #MaxPooling2D(pool_size=(2,2), strides=(2,2)),\n",
    "        \n",
    "        Conv2D(16,(2,2), padding='same', data_format=data_format),\n",
    "        Activation('relu'),\n",
    "        MaxPooling2D(pool_size=(2,2), strides=(2,2)),\n",
    "                \n",
    "        Flatten(),\n",
    "        Dense(3200),\n",
    "        Activation('relu'),\n",
    "        \n",
    "        Dense(classes),\n",
    "        Activation('softmax'),\n",
    "    ]\n",
    "\n",
    "def create_network1(input_shape, data_format, classes):\n",
    "    model = Sequential()\n",
    "    layers = get_layers1(input_shape, data_format, classes)\n",
    "    for i in layers:\n",
    "        model.add(i)\n",
    "    return model                "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# CNN 2 without Batch Normalization\n",
    "def get_layers2(input_shape, data_format, classes):\n",
    "    return [\n",
    "        Conv2D(25, (5, 5), padding='same', data_format=data_format, input_shape=input_shape),\n",
    "        Activation('relu'),\n",
    "        MaxPooling2D(pool_size=(2,2), strides=(2,2)),\n",
    "        \n",
    "        Conv2D(50, (3, 3), padding='same', data_format=data_format),\n",
    "        Activation('relu'),\n",
    "        MaxPooling2D(pool_size=(2,2), strides=(2,2)),\n",
    "                \n",
    "        Conv2D(100, (2,2), padding='same', data_format=data_format),\n",
    "        Activation('relu'),\n",
    "        MaxPooling2D(pool_size=(2,2), strides=(2,2)),\n",
    "        \n",
    "        Conv2D(200, (2,2), padding='same', data_format=data_format),\n",
    "        Activation('relu'),\n",
    "        MaxPooling2D(pool_size=(2,2), strides=(2,2)),\n",
    "        \n",
    "        Flatten(),\n",
    "        Dense(3200),\n",
    "        Activation('relu'),\n",
    "        \n",
    "        Dense(1600),\n",
    "        Activation('relu'),\n",
    "        \n",
    "        Dense(classes),\n",
    "        Activation('softmax'),\n",
    "    ]\n",
    "\n",
    "def create_network2(input_shape, data_format, classes):\n",
    "    model = Sequential()\n",
    "    layers = get_layers2(input_shape, data_format, classes)\n",
    "    for i in layers:\n",
    "        model.add(i)\n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "# CNN 2 with Batch Normalization\n",
    "def get_layers3(input_shape, data_format, classes):\n",
    "    return [\n",
    "        Conv2D(25, (5, 5), padding='same', data_format=data_format, input_shape=input_shape),\n",
    "        BatchNormalization(),\n",
    "        Activation('relu'),\n",
    "        MaxPooling2D(pool_size=(2,2), strides=(2,2)),\n",
    "        \n",
    "        Conv2D(50, (3, 3), padding='same', data_format=data_format),\n",
    "        BatchNormalization(),\n",
    "        Activation('relu'),\n",
    "        MaxPooling2D(pool_size=(2,2), strides=(2,2)),\n",
    "                \n",
    "        Conv2D(100, (2,2), padding='same', data_format=data_format),\n",
    "        BatchNormalization(),\n",
    "        Activation('relu'),\n",
    "        MaxPooling2D(pool_size=(2,2), strides=(2,2)),\n",
    "        \n",
    "        Conv2D(200, (2,2), padding='same', data_format=data_format),\n",
    "        BatchNormalization(),\n",
    "        Activation('relu'),\n",
    "        MaxPooling2D(pool_size=(2,2), strides=(2,2)),\n",
    "        \n",
    "        Flatten(),\n",
    "        Dense(3200),\n",
    "        Activation('relu'),\n",
    "        \n",
    "        Dense(1600),\n",
    "        Activation('relu'),\n",
    "        \n",
    "        Dense(classes),\n",
    "        Activation('softmax'),\n",
    "    ]\n",
    "\n",
    "def create_network3(input_shape, data_format, classes):\n",
    "    model = Sequential()\n",
    "    layers = get_layers3(input_shape, data_format, classes)\n",
    "    for i in layers:\n",
    "        model.add(i)\n",
    "    return model                "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_layers_ann1(input_shape, classes):\n",
    "    return [\n",
    "        Dense(1024,\n",
    "              input_shape=input_shape,\n",
    "              kernel_initializer=keras.initializers.lecun_normal(seed=25)\n",
    "             ),\n",
    "        Activation('relu'),\n",
    "        Dropout(0.2),\n",
    "        Dense(1024,\n",
    "              kernel_initializer=keras.initializers.lecun_normal(seed=26)\n",
    "             ),\n",
    "        Activation('relu'),\n",
    "        Dropout(0.2),\n",
    "        Dense(2048,\n",
    "              kernel_initializer=keras.initializers.lecun_normal(seed=27)\n",
    "             ),\n",
    "        Activation('relu'),\n",
    "        Dropout(0.2),\n",
    "        Dense(classes,\n",
    "              kernel_initializer=keras.initializers.lecun_normal(seed=29)\n",
    "             ),\n",
    "        Activation('softmax')\n",
    "    ]\n",
    "\n",
    "def create_network_ann1(input_shape, classes):\n",
    "    model = Sequential()\n",
    "    layers = get_layers_ann1(input_shape, classes)\n",
    "    for i in layers:\n",
    "        model.add(i)\n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_layers_ann2(input_shape, classes):\n",
    "    return [\n",
    "        Dense(1024, \n",
    "              input_shape=input_shape, \n",
    "              kernel_initializer=keras.initializers.lecun_normal(seed=25), \n",
    "              kernel_regularizer=keras.regularizers.l2(0.01)\n",
    "             ),\n",
    "        Activation('relu'),\n",
    "        Dropout(0.1),\n",
    "        Dense(1024, \n",
    "              kernel_initializer=keras.initializers.lecun_normal(seed=26),\n",
    "              kernel_regularizer=keras.regularizers.l2(0.01)\n",
    "             ),\n",
    "        Activation('relu'),\n",
    "        Dropout(0.1),\n",
    "        Dense(2048,\n",
    "              kernel_initializer=keras.initializers.lecun_normal(seed=27),\n",
    "              kernel_regularizer=keras.regularizers.l2(0.01)\n",
    "             ),\n",
    "        Activation('relu'),\n",
    "        Dropout(0.1),\n",
    "        Dense(classes,\n",
    "              kernel_initializer=keras.initializers.lecun_normal(seed=29)\n",
    "             ),\n",
    "        Activation('softmax'),\n",
    "    ]\n",
    "\n",
    "def create_network_ann2(input_shape, classes):\n",
    "    model = Sequential()\n",
    "    layers = get_layers_ann2(input_shape, classes)\n",
    "    for i in layers:\n",
    "        model.add(i)\n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "# LeNet\n",
    "def get_layers6(input_shape, classes):\n",
    "    return [\n",
    "        # 1st Layer: Conv2D\n",
    "        Conv2D(\n",
    "            filters=6, \n",
    "            kernel_size=(5, 5), \n",
    "            strides=1,\n",
    "            padding='same', \n",
    "            data_format=data_format, \n",
    "            input_shape=input_shape\n",
    "        ),\n",
    "        Activation('tanh'),\n",
    "        \n",
    "        # 2nd Layer: AveragePooling2D\n",
    "        AveragePooling2D(\n",
    "            pool_size=(2,2), \n",
    "            strides=(1,1),\n",
    "            padding='valid'\n",
    "        ),\n",
    "        \n",
    "        # 3rd Layer: Conv2D\n",
    "        Conv2D(\n",
    "            filters=16, \n",
    "            kernel_size=(5, 5),\n",
    "            strides=1,\n",
    "            padding='valid', \n",
    "            data_format=data_format\n",
    "        ),\n",
    "        Activation('tanh'),\n",
    "        \n",
    "        # 4th Layer: AveragePooling2D\n",
    "        AveragePooling2D(\n",
    "            pool_size=(2,2), \n",
    "            strides=(2,2),\n",
    "            padding='valid'\n",
    "        ),\n",
    "        \n",
    "        # 5th Layer: Conv2D\n",
    "        Conv2D(\n",
    "            filters=120, \n",
    "            kernel_size=(5, 5),\n",
    "            strides=1,\n",
    "            padding='valid', \n",
    "            data_format=data_format\n",
    "        ),\n",
    "        Activation('tanh'),\n",
    "        \n",
    "        # Flatten output\n",
    "        Flatten(),\n",
    "        \n",
    "        # 6th Layer: FC 1\n",
    "        Dense(1024),\n",
    "        Activation('tanh'),\n",
    "        Dropout(0.2),\n",
    "        \n",
    "        # 7th Layer: FC 2 (Added this extra layer to account for more classes)\n",
    "        Dense(512),\n",
    "        Activation('tanh'),\n",
    "        Dropout(0.2),\n",
    "        \n",
    "        # 8th layer: Output layer\n",
    "        Dense(classes),\n",
    "        Activation('softmax'),\n",
    "    ]\n",
    "\n",
    "def create_network6(input_shape, classes):\n",
    "    model = Sequential()\n",
    "    layers = get_layers6(input_shape, classes)\n",
    "    for i in layers:\n",
    "        model.add(i)\n",
    "    return model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Evaluation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "def t_test(test_x, test_y, models, sym2latex_dict):\n",
    "    if len(models) < 0:\n",
    "        raise ValueError(\"No models found in models variable\")\n",
    "    if len(test_x) != len(test_y):\n",
    "        raise ValueError(\"Varibales test_x and test_y are of different length\")\n",
    "    \n",
    "    # convert test values from one hot encoded to label value\n",
    "    y_true = []\n",
    "    for i in range(len(test_y)):\n",
    "        # print(one_hot_encode_to_char(res[i], threshold = 0.5, get_max=False))\n",
    "        val = one_hot_encode_to_char(test_y[i], threshold = 0.9, get_max=True)\n",
    "        if len(val) > 0:\n",
    "            y_true.append(val[0])\n",
    "        else:\n",
    "            y_true.append(None)\n",
    "            print(\":(\")\n",
    "    \n",
    "    # test_x_ann for ann with shape (samples, 32*32)\n",
    "    # test_x for cnn with shape (samples, 1, 32, 32)\n",
    "    test_x_ann = test_x.reshape(test_x.shape[0], test_x.shape[-1] * test_x.shape[-2])\n",
    "    no_models = len(models)\n",
    "    cols = ['symbol','latex']\n",
    "    preds = []\n",
    "    # convert predictions from one hot encoded to label value\n",
    "    for i in range(no_models):\n",
    "        print(\"Predictng labels for Model \"+str(i+1))\n",
    "        cols.append('model'+str(i+1))\n",
    "        \n",
    "        # check if model is cnn or ann with a cnn flag\n",
    "        if models[i][1] == 1:\n",
    "            res = models[i][0].predict(test_x)\n",
    "        else:\n",
    "            res = models[i][0].predict(test_x_ann)\n",
    "        \n",
    "        y_pred = []\n",
    "        for j in range(len(res)):\n",
    "            # print(one_hot_encode_to_char(res[i], threshold = 0.5, get_max=False))\n",
    "            val = one_hot_encode_to_char(res[j], threshold = 0.1, get_max=True)\n",
    "            if len(val) > 0:\n",
    "                y_pred.append(val[0])\n",
    "            else:\n",
    "                y_pred.append(None)\n",
    "                print(\":(\")\n",
    "        preds.append(y_pred)\n",
    "    print(\"Predictions Done\")\n",
    "    print(\"Comparing Results...\")\n",
    "    \n",
    "    # perform t test\n",
    "    t_test_result = pd.DataFrame(columns=cols)\n",
    "    for i in range(len(y_true)):\n",
    "        t_test_result.loc[i,'symbol'] = y_true[i]\n",
    "        t_test_result.loc[i,'latex'] = sym2latex_dict[y_true[i]]\n",
    "        for j in range(no_models):\n",
    "            t_test_result.loc[i, 'model'+str(j+1)] = sym2latex_dict[preds[j][i]]\n",
    "    print(\"Done\")    \n",
    "    return t_test_result\n",
    "\n",
    "def create_t_test_report(test_res):\n",
    "    models_names = list(test_res.columns.values[2:])\n",
    "    res = pd.DataFrame(columns=['latex','total_count']+models_names)\n",
    "    for i in range(len(symbols)):\n",
    "        res.loc[i,'latex'] = symbols.loc[i,'latex']\n",
    "        res.loc[i,'total_count'] = 0\n",
    "        for j in models_names:\n",
    "            res.loc[i,j] = 0\n",
    "    for i in range(len(test_res)):\n",
    "        res.loc[test_res.loc[i,'symbol'],'total_count'] += 1\n",
    "        for j in models_names:\n",
    "            if test_res.loc[i,'latex'] == test_res.loc[i,j]:\n",
    "                res.loc[latex2sym[test_res.loc[i,j]],j] += 1\n",
    "    for i in range(len(res)):\n",
    "        for j in models_names:\n",
    "            res.loc[i,j+'_acc'] = round((res.loc[i,j]/res.loc[i,'total_count']) * 100, 2)\n",
    "    return res"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Stratified Train Test Split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "symbols = read_csv(processed_data_dir+'symbols.csv')\n",
    "symbols_dict = {}\n",
    "for i in range(len(symbols)):\n",
    "    symbols_dict[symbols['old_symbol'][i]] = symbols['new_id'][i]\n",
    "# symbols_list = np.array(whole_dataset['symbol_id']).reshape(-1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Use this cell to load csvs\n",
    "train = read_csv(processed_data_dir+'train.csv').drop(['Unnamed: 0'], axis=1).reset_index(drop=True)\n",
    "test = read_csv(processed_data_dir+'test.csv').drop(['Unnamed: 0'], axis=1).reset_index(drop=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 137,
   "metadata": {},
   "outputs": [],
   "source": [
    "# (train_count, test_count, percentage of train count to total)\n",
    "labels_count = get_label_count_train_test_dfs(train, test)\n",
    "# pprint(labels_count)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "One Hot Encode labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "No of categories is  107\n"
     ]
    }
   ],
   "source": [
    "num_classes = len(symbols)\n",
    "print('No of categories is ', num_classes)\n",
    "train_one_hot_symbols = keras.utils.to_categorical(train['symbol_id'], num_classes)\n",
    "test_one_hot_symbols = keras.utils.to_categorical(test['symbol_id'], num_classes)\n",
    "\n",
    "# whole_dataset['symbol_id_ohe'] = [list(one_hot_symbols[i]) for i in range(len(whole_dataset))]\n",
    "train['symbol_id_ohe'] = [list(train_one_hot_symbols[i]) for i in range(len(train))]\n",
    "test['symbol_id_ohe'] = [list(test_one_hot_symbols[i]) for i in range(len(test))]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load all images into memory\n",
    "train = populate_images(train)\n",
    "test = populate_images(test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>path</th>\n",
       "      <th>symbol_id</th>\n",
       "      <th>latex</th>\n",
       "      <th>symbol_id_ohe</th>\n",
       "      <th>img</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>hasy-data/v2-71022.png</td>\n",
       "      <td>89</td>\n",
       "      <td>\\#</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...</td>\n",
       "      <td>[[255, 255, 255, 255, 255, 255, 255, 255, 255,...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>hasy-data/v2-70169.png</td>\n",
       "      <td>88</td>\n",
       "      <td>\\&amp;</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...</td>\n",
       "      <td>[[255, 255, 255, 255, 255, 255, 255, 255, 255,...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>hasy-data/v2-66865.png</td>\n",
       "      <td>85</td>\n",
       "      <td>\\$</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...</td>\n",
       "      <td>[[255, 255, 255, 255, 255, 255, 255, 255, 255,...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>hasy-data/v2-13825.png</td>\n",
       "      <td>102</td>\n",
       "      <td>\\infty</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...</td>\n",
       "      <td>[[255, 255, 255, 255, 255, 255, 255, 255, 255,...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>hasy-data/v2-138917.png</td>\n",
       "      <td>102</td>\n",
       "      <td>\\infty</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...</td>\n",
       "      <td>[[255, 255, 255, 255, 255, 255, 255, 255, 255,...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                      path  symbol_id   latex  \\\n",
       "0   hasy-data/v2-71022.png         89      \\#   \n",
       "1   hasy-data/v2-70169.png         88      \\&   \n",
       "2   hasy-data/v2-66865.png         85      \\$   \n",
       "3   hasy-data/v2-13825.png        102  \\infty   \n",
       "4  hasy-data/v2-138917.png        102  \\infty   \n",
       "\n",
       "                                       symbol_id_ohe  \\\n",
       "0  [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...   \n",
       "1  [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...   \n",
       "2  [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...   \n",
       "3  [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...   \n",
       "4  [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...   \n",
       "\n",
       "                                                 img  \n",
       "0  [[255, 255, 255, 255, 255, 255, 255, 255, 255,...  \n",
       "1  [[255, 255, 255, 255, 255, 255, 255, 255, 255,...  \n",
       "2  [[255, 255, 255, 255, 255, 255, 255, 255, 255,...  \n",
       "3  [[255, 255, 255, 255, 255, 255, 255, 255, 255,...  \n",
       "4  [[255, 255, 255, 255, 255, 255, 255, 255, 255,...  "
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>path</th>\n",
       "      <th>symbol_id</th>\n",
       "      <th>latex</th>\n",
       "      <th>symbol_id_ohe</th>\n",
       "      <th>img</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>hasy-data/v2-61274.png</td>\n",
       "      <td>80</td>\n",
       "      <td>\\subset</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...</td>\n",
       "      <td>[[255, 255, 255, 255, 255, 255, 255, 255, 255,...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>hasy-data/v2-61852.png</td>\n",
       "      <td>81</td>\n",
       "      <td>\\supset</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...</td>\n",
       "      <td>[[255, 255, 255, 255, 255, 255, 255, 255, 255,...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>hasy-data/v2-51999.png</td>\n",
       "      <td>74</td>\n",
       "      <td>\\omega</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...</td>\n",
       "      <td>[[255, 255, 255, 255, 255, 255, 255, 255, 255,...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>hasy-data/v2-33163.png</td>\n",
       "      <td>63</td>\n",
       "      <td>\\gamma</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...</td>\n",
       "      <td>[[255, 255, 255, 255, 255, 255, 255, 255, 255,...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>hasy-data/v2-134232.png</td>\n",
       "      <td>101</td>\n",
       "      <td>\\|</td>\n",
       "      <td>[0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...</td>\n",
       "      <td>[[255, 255, 255, 255, 255, 255, 255, 255, 255,...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                      path  symbol_id    latex  \\\n",
       "0   hasy-data/v2-61274.png         80  \\subset   \n",
       "1   hasy-data/v2-61852.png         81  \\supset   \n",
       "2   hasy-data/v2-51999.png         74   \\omega   \n",
       "3   hasy-data/v2-33163.png         63   \\gamma   \n",
       "4  hasy-data/v2-134232.png        101       \\|   \n",
       "\n",
       "                                       symbol_id_ohe  \\\n",
       "0  [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...   \n",
       "1  [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...   \n",
       "2  [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...   \n",
       "3  [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...   \n",
       "4  [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...   \n",
       "\n",
       "                                                 img  \n",
       "0  [[255, 255, 255, 255, 255, 255, 255, 255, 255,...  \n",
       "1  [[255, 255, 255, 255, 255, 255, 255, 255, 255,...  \n",
       "2  [[255, 255, 255, 255, 255, 255, 255, 255, 255,...  \n",
       "3  [[255, 255, 255, 255, 255, 255, 255, 255, 255,...  \n",
       "4  [[255, 255, 255, 255, 255, 255, 255, 255, 255,...  "
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# CNN"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Split set for CNN Models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 255,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CNN Input Shape is (1, 32, 32)\n"
     ]
    }
   ],
   "source": [
    "input_shape = (1, 32, 32)\n",
    "print(\"CNN Input Shape is\", input_shape)\n",
    "\n",
    "data_format = 'channels_first'\n",
    "\n",
    "X_train, X_test, y_train, y_test = process_x_y_train_test_stratified_2df(train, test, input_shape=input_shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model 1: CNN 1 (based on LeNet-1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 256,
   "metadata": {},
   "outputs": [],
   "source": [
    "# optimizer\n",
    "lr = 0.01\n",
    "mtn=0.9\n",
    "optimizer = SGD(lr=lr, momentum=mtn)\n",
    "# optimizer = RMSprop(lr=lr)\n",
    "\n",
    "# loss function\n",
    "loss = 'categorical_crossentropy'\n",
    "\n",
    "batch_size = 512\n",
    "epochs = 30"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 257,
   "metadata": {},
   "outputs": [],
   "source": [
    "# create network\n",
    "model1 = create_network1(input_shape, data_format, num_classes)\n",
    "# compile network\n",
    "model1.compile(loss=loss, optimizer=optimizer, metrics = ['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 258,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model 1 summary\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d_1 (Conv2D)            (None, 4, 32, 32)         20        \n",
      "_________________________________________________________________\n",
      "activation_5 (Activation)    (None, 4, 32, 32)         0         \n",
      "_________________________________________________________________\n",
      "conv2d_2 (Conv2D)            (None, 16, 32, 32)        272       \n",
      "_________________________________________________________________\n",
      "activation_6 (Activation)    (None, 16, 32, 32)        0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_1 (MaxPooling2 (None, 16, 16, 16)        0         \n",
      "_________________________________________________________________\n",
      "flatten_1 (Flatten)          (None, 4096)              0         \n",
      "_________________________________________________________________\n",
      "dense_5 (Dense)              (None, 3200)              13110400  \n",
      "_________________________________________________________________\n",
      "activation_7 (Activation)    (None, 3200)              0         \n",
      "_________________________________________________________________\n",
      "dense_6 (Dense)              (None, 107)               342507    \n",
      "_________________________________________________________________\n",
      "activation_8 (Activation)    (None, 107)               0         \n",
      "=================================================================\n",
      "Total params: 13,453,199\n",
      "Trainable params: 13,453,199\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "# Model 1 Summary\n",
    "print('Model 1 summary')\n",
    "print(model1.summary())\n",
    "# plot_model(model1, to_file=model_dir+'model1.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 259,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 38484 samples, validate on 4276 samples\n",
      "Epoch 1/30\n",
      "38484/38484 [==============================] - 4s 104us/step - loss: 3.3721 - acc: 0.2717 - val_loss: 1.7830 - val_acc: 0.6239\n",
      "Epoch 2/30\n",
      "38484/38484 [==============================] - 4s 95us/step - loss: 1.2611 - acc: 0.7231 - val_loss: 0.9426 - val_acc: 0.7823\n",
      "Epoch 3/30\n",
      "38484/38484 [==============================] - 4s 95us/step - loss: 0.7826 - acc: 0.8167 - val_loss: 0.7371 - val_acc: 0.8258\n",
      "Epoch 4/30\n",
      "38484/38484 [==============================] - 4s 95us/step - loss: 0.5962 - acc: 0.8579 - val_loss: 0.6177 - val_acc: 0.8562\n",
      "Epoch 5/30\n",
      "38484/38484 [==============================] - 4s 95us/step - loss: 0.4747 - acc: 0.8881 - val_loss: 0.5590 - val_acc: 0.8660\n",
      "Epoch 6/30\n",
      "38484/38484 [==============================] - 4s 95us/step - loss: 0.3804 - acc: 0.9078 - val_loss: 0.5214 - val_acc: 0.8768\n",
      "Epoch 7/30\n",
      "38484/38484 [==============================] - 4s 95us/step - loss: 0.3029 - acc: 0.9283 - val_loss: 0.4861 - val_acc: 0.8866\n",
      "Epoch 8/30\n",
      "38484/38484 [==============================] - 4s 95us/step - loss: 0.2414 - acc: 0.9432 - val_loss: 0.4612 - val_acc: 0.8952\n",
      "Epoch 9/30\n",
      "38484/38484 [==============================] - 4s 95us/step - loss: 0.1841 - acc: 0.9562 - val_loss: 0.4597 - val_acc: 0.8950- loss: 0.1865 -\n",
      "Epoch 10/30\n",
      "38484/38484 [==============================] - 4s 95us/step - loss: 0.1390 - acc: 0.9675 - val_loss: 0.4536 - val_acc: 0.8990\n",
      "Epoch 11/30\n",
      "38484/38484 [==============================] - 4s 95us/step - loss: 0.1016 - acc: 0.9777 - val_loss: 0.4509 - val_acc: 0.9011 - loss: 0.1016 - acc: 0.9\n",
      "Epoch 12/30\n",
      "38484/38484 [==============================] - 4s 95us/step - loss: 0.0768 - acc: 0.9841 - val_loss: 0.4577 - val_acc: 0.9039\n",
      "Epoch 13/30\n",
      "38484/38484 [==============================] - 4s 95us/step - loss: 0.0561 - acc: 0.9896 - val_loss: 0.4796 - val_acc: 0.9018\n",
      "Epoch 14/30\n",
      "38484/38484 [==============================] - 4s 95us/step - loss: 0.0441 - acc: 0.9918 - val_loss: 0.4666 - val_acc: 0.9053\n",
      "Epoch 15/30\n",
      "38484/38484 [==============================] - 4s 95us/step - loss: 0.0319 - acc: 0.9949 - val_loss: 0.4832 - val_acc: 0.9039\n",
      "Epoch 16/30\n",
      "38484/38484 [==============================] - 4s 95us/step - loss: 0.0279 - acc: 0.9953 - val_loss: 0.4811 - val_acc: 0.9043\n",
      "Epoch 17/30\n",
      "38484/38484 [==============================] - 4s 95us/step - loss: 0.0308 - acc: 0.9949 - val_loss: 0.4820 - val_acc: 0.9104\n",
      "Epoch 18/30\n",
      "38484/38484 [==============================] - 4s 95us/step - loss: 0.0205 - acc: 0.9969 - val_loss: 0.4885 - val_acc: 0.9109\n",
      "Epoch 19/30\n",
      "38484/38484 [==============================] - 4s 95us/step - loss: 0.0164 - acc: 0.9980 - val_loss: 0.4906 - val_acc: 0.9072\n",
      "Epoch 20/30\n",
      "38484/38484 [==============================] - 4s 95us/step - loss: 0.0160 - acc: 0.9980 - val_loss: 0.5020 - val_acc: 0.9097\n",
      "Epoch 21/30\n",
      "38484/38484 [==============================] - 4s 95us/step - loss: 0.0145 - acc: 0.9978 - val_loss: 0.4993 - val_acc: 0.9104\n",
      "Epoch 22/30\n",
      "38484/38484 [==============================] - 4s 95us/step - loss: 0.0131 - acc: 0.9981 - val_loss: 0.5081 - val_acc: 0.9074\n",
      "Epoch 23/30\n",
      "38484/38484 [==============================] - 4s 95us/step - loss: 0.0112 - acc: 0.9988 - val_loss: 0.5105 - val_acc: 0.9104\n",
      "Epoch 24/30\n",
      "38484/38484 [==============================] - 4s 95us/step - loss: 0.0129 - acc: 0.9978 - val_loss: 0.5150 - val_acc: 0.9093\n",
      "Epoch 25/30\n",
      "38484/38484 [==============================] - 4s 95us/step - loss: 0.0107 - acc: 0.9987 - val_loss: 0.5174 - val_acc: 0.9116\n",
      "Epoch 26/30\n",
      "38484/38484 [==============================] - 4s 95us/step - loss: 0.0102 - acc: 0.9986 - val_loss: 0.5188 - val_acc: 0.9093\n",
      "Epoch 27/30\n",
      "38484/38484 [==============================] - 4s 96us/step - loss: 0.0100 - acc: 0.9988 - val_loss: 0.5211 - val_acc: 0.9116\n",
      "Epoch 28/30\n",
      "38484/38484 [==============================] - 4s 96us/step - loss: 0.0107 - acc: 0.9985 - val_loss: 0.5213 - val_acc: 0.9093\n",
      "Epoch 29/30\n",
      "38484/38484 [==============================] - 4s 97us/step - loss: 0.0100 - acc: 0.9987 - val_loss: 0.5230 - val_acc: 0.9104\n",
      "Epoch 30/30\n",
      "38484/38484 [==============================] - 4s 97us/step - loss: 0.0096 - acc: 0.9988 - val_loss: 0.5274 - val_acc: 0.9118\n",
      "Training completed on 2020-04-23, 20-15\n"
     ]
    }
   ],
   "source": [
    "# Fit dataset\n",
    "tensorboard_callback1 = keras.callbacks.TensorBoard(os.path.join(model_dir+'logs','model1'))\n",
    "history1 = model1.fit(X_train, \n",
    "                      y_train, \n",
    "                      validation_data=(X_test, y_test), \n",
    "                      batch_size=batch_size, \n",
    "                      epochs=epochs, \n",
    "                      verbose=1, \n",
    "                      callbacks=[tensorboard_callback1])\n",
    "model1.save(model_dir+'model1.h5')\n",
    "print(f\"Training completed on {datetime.datetime.now().strftime('%Y-%m-%d, %H-%M')}\")\n",
    "df_history1 = pd.DataFrame(data=history1.history)\n",
    "df_history1.to_csv(model_logs+'logs_model1.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Metrics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 260,
   "metadata": {},
   "outputs": [],
   "source": [
    "model1 = load_model(model_dir+'model1.h5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 261,
   "metadata": {},
   "outputs": [],
   "source": [
    "res = model1.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 262,
   "metadata": {},
   "outputs": [],
   "source": [
    "# convert predictions from one hot encoded to label value\n",
    "y_pred = convert_pred_list_ohe_to_labels(res, threshold=0.1, get_max=True)\n",
    "# convert test values from one hot encoded to label value\n",
    "y_true = convert_pred_list_ohe_to_labels(y_test, threshold=0.9, get_max=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Confusion Matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 263,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAABCoAAARrCAYAAABFUUcDAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8QZhcZAAAgAElEQVR4nOzdfZRldX3n+/cHqpNqBUMlQCTQszATJTpEm4TmYrgqIvG2JPEpMZGZEKIklWSsBLwY40MiOiZZxgccnc7DPW0hOCFERiAxKIm9HAhxRpsusIFuGx+SGG0hdrw2Aa6A0/C9f5yDU9NT3dXWPrtO1T7v11p79Tm/vc/+fTm76tTZP36/7zdVhSRJkiRJ0kpw2KgDkCRJkiRJeowDFZIkSZIkacVwoEKSJEmSJK0YDlRIkiRJkqQVw4EKSZIkSZK0YjhQIUmSJEmSVozWBiqSbEzy2SRfSPK6tvqRJEmSJEnLL8llSfYk2XGA/Uny3sG4wB1JfvhQztvKQEWSw4E/AF4APA04N8nT2uhLkiRJkiSNxOXAxoPsfwHw5ME2DfzRoZy0rRkVpwFfqKq/r6pvAn8GvKilviRJkiRJ0jKrqpuBrx/kkBcBH6i+TwFHJTlusfO2NVBxPPDlec93D9okSZIkSdJ4WNLYwERLwWSBtvpfDkim6U/9YOKEM39k4uh/c8CT7d22aajBSZIkSZKWz+TEgveInbH2lJla/KjV56Htf/DLDO7bB3pV1fs2TrHo2MBC2hqo2A2sm/f8BODu+QcM/uN60N2LKkmSJEnSajX/vn2JFh0bWEhbSz+2AU9O8qQk3wG8HPhwS31JkiRJkqSV58PAzw+qf5wO/EtV3bPYi1qZUVFV+5LMAH8NHA5cVlU72+hLkiRJkiQtvyRXAWcCRyfZDVwCrAGoqj8GPgqcA3wB+AbwikM6b9XoV108tO/ga1SmNswc9PWHksNiGOeQtLot9jkAfhZIkiS1ofM5Kn7410d/Y92CB29770iuW1tLPyRJkiRJkr5trQxUJLksyZ4kO9o4vyRJkiRJ6qa2ZlRcDmxs6dySJEmSJKmjWhmoqKqbga+3cW5JkiRJktRdrVT9kCRJkiRpbKTTuUKX3ciSaSaZTjKXZG52c29UYUiSJEmSpBVkZDMqqqoH9GDx8qSSJEmSJGk8WJ5UkiRJkiStGK3MqEhyFXAmcHSS3cAlVTXbRl+SJEmSJI1UnAMwTKka/aqLYSz9mNowc9D9e7dtatqFJEmSVqHFvieC3xWltk1O0Olsk2tPffXob6xb8ODcu0dy3Tox7HMof3wkSZIkSdLK14mBCkmSJEmS1A1t5ahYB3wAeCLwKNCrqve00ZckSZIkSSOVTq9sWXZtlSfdB1xcVbclORK4NcmWqvpMS/1JkiRJkqQOaGXpR1XdU1W3DR7fD+wCjm+jL0mSJEmS1B2t56hIciJwCrB1v/bpJHNJ5mY399oOQ5IkSZIkrQJtLf0AIMkRwDXARVV13/x9VdUDejCc8qSSJEmSJI1ErFMxTK29m0nW0B+kuLKqrm2rH0mSJEmS1B2tDFQkCTAL7KqqS9voQ5IkSZIkdU9bMyrOAM4DzkqyfbCd01JfkiRJkiSpI1I1+vQQy5GjYmrDzEH37922qe0QpJHx51+SJEmjNDlBRh1Dm9ae9prR31i34MFb3jmS69ZqMk1JkiRJkjovnR6HWXZt5aiYTHJLktuT7Ezyljb6kSRJkiRJ3dLWjIqHgbOq6oFB9Y9PJLmhqj7VUn+SJEmSJKkDWhmoqH7iiwcGT9cMtk6u2ZEkSZIkScPTVtUPkhyeZDuwB9hSVVv32z+dZC7J3OzmXlthSJIkSZLUrhzWzW1EWkumWVWPAOuTHAVcl+Tkqtoxb38P6MHyVP2QJEmSJEkrX+tDJFV1L3ATsLHtviRJkiRJ0urWVtWPYwYzKUiyFjgbuKuNviRJkiRJUne0tfTjOOCKJIfTHwy5uqqub6kvSZIkSZJGJxl1BJ2SfoGO0VoJOSqmNsw0PsfebZuGEIkkSVrtFvte4XcGSeNmcoJO38mvfebrRn5P24YHP/m2kVy30aXxlCRJkiRJ2k/b5Uk/ncQlH5IkSZIk6ZC0Vp4UuBDYBTyhxT4kSZIkSRqtuFhhmNqq+nEC8OPA+9o4vyRJkiRJ6qa2hn3+I/Ba4NGWzi9JkiRJkjpo6AMVSX4C2FNVty5y3HSSuSRzs5t7ww5DkiRJkiStQm3kqDgDeGGSc4BJ4AlJ/qSqfm7+QVXVA3qwMsqTSpIkSZKk0Rv6QEVVvR54PUCSM4HX7D9IIUmSJElSZySjjqBTTE0qSZIkSZJWjDbLk1JVNwE3tdnHsOzdtumg+6c2zCxTJJIkabVb7HvFcjiU7y4rIU5ptfJ3TGqPMyokSZIkSdKK0dqMiiRfBO4HHgH2VdWpbfUlSZIkSdLIxDkAw9Tq0g/guVX1tZb7kCRJkiRJHeGwjyRJkiRJWjHaHKgo4GNJbk0y3WI/kiRJkiSpI9ocqDijqn4YeAHwqiTPnr8zyXSSuSRzs5t7LYYhSZIkSVKLkm5uI9Jajoqqunvw754k1wGnATfP298DegAP7aPaikOSJEmSJK0ercyoSPL4JEc+9hh4PrCjjb4kSZIkSVJ3tDWj4nuB69KfKjIB/GlV/VVLfUmSJEmSpI5oZaCiqv4eeEYb55YkSZIkaUWJBTWHKVWjTw/RlRwVUxtmDrp/77ZNyxSJJEnjy7/HkrTyTE4wusyMy2Dts9/ciXva/T1485tHct0c9pEkSZIkSStGawMVSY5K8qEkdyXZleSZbfUlSZIkSZK6obXypMB7gL+qqp9O8h3A41rsS5IkSZIkdUArAxVJngA8G/gFgKr6JvDNNvqSJEmSJGmkTKY5VG29m98P/DPw/iSfTvK+JI+ff0CS6SRzSeZmN/daCkOSJEmSJK0mbS39mAB+GPi1qtqa5D3A64DffuyAquoBPehO1Q9JkiRJktRMWzMqdgO7q2rr4PmH6A9cSJIkSZIkHVArMyqq6p+SfDnJSVX1WeB5wGfa6EuSJEmSpJE6LKOOoFParPrxa8CVg4offw+8osW+JEmSJElSB6Rq9OkhxiVHxdSGmYPu37tt0zJFIkmSJOlg/O4+XJMTdHrKwdrnvrWT97QP3vjbI7lu1lCRJEmSJEkrRitLP5KcBHxwXtP3A2+qqv/YRn+SJEmSJI1MnAMwTG0l0/wssB4gyeHAV4Dr2uhLkiRJkiR1x3IM+zwP+Luq+sdl6EuSJEmSJK1iyzFQ8XLgqv0bk0wnmUsyN7u5twxhSJIkSZKkla7N8qQMSpO+EHj9/vuqqgf0YHyqfkiSJEmSOiidLmqy7NqeUfEC4Laq+mrL/UiSJEmSpA5oe6DiXBZY9iFJkiRJkrSQ1gYqkjwO+DHg2rb6kCRJkiRJ3dJajoqq+gbwPW2dfzXau23TQfdPbZhpfA5JkiRJzfm9WxqdVpNpSpIkSZLUeVmOgprjo82lH69OsjPJjiRXJZlsqy9JkiRJktQNrQxUJDke+HXg1Ko6GTgceHkbfUmSJEmSpO5oc37KBLA2yQTwOODuFvuSJEmSJEkd0EqOiqr6SpJ3Al8CHgQ+VlUfa6MvSZIkSZJGKhl1BJ3S1tKPKeBFwJOA7wMen+Tn9jtmOslckrnZzb02wpAkSZIkSatMW1U/zgb+oar+GSDJtcCPAn/y2AFV1QN6AA/to1qKQ5IkSZIkrSJt5aj4EnB6ksclCfA8YFdLfUmSJEmSpI5oK0fF1iQfAm4D9gGfZjB7QpIkSZKkTkmbdSrGT6pGv+rCpR+HbmrDzEH37922aZkikSRp+S32dxD8W6jx5ndFrVSTE3Q62+Ta57+jk/e0D37sN0Zy3Rz2kSRJkiRJK4YDFZIkSZIkacVoqzzphUl2JNmZ5KI2+pAkSZIkaUVIurmNyNAHKpKcDPwScBrwDOAnkjx52P1IkiRJkqTuaWNGxVOBT1XVN6pqH/A3wEta6EeSJEmSJHVMGwMVO4BnJ/meJI8DzgHW7X9Qkukkc0nmZjdbuVSSJEmSJMHEsE9YVbuS/D6wBXgAuB3Yt8BxPaAHlieVJEmSJEl9Qx+oAKiqWWAWIMnvAbvb6EeSJEmSpJGLBTWHqZWBiiTHVtWeJP8KeCnwzDb6kSRJkiRJ3dLKQAVwTZLvAf4H8Kqq2ttSP5IkSZIkqUNSNfr0EOaoGJ6pDTMH3b9326ZliqR94/TfKknSauXfa0kAkxNk1DG0ae3GSzt5T/vgX/3fI7lubc2okCRJkiRpPKTT4zDLrlHGjySXJdmTZMe8tu9OsiXJ5wf/TjUPU5IkSZIkjYOmqUkvBzbu1/Y64ONV9WTg44PnkiRJkiRJi2o0UFFVNwNf36/5RcAVg8dXAC9u0ockSZIkSRofbRR7/d6qugdg8O+xCx2UZDrJXJK52c29FsKQJEmSJGkZ5LBubiMysmSaVdUDemDVD0mSJEmS1NfGEMlXkxwHMPh3Twt9SJIkSZKkDmpjoOLDwPmDx+cDf9FCH5IkSZIkqYMaLf1IchVwJnB0kt3AJcDbgKuTXAB8CXhZ0yAlSZIkSVqxklFH0CmpGn16CHNULJ+pDTOLHrN326ZliESSJEldtNj3Tb9rjqfJCTp9J7/2x9/byXvaBz/y6yO5bqNL4ylJkiRJkrSfRgMVSS5LsifJjnltL0uyM8mjSU5tHqIkSZIkSRoXTWdUXA5s3K9tB/BS4OaG55YkSZIkSWOmUTLNqro5yYn7te0CiMlEJEmSJEnjIGZVGKaRvZtJppPMJZmb3dwbVRiSJEmSJGkFaTSjoomq6gE9sOqHJEmSJEnqc36KJEmSJElaMUY2o0KSJEmSpE4wR8VQNS1PehXwSeCkJLuTXJDkJUl2A88EPpLkr4cRqCRJkiRJ6r5UjT49hDkqVpapDTMH3b9326ZlikSSpPHl32NJXTI5QafLQq79yT/s5D3tg3/570dy3ZyfIkmSJEmSVoymSz8uS7InyY55be9IcleSO5Jcl+So5mFKkiRJkrRCJd3cRqTpjIrLgY37tW0BTq6qpwOfA17fsA9JkiRJkjQmGg1UVNXNwNf3a/tYVe0bPP0UcEKTPiRJkiRJ0vhoO0fFK4EbWu5DkiRJkiR1RGsDFUneCOwDrjzA/ukkc0nmZjf32gpDkiRJkqR25bBubiMy0cZJk5wP/ATwvDpA/dOq6gE9sDypJEmSJEnqG/pARZKNwG8Cz6mqbwz7/JIkSZIkqbualie9CvgkcFKS3UkuADYBRwJbkmxP8sdDiFOSJEmSJI2BRjMqqurcBZpnm5xTo7d326aD7p/aMNP4HJIkaeXzb74WstjPhT8TkppqJUeFJEmSus8bUkkaSEYdQaeMLo2nJEmSJEnSfprmqLgsyZ4kO+a1vTXJHYP8FB9L8n3Nw5QkSZIkSeOg6YyKy4GN+7W9o6qeXlXrgeuBNzXsQ5IkSZIkjYmmyTRvTnLifm33zXv6eKCa9CFJkiRJ0ooWsyoMUyvvZpLfTfJl4N9xgBkVSaaTzCWZm93cayMMSZIkSZK0yrRS9aOq3gi8McnrgRngkgWO6QE9gIf2OetCkiRJkiS1X/XjT4GfarkPSZIkSZLUEUOfUZHkyVX1+cHTFwJ3DbsPSZIkSZJWjGTUEXRKo4GKJFcBZwJHJ9lNf4nHOUlOAh4F/hH4laZBSpIkSZKk8ZCq0aeHMEdF90xtmDno/r3bNi1TJAKvxzD5Xo6fxa45eN0lSVrM5ASdnnKw9qWznbynffDaC0Zy3ayhIkmSJEmSVoxGAxVJLkuyJ8mOBfa9JkklObpJH5IkSZIkrWRJOrmNStMZFZcDG/dvTLIO+DHgSw3PL0mSJEmSxkijgYqquhn4+gK73g28Fsw9IUmSJEmSDt3Qc1QkeSHwlaq6fZHjppPMJZmb3dwbdhiSJEmSJGkValSedH9JHge8EXj+YsdWVQ/ogVU/JEmSJEmr1yjzOXTRsGdU/GvgScDtSb4InADcluSJQ+5HkiRJkiR10FBnVFTVncCxjz0fDFacWlVfG2Y/kiRJkiSpm5qWJ70K+CRwUpLdSS4YTliSJEmSJGkcpWr06SHMUTF+pjbMHHT/3m2blikSSZL+V4v9jYLV8XeqK/8dkrphcoJOJ3F4/E+/v5P3tP/fh14xkus21KUfkiRJkiSNnU4Pwyy/pks/LkuyJ8mOeW1vTvKVJNsH2znNw5QkSZIkSeOgadWPy4GNC7S/u6rWD7aPNuxDkiRJkiSNiUYDFVV1M/D1IcUiSZIkSZLGXNMZFQcyk+SOwdKQqYUOSDKdZC7J3OzmXkthSJIkSZLUriSd3EaljWSafwS8FajBv+8CXrn/QVXVA3pg1Q9JkiRJktQ39BkVVfXVqnqkqh4FNgOnDbsPSZIkSZLUTUMfqEhy3LynLwF2HOhYSZIkSZKk+Rot/UhyFXAmcHSS3cAlwJlJ1tNf+vFF4JcbxihJkiRJ0oo1ynwOXZSq0aeH6EqOiqkNMwfdv3fbpmWKZPVb7L0E30+tTP7sSuoSP9MkDcvkBJ2+kz/yZ6/oxD3t/u7/4PkjuW5tVf2QJElSxzlIIUlqQ6OBikH50T1JduzX/mtJPptkZ5K3NwtRkiRJkiSNi6blSS8HNgEfeKwhyXOBFwFPr6qHkxzbsA9JkiRJklYsc1QMV6MZFVV1M/D1/Zp/FXhbVT08OGZPkz4kSZIkSdL4aCNHxVOAZyXZmuRvkmxooQ9JkiRJktRBbQxUTABTwOnAbwBXZ4F5MEmmk8wlmZvd3GshDEmSJEmStNo0zVGxkN3AtdWve3pLkkeBo4F/nn9QVfWAHnSnPKkkSZIkSWqmjYGKPwfOAm5K8hTgO4CvtdCPJEmSJEkjZzLN4Wo0UJHkKuBM4Ogku4FLgMuAywYlS78JnD+YXSFJkiRJknRQWQljCC790FJMbZg56P692zYtUySSJEmrg9+fNCqTE3R6ysF3nfufO3lP+y9XnTeS69ZGMk1JkiRJkqQlaSNHhSRJkiRJ46PT80WWX6MZFUkuS7JnkI/isbYPJtk+2L6YZHvzMCVJkiRJ0jhoOqPicmAT8IHHGqrqZx97nORdwL807EOSJEmSJI2JRgMVVXVzkhMX2pd+fZafoV+qVJIkSZIkaVFtJtN8FvDVqvr8QjuTTCeZSzI3u7nXYhiSJEmSJLUnSSe3UWkzmea5wFUH2llVPaAHlieVJEmSJEl9rQxUJJkAXgr8SBvnlyRJkiRJ3dTW0o+zgbuqandL55ckSZIkSR3UaEZFkquAM4Gjk+wGLqmqWeDlHGTZhyRJkiRJXTHKfA5dlKrRp4cwR4XaMLVh5qD7927btEyRSFJzfqYdOt8rtcGfK6mZyQk6fSc/9XNXdvKedu+f/LuRXLc2q35IkiRJkiR9WxoNVCS5LMmeJDvmta1P8qkk2wflR09rHqYkSZIkSRoHTWdUXA5s3K/t7cBbqmo98KbBc0mSJEmSpEU1SqZZVTcnOXH/ZuAJg8ffBdzdpA9JkiRJklYyk2kOVxs5Ki4C3pHky8A7gdcvdFCS6cHSkLnZzb0WwpAkSZIkSatNoxkVB/CrwKur6pokPwPMAmfvf1BV9YAeWPVDkiRJkiT1tTGj4nzg2sHj/wKYTFOSJEmSJB2SNmZU3A08B7gJOAv4fAt9SJIkSZK0IpijYrgaDVQkuQo4Ezg6yW7gEuCXgPckmQAeAqabBilJkiRJksZDqkafHsIcFX1TG2Yan2Pvtk1DiGQ8HMr77fspSSuLn92StDpNTtDpKQff8/NXdfKe9v/9wLkjuW5t5KiQJEmSJElakkYDFUkuS7InyY55bc9I8skkdyb5yyRPaB6mJEmSJEkrVDq6jUjTGRWXAxv3a3sf8Lqq+iHgOuA3GvYhSZIkSZLGRKOBiqq6Gfj6fs0nATcPHm8BfqpJH5IkSZIkaXy0kaNiB/DCweOXAesWOijJdJK5JHOzm3sthCFJkiRJklabRuVJD+CVwHuTvAn4MPDNhQ6qqh7QA6t+SJIkSZJWr6TTRU2W3dAHKqrqLuD5AEmeAvz4sPuQJEmSJEndNPSlH0mOHfx7GPBbwB8Puw9JkiRJktRNTcuTXgV8Ejgpye4kFwDnJvkccBdwN/D+5mFKkiRJkqRxkKrRp4cwR4VWqqkNMwfdv3fbpmWKRJIkqbnl+G7j9yctZHKCTidxOPoX/qyT97Rfu/zlI7lubSTTlCRJkiRpbJhMc7iWvPQjybokNybZlWRnkgsH7d+dZEuSzw/+nRpeuJIkSZIkqcua5KjYB1xcVU8FTgdeleRpwOuAj1fVk4GPD55LkiRJkiQtaskDFVV1T1XdNnh8P7ALOB54EXDF4LArgBc3DVKSJEmSJI2HoeSoSHIicAqwFfjeqroH+oMZj5UrlSRJkiSpi8xRMVyNypMCJDkCuAa4qKru+zZeN51kLsnc7OZe0zAkSZIkSVIHNJpRkWQN/UGKK6vq2kHzV5McN5hNcRywZ6HXVlUP6IHlSSVJkiRJUl+Tqh8BZoFdVXXpvF0fBs4fPD4f+IulhydJkiRJksZJkxkVZwDnAXcm2T5oewPwNuDqJBcAXwJe1ixESZIkSZJWMFNUDFWqRr/qwqUfWq2mNswcdP/ebZuWKRJJGg+Lfe6Cn72ShsPvecM1OdHtW/ljL7i6k/e0e2Z/ZiTXrXEyTUmSJEmSpGFxoEKSJEmSJK0YTZJprktyY5JdSXYmuXDQ/rLB80eTnDq8UCVJkiRJWnmSdHIblSbJNPcBF1fVbUmOBG5NsgXYAbwU+H+GEaAkSZIkSRofS55RUVX3VNVtg8f3A7uA46tqV1V9dlgBSpIkSZKklSnJxiSfTfKFJK9bYP+/GqzG+HSSO5Kcs9g5h5KjIsmJwCnA1m/jNdNJ5pLMzW7uDSMMSZIkSZK0TJIcDvwB8ALgacC5SZ6232G/BVxdVacALwf+cLHzNln68VhgRwDXABdV1X2H+rqq6gE9sDypJEmSJEmr0GnAF6rq7wGS/BnwIuAz844p4AmDx98F3L3YSRsNVCRZQ3+Q4sqqurbJuSRJkiRJWo1GmXiyTUmmgel5Tb3BpIPHHA98ed7z3cD/sd9p3gx8LMmvAY8Hzl6s3yUPVKR/JWaBXVV16VLPI0mSJEmSVp75KyEOYKERmv1XTJwLXF5V70ryTOA/Jzm5qh490Emb5Kg4AzgPOCvJ9sF2TpKXJNkNPBP4SJK/btCHJEmSJElamXYD6+Y9P4H/fWnHBcDVAFX1SWASOPpgJ03V6NNDmKNCXTW1YWbRY/Zu27QMkUjS4hb7zPLzavz4d0wL8bNCSzE5seD/ee+MJ/7Shzp5T/tPm3/6oNctyQTwOeB5wFeAbcC/raqd8465AfhgVV2e5KnAx+lXDD3ge9Y4maYkSZLGkzekktTX1RwVi6mqfUlmgL8GDgcuq6qdSf4DMFdVHwYuBjYneTX9ZSG/cLBBCmiWo2Id8AHgicCj9JNqvCfJO4CfBL4J/B3wiqq6d6n9SJIkSZKklamqPgp8dL+2N817/Bn6qSMOWZMcFfuAi6vqqcDpwKsG9VK3ACdX1dPpTwF5fYM+JEmSJEnSGFnyQEVV3VNVtw0e3w/sor/O5GNVtW9w2KfoJ9OQJEmSJElaVJMZFd+S5ETgFGDrfrteCdxwgNdMJ5lLMje7+WDVTiRJkiRJWrmSdHIblcbJNJMcAVwDXFRV981rfyP95SFXLvS6+fVYrfohSZIkSZKg4UBFkjX0BymurKpr57WfD/wE8LzFsnlKkiRJkiQ9pknVjwCzwK6qunRe+0bgN4HnVNU3mocoSZIkSZLGRZMZFWcA5wF3Jtk+aHsD8F7gO4EtgzUtn6qqX2kUpSRJkiRJK9Xo0jl0UlbCygxzVGicTW2YOej+vds2LVMkkiRJ7Vvsuw/4/aeLJie6fSv/fb9ybSfvae/+45eO5LoNpeqHJEmSJEnSMCx5oCLJuiQ3JtmVZGeSCwftb01yR5LtST6W5PuGF64kSZIkSeqyJjMq9gEXV9VTgdOBVyV5GvCOqnp6Va0HrgfeNIQ4JUmSJEnSGFhyMs2quge4Z/D4/iS7gOOr6jPzDns8mH9CkiRJktRdg0ISGpImVT++JcmJwCnA1sHz3wV+HvgX4LnD6EOSJEmSJHVf42SaSY4ArgEuqqr7AKrqjVW1DrgSWDCtb5LpJHNJ5mY395qGIUmSJEmSOqDRjIoka+gPUlxZVdcucMifAh8BLtl/R1X1gB5YnlSSJEmSJPUteaAi/UU4s8Cuqrp0XvuTq+rzg6cvBO5qFqIkSZIkSSuXOSqGq8mMijOA84A7k2wftL0BuCDJScCjwD8Cv9IsREmSJEmSNC5SNfpVFy79kA5sasOCaV6+Ze+2TcsUiaS2LPZ7Dv6uP8b36tD5XklaSSYn6PSUgxP+/Z938p529x++eCTXrXEyTUmSJEmSpGFZ8kBFknVJbkyyK8nOJBfut/81SSrJ0c3DlCRJkiRpZUrSyW1UmuSo2AdcXFW3JTkSuDXJlqr6TJJ1wI8BXxpKlJIkSZIkaSwseUZFVd1TVbcNHt8P7AKOH+x+N/BaMPeEJEmSJEk6dEPJUZHkROAUYGuSFwJfqarbh3FuSZIkSZI0PhoPVCQ5ArgGuIj+cpA3Am86hNdNJ5lLMje7udc0DEmSJEmSRiMd3UakSY4KkqyhP0hxZVVdm+SHgCcBtw8Sb5wA3JbktKr6p/mvraoe0APLk0qSJEmSpL4lD1SkPxIxC+yqqksBqupO4Nh5x3wROLWqvtYwTkmSJEmSNAaaLP04AzgPOCvJ9sF2zpDikiRJkiRJYyhVo1914dIPaemmNswseszebZuWIRJJkjRqi30vGMZ3guXoYzl05b9jtZicGGXGg/atm/mLTt7TfnnTi0Zy3RrlqJAkSZIkadwNcjRqSIZSnlSSJEmSJGkYljxQkWRdkhuT7EqyM8mFg/Y3J/mKeSskSZIkSdK3q8nSj33AxVV1W5IjgVuTbBnse3dVvbN5eJIkSZIkaZwseaCiqu4B7hk8vj/JLuD4YQUmSZIkSdJqYI6K4RpKjookJwKnAFsHTTNJ7khyWbLR0eYAACAASURBVJKpA7xmOslckrnZzb1hhCFJkiRJkla5xlU/khwBXANcVFX3Jfkj4K1ADf59F/DK/V9XVT2gB5YnlSRJkiRJfY1mVCRZQ3+Q4sqquhagqr5aVY9U1aPAZuC05mFKkiRJkqRxsOQZFekvwpkFdlXVpfPajxvkrwB4CbCjWYiSJEmSJK1c5qgYriZLP84AzgPuTLJ90PYG4Nwk6+kv/fgi8MuNIpQkSZIkSWMjVaNPD2GOiu6Z2jBz0P17t21apkgEXg+picV+f8DfoXHk56okfXsmJ+j0lIMTL7y+k/e0X3zPT4zkug2l6ockSZIkSdIwLHmgIsm6JDcm2ZVkZ5IL5+37tSSfHbS/fTihSpIkSZK08iTp5DYqTXJU7AMurqrbkhwJ3JpkC/C9wIuAp1fVw0mOHUagkiRJkiSp+5Y8UDGo7HHP4PH9SXYBxwO/BLytqh4e7NszjEAlSZIkSVL3DSVHRZITgVOArcBTgGcl2Zrkb5JsOMBrppPMJZmb3dwbRhiSJEmSJGmVa7L0A4AkRwDXABdV1X1JJoAp4HRgA3B1ku+v/cqLVFUP6IFVPyRJkiRJUl+jgYoka+gPUlxZVdcOmncD1w4GJm5J8ihwNPDPjSKVJEmSJGkl6nTx1eXXpOpHgFlgV1VdOm/XnwNnDY55CvAdwNeaBClJkiRJksZDkxkVZwDnAXcm2T5oewNwGXBZkh3AN4Hz91/2IUmSJEmStJCshDEEc1T0TW2YWfSYvds2LUMkGjeL/ez5cydJkvTt8zvW/zQ50e3FEU969Uc6eU/7D+/+8ZFct8bJNCVJkiRJGmf9zAgaliUPVCRZB3wAeCLwKNCrqvck+SBw0uCwo4B7q2p940glSZIkSVLnNZlRsQ+4uKpuS3IkcGuSLVX1s48dkORdwL80DVKSJEmSJI2HJQ9UVNU9wD2Dx/cn2QUcD3wGvlUV5GcYVACRJEmSJElazFByVCQ5ETgF2Dqv+VnAV6vq88PoQ5IkSZKklcgcFcN1WNMTJDkCuAa4qKrum7frXOCqg7xuOslckrnZzb2mYUiSJEmSpA5oNKMiyRr6gxRXVtW189ongJcCP3Kg11ZVD+iB5UklSZIkSVLfkmdUDHJQzAK7qurS/XafDdxVVbubBCdJkiRJksZLkxkVZwDnAXcm2T5oe0NVfRR4OQdZ9iFJkiRJUleYomK4UjX6VRcu/ZBWtqkNM4ses3fbpmWIRMvFay61y98xafQW+z30d3C4Jifo9K38D7zmhk7e037hnS8YyXVrnExTkiRJkiRpWJrkqFiX5MYku5LsTHLhoH19kk8l2T6o6nHa8MKVJEmSJEld1iRHxT7g4qq6LcmRwK1JtgBvB95SVTckOWfw/MzmoUqSJEmSpK5b8kBFVd0D3DN4fH+SXcDxQAFPGBz2XcDdTYOUJEmSJGmlitk0h6rJjIpvSXIicAqwFbgI+Osk76S/tORHh9GHJEmSJEnqvsbJNJMcAVwDXFRV9wG/Cry6qtYBrwZmD/C66UEOi7nZzb2mYUiSJEmSpA5oNKMiyRr6gxRXVtW1g+bzgQsHj/8L8L6FXltVPaAHlieVJEmSJEl9Sx6oSH8Rziywq6ounbfrbuA5wE3AWcDnmwQoSZIkSdJKZoqK4Woyo+IM4DzgziTbB21vAH4JeE+SCeAhYLpZiJIkSZIkaVykavSrLlz6Ia1+UxtmDrp/77ZNyxRJM13575BWK38HJa0Wi31egZ9Z801O0Ok5B0957V918p72c2/fOJLr1jiZpiRJkiRJ0rAMpTypJEmSJEnjKiapGKolz6hIsi7JjUl2JdmZ5MJB+zOSfDLJnUn+MskThheuJEmSJEnqsiZLP/YBF1fVU4HTgVcleRr9cqSvq6ofAq4DfqN5mJIkSZIkaRwseaCiqu6pqtsGj+8HdgHHAycBNw8O2wL8VNMgJUmSJEnSeBhKMs0kJwKnAFuBHcALB7teBqw7wGumk8wlmZvd3BtGGJIkSZIkLbukm9uoNE6mmeQI4Brgoqq6L8krgfcmeRPwYeCbC72uqnpADyxPKkmSJEmS+hoNVCRZQ3+Q4sqquhagqu4Cnj/Y/xTgx5sGKUmSJEmSxkOTqh8BZoFdVXXpvPZjB/8eBvwW8MdNg5QkSZIkSeOhSY6KM4DzgLOSbB9s5wDnJvkccBdwN/D+IcQpSZIkSZLGQKpGnx5i7SkzBw1i77ZNyxWKpJZMbZg56H5/z1cfr6kkLa/FPndh8c9eP7s1KpMTjDA1Y/ue9oaPjf7GugWf+b3nj+S6DaXqhyRJkiRJ0jA0yVExmeSWJLcn2ZnkLYP2JyXZmuTzST6Y5DuGF64kSZIkSeqyJjMqHgbOqqpnAOuBjUlOB34feHdVPRnYC1zQPExJkiRJkjQOljxQUX0PDJ6uGWwFnAV8aNB+BfDiRhFKkiRJkrSCJd3cRqVRjookhyfZDuwBtgB/B9xbVfsGh+wGjj/Aa6eTzCWZ2/e1nU3CkCRJkiRJHdFooKKqHqmq9cAJwGnAUxc67ACv7VXVqVV16sTR/6ZJGJIkSZIkqSOGUvWjqu4FbgJOB45KMjHYdQJw9zD6kCRJkiRJ3dek6scxSY4aPF4LnA3sAm4Efnpw2PnAXzQNUpIkSZKklSpJJ7dRmVj8kAM6DrgiyeH0Bzyurqrrk3wG+LMkvwN8GpgdQpySJEmSJGkMpGrBFBLL6qF9C+exkDQ+pjbMLHrM3m2bliESSavdYp8nfpZIWi5+Hv1PkxOMsIZE+07+rS2dvKfd8Ts/NpLrNpQcFZIkSZIkScPQJEfFZJJbktyeZGeStwzaZ5J8IUklOXp4oUqSJEmStPIk3dxGpUmOioeBs6rqgSRrgE8kuQH4b8D19KuASJIkSZIkHbIlD1RUP7nFA4OnawZbVdWngZFmCJUkSZIkSatToxwVSQ5Psh3YA2ypqq3DCUuSJEmSJI2jRgMVVfVIVa0HTgBOS3Lyob42yXSSuSRzs5t7TcKQJEmSJEkd0SRHxbdU1b1JbgI2AjsO8TU9oAeWJ5UkSZIkrV6mPhiuJlU/jkly1ODxWuBs4K5hBSZJkiRJksZPk6UfxwE3JrkD2EY/R8X1SX49yW76y0HuSPK+YQQqSZIkSZK6r0nVjzuAUxZofy/w3iZBSRo/e7dtWvSYqQ0zjc8hSVpefnZrnPnzLS3NUHJUSJIkSZI0rsxRMVxNclRMJrklye1JdiZ5y6D9yiSfTbIjyWVJ1gwvXEmSJEmS1GVNclQ8DJxVVc8A1gMbk5wOXAn8IPBDwFrgFxtHKUmSJEmSxkKTHBUFPDB4umawVVV99LFjktxCP6mmJEmSJEnSohrlqEhyOHAr8APAH1TV1nn71gDnARc2ilCSJEmSpBXMFBXD1WTpB1X1SFWtpz9r4rQkJ8/b/YfAzVX1twu9Nsl0krkkc7Obe03CkCRJkiRJHTGUqh9VdW+Sm4CNwI4klwDHAL98kNf0gB7AQ/uoYcQhSZIkSZJWtyZVP45JctTg8VrgbOCuJL8I/F/AuVX16HDClCRJkiRJ46DJjIrjgCsGeSoOA66uquuT7AP+EfjkoJbstVX1H5qHKkmSJEnSyhOTVAxV+sU7RsulH5KGYWrDzEH37922aZkikSRpNPxbqJVqcoJO38mf8pb/2sl72k9fctZIrlujZJqSJEmSJEnD5ECFJEmSJElaMZok05xMckuS25PsTPKWQfvsoO2OJB9KcsTwwpUkSZIkSV3WJJnmw8BZVfVAkjXAJ5LcALy6qu4DSHIpMAO8rXmokiRJkiStPObSHK4lD1RUPwvnA4OnawZbzRukCLAWTJQpSZIkSZIOTaMcFUkOT7Id2ANsqaqtg/b3A/8E/CDwnw7w2ukkc0nmZjf3moQhSZIkSZI6osnSD6rqEWB9kqOA65KcXFU7quoVSQ6nP0jxs8D7F3htD+iB5UklSZIkSVLfUKp+VNW9wE3AxnltjwAfBH5qGH1IkiRJkrQSJenkNipNqn4cM5hJQZK1wNnAZ5P8wKAtwE8Cdw0jUEmSJEmS1H1Nln4cB1wxWOJxGHA18BHgb5M8AQhwO/CrjaOUJEmSJEljIf3iHaNljgpp9ZvaMHPQ/Xu3bVqmSA5ssRhhZcQpqZnV8HkkjYq/HxqVyQk6XcDzR956YyfvaW/97eeO5Lo1SqYpSZIkSdK4G2E6h05qkqNiMsktSW5PsjPJW/bb/5+SPNA8REmSJEmSNC6azKh4GDirqh5Isgb4RJIbqupTSU4FjhpOiJIkSZIkaVwseUZF9T02Y2LNYKtBcs13AK8dQnySJEmSJGmMLHmgAiDJ4Um2A3uALVW1FZgBPlxV9yzy2ukkc0nmZjf3moQhSZIkSdLIJOnkNiqNkmlW1SPA+iRHAdcleTbwMuDMQ3htD+iBVT8kSZIkSVJfoxkVj6mqe4GbgOcCPwB8IckXgccl+cIw+pAkSZIkSd3XpOrHMYOZFCRZC5wN3FpVT6yqE6vqROAbVfUDwwlVkiRJkiR1XZOlH8cBVwySZx4GXF1V1w8nLEmSJEmSNI5SNfr0EOaokEZrasPMQffv3bZpmSIZPd+LvsXeBxif92IY/LmSJI27yQlGl5lxGZz2ezd18p72ljecOZLrNpQcFZIkSZIkScPQJEfFZJJbktyeZGeStwzaL0/yD0m2D7b1wwtXkiRJkiR1WZMcFQ8DZ1XVA0nWAJ9IcsNg329U1YeahydJkiRJksbJkgcqqp/c4oHB0zWDrZPrciRJkiRJOpCk0yk4ll2jHBVJDk+yHdgDbKmqrYNdv5vkjiTvTvKdjaOUJEmSJEljodFARVU9UlXrgROA05KcDLwe+EFgA/DdwG8u9Nok00nmkszNbu41CUOSJEmSJHVEkxwV31JV9ya5CdhYVe8cND+c5P3Aaw7wmh7QA8uTSpIkSZKkviZVP45JctTg8VrgbOCuJMcN2gK8GNgxjEAlSZIkSVqJkm5uo9JkRsVxwBVJDqc/4HF1VV2f5L8mOQYIsB34lSHEKUmSJEmSxkD6xTtGy6UfklaLqQ0zB92/d9umZYpEGk+L/Q4Ow6H8HvtZMDyHck19P6UDWy2fR5MTdLosxulv+5tO3tN+6nXPGcl1a5RMU5IkSZIkaZiWvPQjySRwM/Cdg/N8qKouGeSm+B3gZcAjwB9V1XuHEawkSZIkSStNRpnQoYOa5Kh4GDirqh5Isgb4RJIbgKcC64AfrKpHkxw7jEAlSZIkSVL3LXmgovrJLR4YPF0z2Ar4VeDfVtWjg+P2NA1SkiRJkiSNh0Y5KpIcnmQ7sAfYUlVbgX8N/GySuSQ3JHnyMAKVJEmSJEnd12igoqoeqar1wAnAaUlOpp+z4qGqOhXYDFy20GuTTA8GM+ZmN/eahCFJkiRJkjqiSY6Kb6mqe5PcBGwEdgPXDHZdB7z/AK/pAT2wPKkkSZIkafUyl+ZwLXlGRZJjkhw1eLwWOBu4C/hz4KzBYc8BPtc0SEmSJEmSNB6azKg4DrgiyeH0Bzyurqrrk3wCuDLJq+kn2/zFIcQpSZIkSZLGQPrFO0bLpR+SVoupDTONz7F326YhRKJDtdg183pIktS+yQk6vTjiR99+cyfvaf/7a589kus2lBwVkiRJkiSNq5ikYqgaVf2QJEmSJEkapiXPqEgyCdxMvxzpBPChqrokyd8CRw4OOxa4pape3DhSSZIkSZLUeU2WfjwMnFVVDyRZA3wiyQ1V9azHDkhyDfAXTYOUJEmSJEnjYckDFdXPwvnA4OmawfatBCJJjqRfpvQVTQKUJEmSJGklM0XFcDXKUZHk8CTbgT3AlqraOm/3S4CPV9V9B3jtdJK5JHOzm3tNwpAkSZIkSR3RqOpHVT0CrE9yFHBdkpOrasdg97nA+w7y2h7QA8uTSpIkSZKkvqFU/aiqe4GbgI0ASb4HOA34yDDOL0mSJEmSxkOTqh/HAP+jqu5NshY4G/j9we6XAddX1UNDiFGSJEmSpBUrJqkYqiZLP44DrkhyOP2ZGVdX1fWDfS8H3tY0OEmSJEmSNF7SL94xWuaokDROpjbMHHT/3m2blikSSZKk5TE5QaenHDzrXZ/o5D3t3178f47kug0lR4UkSZKk/5+9u4+2q67vff/+hC2mUCI5VSin4QykAo5RKpEmnOPNkPKgVIEGtajY2gLW7ranoVSvIjm2trXDDo/aCjW92JVEDr1CxaKpHNQUhh5q21FDduRBIPGJRokRo2eEVtuBuYHv/WPNeLZxs3fYc2WvnbneL8Yae82H35zfveZaa2d++f2+P0nSIMw6UZFkYZK7ktyb5IEkf9isPzfJ55Lck+QfkjxncOFKkiRJkqQua1Oj4nvAOVX13SRPA/4hySeB64CLqmprkv8K/C5wWftQJUmSJEmafyymOVizTlRUv7jFd5vFpzWPah6LmvXPAHa2CVCSJEmSJI2OVjUqkhyW5B5gF3BHVW0CXg98IskO4Jd5ktk/kownmUgysX5tr00YkiRJkiSpI9oM/aCqHgeWJjka2JDkVOANwPlVtSnJm4E/pZ+82L9tD+iBs35IkiRJkqS+VomKfarq0SR3Ai8FTmt6VgDcDGwcxDkkSZIkSZqPLFExWG1m/XhW05OCJD8CvAjYCjwjycnNbi9u1kmSJEmSJM2oTY+K44AbkhxGP+Hx4aq6LcmvAR9J8gSwG3jdAOKUJEmSJEkjoM2sH/cBz59i/QZgQ5ugJKnLdm9eM+32xctXtWovSfpBM32vgt+tkjSfDKRGhSRJkiRJoyoWqRioNjUqFia5K8m9SR5I8ofN+nOSfC7J/UluSGIyRJIkSZIkHZBZJyqA7wHnVNVpwFLgJUn+L+AG4JKqOhX4KnBp+zAlSZIkSdIomHWiovq+2yw+rXk8Dnyvqr7YrL8D+IV2IUqSJEmSpFHRpkcFSQ5Lcg+wi35S4i7gaUmWNbtcDBzfLkRJkiRJkuavpJuPYWmVqKiqx6tqKbAEOAP4KeAS4L1J7gK+A+ydqm2S8SQTSSbWr+21CUOSJEmSJHXEQApdVtWjSe4EXlJV7wFeCJDkPODkJ2nTA3oAj+2lBhGHJEmSJEk6tLWZ9eNZSY5unv8I8CJgW5JjmnVPB94CvH8QgUqSJEmSpO5r06PiOOCGJIfRT3h8uKpuS/LuJBc2666rqk8PIlBJkiRJktR9qRr+qAuHfkjSgVu8fNWM++zevGYOIpEkSTowC8cYYmnGg++cP/unTt7Tfvq3XzCU69aqmKYkSZIkSdIgtU5UNFOU3p3ktmb52Uk2JflSkpuTHN4+TEmSJEmSNAoG0aPiSmDrpOX/Dry3qk4CdgO/OoBzSJIkSZKkEdAqUZFkCXABsK5ZDnAOcEuzyw3Ay9qcQ5IkSZKk+Szp5mNY2vaouAa4CniiWf4x4NGq2tss7wB+ouU5JEmSJEnSiJh1oqKZgnRXVW2ZvHqKXaesfppkPMlEkon1a3uzDUOSJEmSJHXIWIu2K4CVSc4HFgKL6PewODrJWNOrYgmwc6rGVdUDeuD0pJIkSZIkqW/WiYqqWg2sBkhyFvCmqvqlJH8NXAx8CLgU+NgA4pQkSZIkaV5aMMyCDh00iFk/9vcW4I1Jvky/ZsX6g3AOSZIkSZLUQW2GfnxfVd0J3Nk8fwg4YxDHlST9sN2b18y4z+LlqwZyHEmzN9PncC4+g/MhhvnC10KSDh0Ho0eFJGme8x/k0sE1H26KDyRhOSp8LSTp0DKQHhWSJEmSJI0qS1QMVuseFUkOS3J3ktua5VVJvpykkjyzfYiSJEmSJGlUDGLox5XA1knL/wi8CPjqAI4tSZIkSZJGSKtERZIlwAXAun3rquruqtreMi5JkiRJkjSC2vaouAa4CnjiqTZMMp5kIsnE+rW9lmFIkiRJkqQumHUxzSQXAruqakuSs55q+6rqAT2Ax/ZSs41DkiRJkqRhitU0B6pNj4oVwMok24EPAeck+eBAopIkSZIkSSNp1omKqlpdVUuq6gTgEuDTVfXagUUmSZIkSZJGziBm/fgBSX47yQ5gCXBfknUztZEkSZIkSQJI1fDLQ1ijQpLm1uLlq6bdvnvzmjmKRHpqZnrvgu/ffXytusdrqkPZwjE6XcThpddt6uQ97Sd/8z8P5boNvEeFJEmSJEnSbLVOVCQ5LMndSW5rlm9M8oUk9yf5QJKntQ9TkiRJkiSNgkH0qLgS2Dpp+UbgucBPAz8CvH4A55AkSZIkSSNgrE3jJEuAC4B3AG8EqKpPTNp+F/2impIkSZIkdVLS6RIcc65tj4prgKuAJ/bf0Az5+GVg41QNk4wnmUgysX5tr2UYkiRJkiSpC2bdoyLJhcCuqtqS5Kwpdvl/gM9U1d9P1b6qekAPnPVDkiRJkqRDUZKXANcChwHrquqdU+zzKuAPgALurapfnO6YbYZ+rABWJjkfWAgsSvLBqnptkt8HngX8eovjS5IkSZKkeSrJYcCfAy8GdgCbk9xaVQ9O2uckYDWwoqp2JzlmpuPOOlFRVaubk9H0qHhTk6R4PfBzwLlV9UNDQiRJkiRJ6pIRLlFxBvDlqnoIIMmHgIuAByft82vAn1fVboCq2jXTQQcx68f+3g8cC/xTknuSvO0gnEOSJEmSJB1Ek2tLNo/x/Xb5CeDhScs7mnWTnQycnOQfk3y2GSoyrVazfuxTVXcCdzbPB3JMSRqGxctXTbt99+Y1cxTJwTXT7zHT63Agx5AOhgN5343K51ijx/eupLk2ubbkk5iqL8n+NSjHgJOAs+jPCvr3SU6tqkef7KAHo0eFJEmSJEnqvh3A8ZOWlwA7p9jnY1X1/1XVPwNfoJ+4eFKtExVJDktyd5LbmuX1Se5Ncl+SW5L8aNtzSJIkSZKkeWczcFKSZyc5HLgEuHW/ff4GOBsgyTPpDwV5aLqDDqJHxZXA1knLb6iq06rqecDXgJn7D0uSJEmSdIhKR/+bSVXtpX/P/7f08wIfrqoHkrw9ycpmt78F/neSB4H/Bby5qv73dMdtVU8iyRLgAuAdwBubQP+12RbgR/jh8SmSJEmSJKkDquoTwCf2W/e2Sc+Lfr7gjQd6zLY9Kq4BrgJ+YBrSJNcDjwDPBd7X8hySJEmSJGlEzDpRkeRCYFdVbdl/W1VdDvxH+l0/Xv0k7b8/zcn6tdMVEZUkSZIkSaOizdCPFcDKJOcDC4FFST5YVa8FqKrHk9wMvBm4fv/Gk6c5eWyvw0MkSZIkSYemBTOXc9BTMOseFVW1uqqWVNUJ9Ct7fhr45STPge/XqPh5YNsgApUkSZIkSd3XqpjmFALckGRR8/xe4DcHfA5JkiRJktRR6RfgHC6HfkjSoWfx8ulnn969ec0cRaJRMtP7Dg6N915Xfg9JOlALxw5grstD2Mre5k7e0946vnwo123QPSokSZIkSRop/coHGpS205OS5LAkdye5bb/170vy3bbHlyRJkiRJo6N1ogK4kv40pN+XZBlw9ACOLUmSJEmSRkirREWSJcAFwLpJ6w4D3g1c1S40SZIkSZI0atrWqLiGfkLiqEnrVgG3VtU3HKcjSZIkSeo6b30Ha9Y9KpJcCOyqqi2T1v1H4JXA+w6g/XiSiSQT69f2ZhuGJEmSJEnqkDY9KlYAK5OcDywEFgEPAN8Dvtz0pjgiyZer6jn7N66qHtADpyeVJEmSJEl9s+5RUVWrq2pJVZ0AXAJ8uqoWV9WPV9UJzfp/nypJIUmSJEmSNJVBzPohSZIkSZI0EG2LaQJQVXcCd06x/kcHcXxJkiRJkuarBVbTHKhUDb88hDUqJKl7Fi9fNeM+uzevmYNIJEnSsC0co9N38q9Yv6WT97Qf/dWfGcp1c+iHJEmSJEmaN1onKpIcluTuJLc1y/8jyT8nuad5LG0fpiRJkiRJGgWDqFFxJbCV/vSk+7y5qm4ZwLElSZIkSZrXLFExWK16VCRZAlwArBtMOJIkSZIkaZS1HfpxDXAV8MR+69+R5L4k703y9KkaJhlPMpFkYv3aXsswJEmSJElSF8x66EeSC4FdVbUlyVmTNq0GHgEOB3rAW4C379++qnrNdmf9kCRJkiRJQLsaFSuAlUnOBxYCi5J8sKpe22z/XpLrgTe1DVKSJEmSpPkqFqkYqFkP/aiq1VW1pKpOAC4BPl1Vr01yHED6V+plwP0DiVSSJEmSJHXeIGb92N+NSZ4FBLgH+I2DcA5JkiRJktRBA0lUVNWdwJ3N83MGcUxJ0qFt9+Y1M+6zePmq1seQNHt+BiVJ89HB6FEhSZIkSdLIsETFYLWdnpQkhyW5O8ltzXKSvCPJF5NsTfLb7cOUJEmSJEmjYBA9Kq4EtgKLmuXLgOOB51bVE0mOGcA5JEmSJEnSCGjVoyLJEuACYN2k1b8JvL2qngCoql1tziFJkiRJkkZH26Ef1wBXAU9MWveTwKuTTCT5ZJKTpmqYZLzZZ2L92l7LMCRJkiRJUhfMeuhHkguBXVW1JclZkzY9HXisqpYleQXwAeCF+7evqh7QA3hsLzXbOCRJkiRJGqYFVtMcqDY1KlYAK5OcDywEFiX5ILAD+Eizzwbg+nYhSpIkSZKkUTHroR9VtbqqllTVCcAlwKer6rXA3wDnNLv9LPDF1lFKkiRJkqSRMIhZP/b3TuDGJG8Avgu8/iCcQ5IkSZIkdVCqhl8ewhoVkqSpLF6+atrtuzevmaNIJElSGwvH6HQRh0tuuLuT97QfuvT5Q7lubWf9kCRJkiRJGpjWQz+SHAZMAF+vqguT/D1wVLP5GOCuqnpZ2/NIkiRJkqTuG0SNiiuBrcAigKr6/lSkST4CfGwA55AkSZIkSSOgVaIiyRLgAuAdwBv323YU/dk/Lm9zDkmSJEmS5rOk0yU45lzbGhXXAFcBT0yx7eXAp6rqX1ueQ5IkSZIkjYhZJyqSXAjsqqotT7LLa4C/mqb9eJKJJBPr1/ZmG4YkSZIkSeqQSAr5/gAAIABJREFUNkM/VgArk5wPLAQWJflgVb02yY8BZ9DvVTGlquoBPXB6UkmSJEmS1DfrREVVrQZWAyQ5C3hTVb222fxK4Laqeqx1hJIkSZIkzWMLLFExUG1rVDyZS5hm2IckSZIkSdJUBjE9KVV1J3DnpOWzBnHcUbN4+aoZ99m9ec0cRCJJ88NM33l+b2q+mum96ftSGg1+F0izc7B6VEiSJEmSJD1lA+lRIUmSJEnSqEosUjFIrXtUJDksyd1JbmuWz03yuST3JPmHJM9pH6YkSZIkSRoFgxj6cSWwddLydcAvVdVS4CbgdwdwDkmSJEmSNAJaJSqSLAEuANZNWl3Aoub5M4Cdbc4hSZIkSZJGR9seFdcAVwFPTFr3euATSXYAvwy8c6qGScaTTCSZWL+21zIMSZIkSZLUBbMuppnkQmBXVW1JctakTW8Azq+qTUneDPwp/eTFD6iqHtADeGwvNds4JEmSJEkaJmtpDlabWT9WACuTnA8sBBYl+Tjw3Kra1OxzM7CxZYySJEmSJGlEzHroR1WtrqolVXUCcAnwaeAi4BlJTm52ezE/WGhTkiRJkiTpSbXpUfFDqmpvkl8DPpLkCWA38LpBnkOSJEmSJHVXqoZfHsIaFfPL4uWrpt2+e/OaOYpEktrzO61bZrqe4DWVpPlo4RidruLwKzfd18l72r/8xecN5bq1nfVDkiRJkiRpYFonKpIcluTuJLc1y+ck+VyS+5PckGSgw0skSZIkSVJ3DaJHxZU0BTOTLABuAC6pqlOBrwKXDuAckiRJkiRpBLRKVCRZAlwArGtW/Rjwvar6YrN8B/ALbc4hSZIkSdJ8tiDdfAzt9WzZ/hrgKuCJZvnbwNOSLGuWLwaOn6phkvEkE0km1q/ttQxDkiRJkiR1wazrRyS5ENhVVVuSnAVQVZXkEuC9SZ4O3A7snap9VfWAHjjrhyRJkiRJ6mtT6HIFsDLJ+cBCYFGSD1bVa4EXAiQ5Dzi5fZiSJEmSJGkUzDpRUVWrgdUATY+KN1XVa5McU1W7mh4VbwHeMZBIJUmSJEmah5IhFnTooEHM+rG/NyfZCtwH/M+q+vRBOIckSZIkSeqgVA2/PIQ1KiRJw7J4+appt+/evGaOItF8MdN7AubmfeF7U/OV78355VC5HgvH6HSXg8s/9PlO3tNef8lPD+W6HYweFZIkSZIkSbPSppgmSbYD3wEeB/ZW1bIk/wG4GTgB2A68qqp2twtTkiRJkiSNgkH0qDi7qpZW1bJm+WrgU1V1EvCpZlmSJEmSpE5KRx/DcjCGflwE3NA8vwF42UE4hyRJkiRJ6qC2iYoCbk+yJcl4s+7YqvoGQPPzmKkaJhlPMpFkYv3aXsswJEmSJElSF7SqUQGsqKqdSY4B7kiy7UAbVlUP6IGzfkiSJEmSpL5WiYqq2tn83JVkA3AG8M0kx1XVN5IcB+waQJySJEmSJM1LC9Lp2Vfn3KyHfiQ5MslR+54D5wH3A7cClza7XQp8rG2QkiRJkiRpNLTpUXEssCH9zNEYcFNVbUyyGfhwkl8Fvga8sn2YkiRJkiRpFKRq+OUhrFHRPYuXr5p2++7Na+YoEoHXY5B8LUfPTNccvO6SJM1k4dhQZ7s86F5/8/2dvKdd9+pTh3Ld2hbTlCRJkiRppFmiYrBaTU+aZHuSzye5J8lEs+6VSR5I8kSSZYMJU5IkSZIkjYJB9Kg4u6q+PWn5fuAVwF8M4NiSJEmSJGmEDHzoR1VtBYh9XyRJkiRJ0lPUNlFRwO1JCviLquoNICZJkiRJkg4Z/o/6wWpVowJYUVWnAy8FfivJmQfaMMl4kokkE+vXmt+QJEmSJEkte1RU1c7m564kG4AzgM8cYNse0AOnJ5UkSZIkSX2z7lGR5MgkR+17DpxHv5CmJEmSJEnSrLQZ+nEs8A9J7gXuAj5eVRuTvDzJDuAFwMeT/O0gApUkSZIkSd2XquGPunDohyTpULZ4+appt+/evGaOIpH+j5nel+B7U9LcWThGp6tN/votD3TynvYvLv6poVy3tsU0JUmSJEmSBsZEhSRJkiRJmjdazfqRZDvwHeBxYG9VLUvybuDngT3AV4DLq+rRtoFKkiRJkqTua5WoaJxdVd+etHwHsLqq9ib578Bq4C0DOI8kSZIkSfPOgnS6BMecG/jQj6q6var2NoufBZYM+hySJEmSJKmb2iYqCrg9yZYk41Nsfx3wyakaJhlPMpFkYv3aXsswJEmSJElSF7Qd+rGiqnYmOQa4I8m2qvoMQJK3AnuBG6dqWFU9oAdOTypJkiRJkvpaJSqqamfzc1eSDcAZwGeSXApcCJxbVSYhJEmSJEmdZYmKwZr10I8kRyY5at9z4Dzg/iQvoV88c2VV/ftgwpQkSZIkSaOgTY+KY4EN6aeOxoCbqmpjki8DT6c/FATgs1X1G60jlSRJkiRJnTfrREVVPQScNsX657SKSJKkjlm8fNW023dvXjNHkehAzHS94NC4ZodCjPpBfldIUl/bYpqSJEmSJI20WKRioFolKpJsB74DPA7sraplSf4IuAh4AtgFXLav6KYkSZIkSdJ0Zl1Mc5Kzq2ppVS1rlt9dVc+rqqXAbcDbBnAOSZIkSZI0AgaRqPgBVfWvkxaPBJyeVJIkSZIkHZC2iYoCbk+yJcn4vpVJ3pHkYeCXeJIeFUnGk0wkmVi/ttcyDEmSJEmS1AVti2muqKqdSY6hPx3ptqr6TFW9FXhrktXAKuD3929YVT2gB/DYXntdSJIkSZIOTQMfqjDiWr2e+4pkVtUuYANwxn673AT8QptzSJIkSZKk0THrREWSI5Mcte85cB5wf5KTJu22EtjWLkRJkiRJkjQq2gz9OBbY0MwXOwbcVFUbk3wkySn0pyf9KvAb7cOUJEmSJEmjIFXDLw9hjYq+xctXzbjP7s1r5iASSdJc8vv/wPlaSdKhaeEYGXYMB9Nv/822Tt7T/tnLnjuU62bND0mSJEmSNG+0SlQk2Z7k80nuSTKx37Y3Jakkz2wXoiRJkiRJGhVtpycFOLuqvj15RZLjgRcDXxvA8SVJkiRJ0ogYRKJiKu8FrgI+dpCOL0mSJEnSvLCg0xU45l7bGhUF3J5kS5JxgCQrga9X1b3TNUwynmQiycT6tb2WYUiSJEmSpC5o26NiRVXtTHIMcEeSbcBbgfNmalhVPaAHzvohSZIkSZL6WvWoqKqdzc9dwAbgZ4FnA/cm2Q4sAT6X5MdbxilJkiRJkkbArHtUJDkSWFBV32menwe8vaqOmbTPdmDZ/sU2JUmSJEnqCmtUDFaboR/HAhuS7DvOTVW1cSBRSZIkSZKkkZSq4ZeHsEaFDobFy1dNu3335jVzFIkOxEzXC7xmkqTh8d8VUjsLx+h0n4M33rqtk/e0f7ryuUO5bm1n/ZAkSZIkSRqYVomKJNuTfD7JPUkmmnV/kOTrzbp7kpw/mFAlSZIkSVLXtZ2eFODsKYplvreq3jOAY0uSJEmSNK81tRs1IA79kCRJkiRJ80bbREUBtyfZkmR80vpVSe5L8oEki1ueQ5IkSZIkjYi2iYoVVXU68FLgt5KcCVwH/CSwFPgG8CdTNUwynmQiycT6tb2WYUiSJEmSpC5oVaOiqnY2P3cl2QCcUVWf2bc9yVrgtidp2wN64PSkkiRJkqRD1wJLVAzUrHtUJDkyyVH7ngPnAfcnOW7Sbi8H7m8XoiRJkiRJGhVtelQcC2xoqpuOATdV1cYk/2+SpfTrV2wHfr11lJIkSZIkaSSkavijLhz6IUlSO4uXr5p2++7Na+YoEh1KfN9ImisLx+j04Ig33/aFTt7TvvvCU4Zy3VrVqJAkSZIkadSl02mYudd21g9JkiRJkqSBaZWoSLI9yeeT3JNkYtL6K5J8IckDSd7VPkxJkiRJkjQKBjH04+yq+va+hSRnAxcBz6uq7yU5ZgDnkCRJkiRJI+Bg1Kj4TeCdVfU9gKradRDOIUmSJEnSvLDAIhUD1bZGRQG3J9mSZLxZdzLwwiSbkvxdkuVTNUwynmQiycT6tb2WYUiSJEmSpC5o26NiRVXtbIZ33JFkW3PMxcB/AZYDH05yYu03D2pV9YAeOD2pJEmSJEnqa9Wjoqp2Nj93ARuAM4AdwEer7y7gCeCZbQOVJEmSJEndN+tERZIjkxy17zlwHnA/8DfAOc36k4HDgW8/2XEkSZIkSZL2aTP041hgQ/pFQ8aAm6pqY5LDgQ8kuR/YA1y6/7APSZIkSZK6om3xR/2gzIccgjUqdKhavHzVtNt3b14zR5FI0vRm+r4Cv7MkDUZX/n3Uld9jvlg4Rqenxfhvn/hiJ+9p//j8k4dy3Uz8SJIkSZKkeaPVrB9JtgPfAR4H9lbVsiQ3A6c0uxwNPFpVS1tFKUmSJEmSRkLb6UkBzq6q7xfLrKpX73ue5E+AfxnAOSRJkiRJmpfS6YEtc28QiYoppV9l81U0M4BIkiRJkiTNpG2NigJuT7Ilyfh+214IfLOqvjRVwyTjSSaSTKxf22sZhiRJkiRJ6oK2PSpWVNXOJMcAdyTZVlWfaba9BvirJ2tYVT2gB876IUmSJEmS+lolKqpqZ/NzV5INwBnAZ5KMAa8AfqZ9iJIkSZIkzV8LLFIxULMe+pHkyCRH7XsOnAfc32x+EbCtqna0D1GSJEmSJI2KNj0qjgU29GtmMgbcVFUbm22XMM2wD0mSJEmSpKmkavjlIaxRoa5avHzVjPvs3rxmDiKRpJnN9J3l99Xo8e/Y6PGa62BZOEanx0b83sYvdfKe9o9ectJQrttBm55UkiRJ3eYNqyT1WaJisFolKpJsB74DPA7sraplSZYC7wcWAnuB/1pVd7UNVJIkSZIkdd8gelScXVXfnrT8LuAPq+qTSc5vls8awHkkSZIkSVLHzXrWj2kUsKh5/gxg50E4hyRJkiRJ6qC2iYoCbk+yJcl4s+53gHcneRh4D7B6qoZJxpNMJJlYv7bXMgxJkiRJktQFbYd+rKiqnUmOAe5Isg24GHhDVX0kyauA9cCL9m9YVT2gB876IUmSJEk6dC2wmOZAtepRUVU7m5+7gA3AGcClwEebXf66WSdJkiRJkjSjWScqkhyZ5Kh9z4HzgPvp16T42Wa3c4AvtQ1SkiRJkiSNhjZDP44FNqQ/YewYcFNVbUzyXeDaJGPAY8D4NMeQJEmSJEn6vlQNvzyENSo0yhYvXzXt9t2b18xRJJI0Pb+vJEH774KZ2h/IMXToWThGp6s4vP2OL3fynvZtL37OUK7bwZieVJIkSZIkaVZazfqRZDvwHeBxYG9VLUtyGvB+4EeB7cAvVdW/toxTkiRJkiSNgEH0qDi7qpZW1bJmeR1wdVX9NP2ZQN48gHNIkiRJkqQRcDCGfpwCfKZ5fgfwCwfhHJIkSZIkzQtJNx/D0jZRUcDtSbYk2Te7x/3Ayub5K4HjW55DkiRJkiSNiLaJihVVdTrwUuC3kpwJvK55vgU4CtgzVcMk40kmkkysX9trGYYkSZIkSeqCVsU0q2pn83NXkg3AGVX1HuA8gCQnAxc8Sdse0AOnJ5UkSZIkSX2zTlQkORJYUFXfaZ6fB7w9yTFN4mIB8Lv0ZwCRJEmSJKmTFgyxnkMXtRn6cSzwD0nuBe4CPl5VG4HXJPkisA3YCVzfPkxJkiRJkjQKUjX8URcO/ZCGa/HyVdNu3715zRxFIulQNtN3Cfh9Ik3Hv8fqsoVjdLrPwTs+9eVO3tO+9dznDOW6HYzpSSVJkiRJkmbFRIUkSZIkSZo3Ws36keRoYB1wKlD0pyb9AnAzcAKwHXhVVe1uFaUkSZIkSfNUuj2yZc617VFxLbCxqp4LnAZsBa4GPlVVJwGfapYlSZIkSZJmNOtERZJFwJnAeoCq2lNVjwIXATc0u90AvKxtkJIkSZIkaTS06VFxIvAt4PokdydZl+RI4Niq+gZA8/OYqRonGU8ykWRi/dpeizAkSZIkSVJXtKlRMQacDlxRVZuSXMtTGOZRVT2gB05PKkmSJEk6dC2wRMVAtelRsQPYUVWbmuVb6CcuvpnkOIDm5652IUqSJEmSpFEx60RFVT0CPJzklGbVucCDwK3Apc26S4GPtYpQkiRJkiSNjFbTkwJXADcmORx4CLicfvLjw0l+Ffga8MqW55AkSZIkSSMiVcMvD2GNCkmDsHj5qmm37968Zo4iGT5fCx2qfO+OnpmuOXjdpS5YOEanqzi86399pZP3tFed/ZNDuW5talRIkiRJkiQNVKtERZKjk9ySZFuSrUlekOSVSR5I8kSSZYMKVJIkSZIkdV/bGhXXAhur6uKmTsURwKPAK4C/aBucJEmSJEkaLbNOVCRZBJwJXAZQVXuAPfQTFSSdHoIkSZIkSRLg/e+gtRn6cSLwLeD6JHcnWZfkyANtnGQ8yUSSifVrey3CkCRJkiRJXdFm6McYcDpwRVVtSnItcDXwewfSuKp6QA+c9UOSJEmSJPW16VGxA9hRVZua5VvoJy4kSZIkSZJmZdaJiqp6BHg4ySnNqnOBBwcSlSRJkiRJGkltZ/24ArixmfHjIeDyJC8H3gc8C/h4knuq6udankeSJEmSpHlpgbU0B6pVoqKq7gGW7bd6Q/OQpDm1e/OaYYcwb/haaJQtXr6qVftBfH5mimGUPqNz8Vp05fVu+96Fufldu/J6zwVfK2l22tSokCRJkiRJGqhWiYokRye5Jcm2JFuTvCDJu5vl+5JsSHL0oIKVJEmSJEnd1rZHxbXAxqp6LnAasBW4Azi1qp4HfBFY3fIckiRJkiTNW0k3H8My60RFkkXAmcB6gKraU1WPVtXtVbW32e2zwJL2YUqSJEmSpFHQpkfFicC3gOuT3J1kXZIj99vndcAnW5xDkiRJkiSNkDaJijHgdOC6qno+8G/A1fs2JnkrsBe4carGScaTTCSZWL+21yIMSZIkSZLUFW2mJ90B7KiqTc3yLTSJiiSXAhcC51ZVTdW4qnpAD+CxvUy5jyRJkiRJ892CYRZ06KBZ96ioqkeAh5Oc0qw6F3gwyUuAtwArq+rfBxCjJEmSJEkaEW16VABcAdyY5HDgIeByYDPwdOCO9LNKn62q32h5HkmSJEmSNALyJCMz5pRDPyTNhcXLV824z+7Na1odY6b2UpcN4jM2iHMc7BikQ9kg/o61PcZcfFcMwnz4m3+ovFYHYuEYnR4bcc3f/3Mn72l/54XPHsp1a9ujQpIkSZKkkbag02mYudcqUZHkaGAdcCpQ9KcjPR+4CHgC2AVcVlU7W8YpSZIkSZJGQJvpSQGuBTZW1XOB04CtwLur6nlVtRS4DXhby3NIkiRJkqQRMeseFUkWAWcClwFU1R5gz367HQnWn5AkSZIkSQemzdCPE4FvAdcnOQ3YAlxZVf+W5B3ArwD/ApzdPkxJkiRJkjQK2gz9GANOB66rqucD/wZcDVBVb62q44EbgSlL1SYZTzKRZGL92l6LMCRJkiRJGp6km49hadOjYgewo6o2Ncu30CQqJrkJ+Djw+/s3rqoe0AOnJ5UkSZIkSX2z7lFRVY8ADyc5pVl1LvBgkpMm7bYS2NYiPkmSJEmSNELazvpxBXBjkvuApcAfA+9Mcn+z7jzgypbnkCRJkiRJ81CSlyT5QpIvJ9l/lMXk/S5OUkmWzXjMquGPunDoR9/i5VOW8/gBuzevmYNIpLl3IO//mczF52OmOP2MSu0M4rtgJoP4nPpdoPmq7WfI964OloVjDLHiwcH35/+4vZP3tL+14oRpr1uSw4AvAi+mXx5iM/Caqnpwv/2Ool8W4nBgVVVNTHfctj0qJEmSJEnSaDoD+HJVPVRVe4APARdNsd8fAe8CHjuQg5qokCRJkiRJP2TybJ3NY3y/XX4CeHjS8o5m3eRjPB84vqpuO9DztkpUJDk6yS1JtiXZmuQFk7a9qRl/8sw255AkSZIkSXOvqnpVtWzSo7ffLlMNDfn+MJgkC4D3Av/3Uzlvm+lJAa4FNlbVxUkOB45ogjme/hiVr7U8viRJkiRJ81o6XYFjWjuA4yctLwF2Tlo+CjgVuDP9F+nHgVuTrJyuTsWse1QkWQScCawHqKo9VfVos/m9wFVgkUxJkiRJkjpqM3BSkmc3nRcuAW7dt7Gq/qWqnllVJ1TVCcBngWmTFNBu6MeJwLeA65PcnWRdkiOTrAS+XlX3Ttd48liX9Wv37z0iSZIkSZLms6raC6wC/hbYCny4qh5I8vYmNzArbYZ+jAGnA1dU1aYk1wJ/QL+XxXkzNW7GtvTA6UklSZIkSToUVdUngE/st+5tT7LvWQdyzDaJih3Ajqra1CzfQj9R8Wzg3mb8yRLgc0nOqKpHWpxLkiRJkqR5acHo1qg4KGY99KNJPDyc5JRm1bnA56rqmEnjT3YAp5ukkCRJkiRJB6LtrB9XADc2RTMeAi5vH5IkSZIkSRpVqRp+eYgfef6qaYPYvXnNXIUiSZI6bvHyVdNun4t/d8wUw1zFIUlzZeEYnR4c8f5/2j78G+uD4DdecMJQrlubWT8kSZIkSZIGqtXQjyRHA+uAU4ECXgf8HPBr9KcuBfhvTRVQSZIkSZI6Z0E63WFkzrWtUXEtsLGqLm7qVBxBP1Hx3qp6T+voJEmSJEnSSJl1oiLJIuBM4DKAqtoD7ImZJEmSJEmSNEttalScSH94x/VJ7k6yLsmRzbZVSe5L8oEki6dqnGQ8yUSSib3ffqBFGJIkSZIkqSvaJCrGgNOB66rq+cC/AVcD1wE/CSwFvgH8yVSNq6pXVcuqatnYM3+qRRiSJEmSJA1P0s3HsLRJVOwAdlTVpmb5FuD0qvpmVT1eVU8Aa4Ez2gYpSZIkSZJGw6wTFVX1CPBwklOaVecCDyY5btJuLwfubxGfJEmSJEkaIW1n/bgCuLGZ8eMh4HLgz5IspT9d6Xbg11ueQ5IkSZIkjYhU1bBj4LG9DD+IQ8Ti5aum3b5785o5iqSdrvweOrTM9L4D33vSfDcXn2O/KzQs/vtIXbZwjE5PD7n+rq918p72V8/4T0O5bm1qVEiSJEmSJA1Uq0RFkqOT3JJkW5KtSV7QrL8iyReSPJDkXYMJVZIkSZIkdV3bGhXXAhur6uKmTsURSc4GLgKeV1XfS3JM6yglSZIkSdJImHWiIski4EzgMoCq2gPsSfKbwDur6nvN+l0DiFOSJEmSpHkpna7AMffaDP04EfgWcH2Su5OsS3IkcDLwwiSbkvxdkuUDiVSSJEmSJHVem0TFGHA6cF1VPR/4N+DqZv1i4L8AbwY+nPxwfinJeJKJJBPr1/ZahCFJkiRJkrqiTY2KHcCOqtrULN9CP1GxA/ho9ec9vSvJE8Az6fe++L6q6gE9cHpSSZIkSZLUN+seFVX1CPBwklOaVecCDwJ/A5wDkORk4HDg2y3jlCRJkiRJI6DtrB9XADc2M348BFxOfwjIB5LcD+wBLm16V0iSJEmS1Dltairoh2U+5BAc+iFpvli8fFWr9rs3rxlQJJLms5m+Kw6V74Ku/B4anLZ/B8H3jaa2cIxOz4vxPzZ/rZP3tJct/09DuW4mfiRJkiRJ0rzRauhHkqOBdcCpQAGvA34H2Fe34mjg0apa2uY8kiRJkiRpNLStUXEtsLGqLm7qVBxRVa/etzHJnwD/0vIckiRJkiTNW0mnR7bMuVknKpIsAs4ELgOoqj30i2fu2x7gVTQzgEiSJEmSJM2kTY2KE4FvAdcnuTvJuiRHTtr+QuCbVfWlVhFKkiRJkqSR0SZRMQacDlxXVc+nPy3p1ZO2vwb4qydrnGQ8yUSSifVrey3CkCRJkiRJXdGmRsUOYEdVbWqWb6FJVCQZA14B/MyTNa6qHtADpyeVJEmSJB26rFAxWLPuUVFVjwAPJ9k3w8e5wIPN8xcB26pqR8v4JEmSJEnSCGk768cVwI3NjB8PAZc36y9hmmEfkiRJkiRJU0nV8EddOPRD0qFi8fJV027fvXnNHEUizT8zfT5g5s/IID5jBxJH23O0jWGUvivmw2sxiPfmfNCV30OjaeFYt0dH/OXEw528p/2VZccP5bq17VEhSZIkSdJIW5BO52HmXJtZPyRJkiRJkgaqVaIiydFJbkmyLcnWJC9IsjTJZ5Pc00w/esaggpUkSZIkSd3WdujHtcDGqrq4Kah5BPBh4A+r6pNJzgfeBZzV8jySJEmSJGkEzDpRkWQRcCZwGUBV7QH2JClgUbPbM4CdLWOUJEmSJEkjos3QjxOBbwHXJ7k7ybokRwK/A7w7ycPAe4DVUzVOMt4MDZlYv7bXIgxJkiRJkoYnHX0MS5uhH2PA6cAVVbUpybXA1fR7Ubyhqj6S5FXAeuBF+zeuqh7QA6cnlSRJkiRJfW16VOwAdlTVpmb5FvqJi0uBjzbr/hqwmKYkSZIkSTogs05UVNUjwMNJTmlWnQs8SL8mxc82684BvtQqQkmSJEmSNDLazvpxBXBjM+PHQ8DlwMeAa5OMAY8B4y3PIUmSJEnSvJVhFnTooFQNvzyENSokSdIoWbx81Yz77N68Zg4ikebeXLz/D+QcbWPwc/zULBwbam3Gg+6mz+3o5D3tL56+ZCjXrU2NCkmSJEmSpIFqlahIcnSSW5JsS7I1yQuSnJbkn5J8Psn/TLJoUMFKkiRJkqRua1uj4lpgY1Vd3NSpOAK4A3hTVf1dktcBbwZ+r+V5JEmSJEmal2KRioGadY+KpqfEmcB6gKraU1WPAqcAn2l2uwP4hbZBSpIkSZKk0dBm6MeJwLeA65PcnWRdkiOB+4GVzT6vBI6fqnGS8SQTSSbWr+21CEOSJEmSJHVFm6EfY8DpwBVVtSnJtcDVwOuAP0vyNuBWYM9UjauqB/TAWT8kSZIkSVJfm0TFDmBHVW1qlm8Brq6q3wPOA0hyMnBBuxAlSZIkSZq/nE5zsGb9elbVI8DDSU5pVp0LPJjkGIAkC4DfBd7fOkpJkiRJkjQS2iZ+rgBuTHJAzjBZAAAgAElEQVQfsBT4Y+A1Sb4IbAN2Ate3PIckSZIkSRoRqRp+eQhrVMwvi5evan2M3ZvXDCASSZJG10x/j/1bq9nw33ndc6h8Vywco9Pzd95899c7eU/76uf/xFCum0NpJEmSJEnSvDHrYppNbYqbJ606EXgb8JfN+hOA7cCrqmr37EOUJEmSJGn+SjrdYWTOtSmm+YWqWlpVS4GfAf4d2EB/itJPVdVJwKeaZUmSJEmSpBkNaujHucBXquqrwEXADc36G4CXDegckiRJkiSp4waVqLgE+Kvm+bFV9Q2A5ucxAzqHJEmSJEnquNaJiiSHAyuBv36K7caTTCSZWL+21zYMSZIkSZKGIh19DMusi2lO8lLgc1X1zWb5m0mOq6pvJDkO2DVVo6rqAT1welJJkiRJktQ3iKEfr+H/DPsAuBW4tHl+KfCxAZxDkiRJkiSNgFaJiiRHAC8GPjpp9TuBFyf5UrPtnW3OIUmSJEmSRkeqhj/qwqEfkiTpQCxevmrGfXZvXjMHkQzfTK/FqLwOGk0H8l0wEz8jc2vh2FBLHhx0t9z7jU7e01582nFDuW6DmvVDkiRJkiSptVknKpKckuSeSY9/TfI7SV6Z5IEkTyRZNshgJUmSJElSt8161o+q+gKwFCDJYcDXgQ3AEcArgL8YRICSJEmSJGl0DGJ6UoBzga9U1Vf3rUg6PQRJkiRJkiTAmgqDNqjX8xJ+cIpSSZIkSZKkp6x1oiLJ4cBK4K+fYrvxJBNJJtav7bUNQ5IkSZIkdcAghn68FPhcVX3zqTSqqh7QA6cnlSRJkiRJfYMY+vEaHPYhSZIkSZIGoFWPiiRHAC8Gfn3SupcD7wOeBXw8yT1V9XOtopQkSZIkaZ5yMonBStXwR1049EPSoWLx8lXTbt+9ec0cRaJDyai8b2b6PaE7v+uhYlTee1KXtf0cz5fv5oVjdPpOfsN9j3Tynvblz/vxoVw3Z1GRJEnqoAO5OZE0v83F59iEpeYjExWSJEmSJGnemHWNiiSnADdPWnUi8DbgJ4CfB/YAXwEur6pH2wQpSZIkSdJ81elxLUMw6x4VVfWFqlpaVUuBnwH+HdgA3AGcWlXPA74IrB5IpJIkSZIkqfMGNfTjXOArVfXVqrq9qvY26z8LLBnQOSRJkiRJUscNKlFxCfBXU6x/Hfz/7N19uJ31Xef796fsYkjtw24wlfJwSpUiFQtWwqHMlDIFHGA6UKiMYKscW4l6iDycy9r2dE456uiBitNxzFHPSjctc6yxPNpqH2wuHKlzRjDbFOiOgaYdSgykhMouWIE2Kd/zx7pjtzFkp+te2WvnXu/Xvu5rrfW7n7573Wvv7PuX3+/75VN72iHJyiTTSaan1vSGFIYkSZIkSTqQDZyjYpckBwPnsdsUjyTvBXYCH9nTflXVA3pgeVJJkiRJ0oErJqkYqtYdFcA5wIaqenRXQ5JLgTcBZ1SVnRCSJEmSJGmfDKOj4hLmTPtIcjbwLuANVfXUEI4vSZIkSZLGRKscFUmWAmcBt81pXg28EFiX5J4kv9fmHJIkSZIkaXxkMczMMEeFJEnSwptcsWqv62fXr16gSKTFx5+P4VoyQaezOPzx5x/t5D3tv/2hl43kug2r6ockSZIkSVJrA+eoSHIs8NE5Ta8E3gcsA84HngW2A/9LVT3SJkhJkiRJkjQeBu6oqKoHgBMBkhwEPAzcDsxW1f/RtF9Bv/Pi59qHKkmSJEmSum4YVT8AzgC+VFUP7db+AjD/hCRJkiRJ2jfDylFxMf+0ROmvJflb4K30R1T8M0lWJplOMj21pjekMCRJkiRJWlhJN5dRaT2iIsnBwHnAe3a1VdV7gfcmeQ+wCrhm9/2qqgf0wKofkiRJkiSpbxgjKs4BNlTVo3tY9wfAW4ZwDkmSJEmSNAaG0VFxCf902scxc9adB9w/hHNIkiRJkqQx0GrqR5KlwFnAz85pvrYpXfos8BBW/JAkSZIkdVgYYUKHDkrV6NNDmKNC2r8mV6za6/rZ9asXKBJp4Y3L539cvs9h8L3ad/O9V+D7JY3agfI7bclEt+/kPzGzvZP3tP/m+OUjuW7DqvohSZKkMbNYboAkSd0ycEdFkmOT3DNneTLJVXPW/2KSSnLocEKVJEmSJEldN3COiqp6ADgRIMlBwMPA7c3rI+nnrtgyhBglSZIkSVq00umJLQtvWFM/zgC+VFUPNa8/APwSmHtCkiRJkiTtu2F1VFxMU6I0yXnAw1V175COLUmSJEmSxkTrjookBwPnATc35UrfC7xvH/ZbmWQ6yfTUml7bMCRJkiRJUgcMnKNijnOADVX1aJIfAo4G7k1/ks4RwIYkJ1fVV+buVFU9oAeWJ5UkSZIkHbie1+3qqwtuGB0Vl9BM+6iqzwPLd61I8mXgpKr66hDOI0mSJEmSOq7V1I9mqsdZwG3DCUeSJEmSJI2zVI1+1oVTPyRJozK5YtVe18+uX71AkUgHJn+GJO2LJRPdnhvx6Y2PdfKe9uwf/J6RXLdhVf2QJEnSmJmvk0KSpEEMnKMiybHAR+c0vZJ+tY+XAJcBjzXt/3tVfXLgCCVJkiRJWsTS6fEiC2/gjoqqegA4ESDJQcDDwO3ATwMfqKrrhxKhJEmSJEkaG8Oa+nEG8KWqemhIx5MkSZIkSWNoWB0VF9OUKG2sSnJfkhuSTA7pHJIkSZIkqeNad1QkORg4D7i5afpd4PvoTwvZBvzmc+y3Msl0kumpNb22YUiSJEmSNBJJN5dRGThHxRznABuq6lGAXY8ASdYAf7KnnaqqB/TA8qSSJEmSJKlvGFM/LmHOtI8kh81ZdwEwM4RzSJIkSZKkMdBqREWSpcBZwM/OaX5/khOBAr682zpJkiRJkqTnlKrRz7pw6ockSVI3Ta5Ytdf1s+tXL8gx2loMMUgHsiUTjDDjwf63btNXO3lPe9Zxh47kug2r6ockSZIkSVJrdlRIkiRJkqRFY+COiiTHJrlnzvJkkquadb+Q5IEkG5O8f3jhSpIkSZKkLhs4mWZVPQCcCJDkIOBh4PYk/wo4H3hNVX0jyfKhRCpJkiRJ0iL0vE5n4Fh4w5r6cQbwpap6CPh54Nqq+gZAVW0f0jkkSZIkSVLHDauj4mJgbfP8VcDrk9yd5M4kK/a0Q5KVSaaTTE+t6Q0pDEmSJEmSdCAbeOrHLkkOBs4D3jPnmJPAKcAK4KYkr6zd6qBWVQ/ogeVJJUmSJElS3zBGVJwDbKiqR5vXW4Hbqu+vgGeBQ4dwHkmSJEmS1HGtR1QAl/DtaR8AfwS8EfjzJK8CDga+OoTzSJIkSZK06ASzaQ5TqxEVSZYCZwG3zWm+AXhlkhngD4FLd5/2IUmSJEmStCdZDH0I5qiQJEkaT5MrVs27zez61a2OMd/+0p74uRquJRPdHnLwZ/f/XSfvad/4A8tGct2GVfVDkiRJkiSptYFzVCQ5FvjonKZXAu8DXgcc27S9BPhaVZ04cISSJEmSJC1i6fR4kYU3cEdFVT0AnAiQ5CDgYeD2qvpPu7ZJ8pvAE22DlCRJkiRJ42EYVT8AzgC+VFUP7WpIEuDf0a8AIkmSJEmSNK9h5ai4mH9aohTg9cCjVbV5TzskWZlkOsn01JrekMKQJEmSJEkHstYjKpIcDJwHvGe3VZfwzzsv/lFV9YAeWPVDkiRJknTgSreLmiy4YUz9OAfYUFWP7mpIMgFcCPzIEI4vSZIkSZLGxDCmfuxp5MSZwP1VtXUIx5ckSZIkSWOiVUdFkqXAWcBtu63aU84KSZIkSZKkvUrV6NNDmKNCkiRJz2Vyxaq9rp9dv3qBIhmt+d4HGJ/3QgeeJRPdTuLw2S883sl72tNe9dKRXLdhVf2QJEmSJElqbeCOiiTHJrlnzvJkkquSnJjkrqZtOsnJwwxYkiRJkiR118BVP6rqAeBEgCQHAQ8DtwNrgF+uqk8lORd4P3B6+1AlSZIkSVLXDWvqxxnAl6rqIaCAFzXtLwYeGdI5JEmSJElSxw08omI3c6t8XAX8aZLr6XeEnDqkc0iSJEmStOik27lCF1zrERVJDgbOA25umn4euLqqjgSuBqaeY7+VTQ6L6ak1vbZhSJIkSZKkDhjGiIpzgA1V9Wjz+lLgyub5zcAH97RTVfWAHlieVJIkSZIk9Q0jR8UlfHvaB/RzUryhef5GYPMQziFJkiRJksZAqxEVSZYCZwE/O6f5MuC3kkwAzwAr25xDkiRJkqTFLKaoGKpUjX7WhVM/JEnSQplcsWqv62fXr16gSDQsXlNp8Vsy0e1sk/9t82wn72n/5TGTI7luwypPKkmSJEmS1NrAHRVJjk1yz5zlySRXJTkhyV8m+XySP07yomEGLEmSJEmSumvgjoqqeqCqTqyqE4EfAZ4Cbqdf5ePdVfVDzet3DiVSSZIkSZIWoXR0GZVhTf04A/hSVT0EHAt8tmlfB7xlSOeQJEmSJEkdN6yOiov5donSGeC85vlFwJFDOockSZIkSeq41h0VSQ6m3zFxc9P0duDyJH8NvBD45nPstzLJdJLpqTW9tmFIkiRJkqQOmBjCMc4BNlTVowBVdT/wowBJXgX8mz3tVFU9oAeWJ5UkSZIkHbiel05XX11ww5j6cQnfnvZBkuXN4/OAfw/83hDOIUmSJEmSxkCrjookS4GzgNvmNF+S5AvA/cAjwIfanEOSJEmSJI2PVlM/quopYNlubb8F/Fab40qSJEmSpPGUqtGnh2ibo2Jyxap5t5ldv7rNKSRpwcz3O83fZ5L0nfFvRWn0lkzQ6SQOd33xa6O/sd4PTvn+l4zkug2rPKkkSZIkSVJrbXNUXJ1kY5KZJGuTLElydJK7k2xO8tGmfKkkSZIkSdK8Bu6oSHI4cAVwUlUdDxwEXAxcB3ygqo4BZoF3DCNQSZIkSZLUfW2nfkwAhySZAJYC24A3Arc0628E3tzyHJIkSZIkaUwM3FFRVQ8D1wNb6HdQPAH8NfC1qtrZbLYVOHxP+ydZmWQ6yfTUmt6gYUiSJEmSNFrp6DIiA5cnTTIJnA8cDXwNuBk4Zw+b7jH7aVX1gB60r/ohSZIkSZK6oc3UjzOBB6vqsaraAdwGnAq8pJkKAnAE8EjLGCVJkiRJ0pho01GxBTglydIkAc4A/gb4r8CPNdtcCnysXYiSJEmSJGlcDDz1o6ruTnILsAHYCXyO/lSOTwB/mOQ/NG1TwwhUkiRJkqTFKKNM6NBBqRp9eghzVEiSJI2nyRWr5t1mdv3qkcexEDFo/IzT527JRLfv5O/+0hOdvKf9n7/vxSO5bm3Lk0qSJEmSJA1Nq46KJFcn2ZhkJsnaJEuSrEryxSSV5NBhBSpJkiRJkrqvTXnSw4ErgFdX1dNJbgIuBv4/4E+APx9KhJIkSZIkLWLp9MSWhTdwR8Wc/Q9JsgNYCjxSVZ8DiFdKkiRJkiR9hwae+lFVDwPX0y9Tug14oqo+s6/7J1mZZDrJ9NSa3qBhSJIkSZKkDmkz9WMSOB84GvgacHOSt1XV7+/L/lXVo1/O1KofkiRJkiQJaJdM80zgwap6rKp2ALcBpw4nLEmSJEmSDgzp6DIqbToqtgCnJFmafkKKM4BNwwlLkiRJkiSNozY5Ku4GbgE2AJ9vjtVLckWSrcARwH1JPjiUSCVJkiRJUuelavTpIcxRIUmS1E2TK1btdf3s+tULFEk7Xfk+pFFZMjHSmQT73fr/8UQn72lXvPLFI7lubaZ+SJIkSZIkDVWrjookVyfZmGQmydokS5J8JMkDTdsNSZ4/rGAlSZIkSVp0Rp31smPZNAfuqEhyOHAFcFJVHQ8cBFwMfAT4AeCHgEOAnxlCnJIkSZIkaQxMDGH/Q5LsAJYCj1TVZ3atTPJX9JNqSpIkSZIkzatN1Y+HgevplyndBjyxWyfF84GfBD7dNkhJkiRJkjQe2kz9mATOB44GXg68IMnb5mzyO8Bnq+ovnmP/lUmmk0xPrekNGoYkSZIkSSOVjn6NSpupH2cCD1bVYwBJbgNOBX4/yTXA9wA/+1w7V1UP6IHlSSVJkiRJUl+bjootwClJlgJPA2cA00l+BvjXwBlV9ewQYpQkSZIkSWNi4I6Kqro7yS3ABmAn8Dn6IyT+AXgI+MskALdV1a8MIVZJkiRJktRxqRr9rAunfkiSJOlANrli1bzbzK5fvQCRSIvTkokRJjxYAH/95Sc7eU/7I6940Uiu28DJNCVJkiRJkoatVUdFkquTbEwyk2RtkiVJppLcm+S+JLck+e5hBStJkiRJkrqtTXnSw4ErgJOq6njgIOBi4OqqOqGqXkM/4eb84+AkSZIkSZJoV/Vj1/6HJNkBLAUeqaonAdLPpHkImH9CkiRJktRdnU7AMQIDj6ioqoeB6+mPmtgGPFFVnwFI8iHgK8APAL89hDglSZIkSdIYaDP1YxI4HzgaeDnwgiRvA6iqn27aNgE//hz7r0wynWR6ak1v0DAkSZIkSVKHtJn6cSbwYFU9BpDkNuBU4PcBqupbST4KvBP40O47V1UP6IHlSSVJkiRJUl+bqh9bgFOSLG3yUZwBbEry/fCPOSr+LXB/+zAlSZIkSdI4GHhERVXdneQWYAOwE/gc/RESf5bkRfTzidwL/PwwApUkSZIkaVEym+ZQtar6UVXXANfs1vwv2hxTkiRJkiSNr7blSSVJkqSxN7t+9bzbTK5Y1foYkjQO2uSokCRJkiRJGqpWHRVJrk6yMclMkrVJlsxZ99tJvt4+REmSJEmSFq909GtUBu6oSHI4cAVwUlUdDxwEXNysOwl4yVAilCRJkiRJY6Pt1I8J4JAkE8BS4JEkBwG/AfxS2+AkSZIkSdJ4GbijoqoeBq4HtgDbgCeq6jPAKuDjVbVtb/snWZlkOsn01JreoGFIkiRJkqQOGbjqR5JJ4HzgaOBrwM1Jfgq4CDh9vv2rqgf0AJ7ZSQ0ahyRJkiRJo5TRpXPopDblSc8EHqyqxwCS3Ab8MnAI8MX0r9TSJF+squ9vHakkSZIkSeq8NjkqtgCnJFmafq/EGcB/rKrvrapXVNUrgKfspJAkSZIkSfuqTY6Ku4FbgA3A55tjmWxCkiRJkiQNLFWjTw9hjgpJ0v4yuWLVXtfPrl99QJxjXAzjvfR67Dvfq8XF31fqsiUTdDqLw71b/r6T97QnHPXCea9bkrOB3wIOAj5YVdfutv5/A34G2Ak8Bry9qh7a2zHblieVJEmSJEljKMlBwP8NnAO8Grgkyat32+xzwElV9Rr6szLeP99xW3VUJLk6ycYkM0nWJlmS5MNJHkxyT7Oc2OYckiRJkiRpUToZ+GJV/Y+q+ibwh/Srg/6jqvqvVfVU8/Iu4Ij5DjpwR0WSw4Er6PeMHE9/mMfFzep3VtWJzXLPoOeQJEmSJEmjkWRlkuk5y8rdNjkc+Ns5r7c2bc/lHcCn5jtvm/Kku/Y/JMkOYCnwSMvjSZIkSZKkRaCqeuy9aMaecljsMV9HkrcBJwFvmO+8bap+PAxcT79M6Tbgiar6TLP615Lcl+QDSb7rOYL8x56ZqTUWC5EkSZIkHaDS0WV+W4Ej57w+gj0MYEhyJvBe4Lyq+sZ8Bx14REWSSfpzT44Gvgbc3PSQvAf4CnAw/Z6XdwG/svv+c3tmrPohSZIkSdIBZz1wTJKjgYfpp4P4ibkbJPlh4P8Bzq6q7fty0DbJNM8EHqyqx6pqB3AbcGpVbau+bwAfop9cQ5IkSZIkdUhV7QRWAX8KbAJuqqqNSX4lyXnNZr8BfDf9wQ33JPn4fMdtk6NiC3BKkqXA08AZwHSSw6pqW5IAbwZmWpxDkiRJkiQtUlX1SeCTu7W9b87zM7/TYw7cUVFVdye5BdgA7KRfG7UHfCrJ99Cf0XIP8HODnkOSJEmSpMUu+5jQQfumVdWPqroGuGa35je2OabUJZMrVs27zez61QsQidRNi+VnzJ/j4ZnvvVws11wLa77r3pVrfqB8/sflenSF10sHojY5KiRJkiRJkoaqVUdFkquTbEwyk2RtkiXp+7UkX0iyKckVwwpWkiRJkiR1W5vypIcDVwCvrqqnk9xEvxRJ6NdR/YGqejbJ8uGEKkmSJEnS4hNTVAxVqxwVzf6HJNkBLAUeAf4D8BNV9SzAvtZJlSRJkiRJGnjqR1U9DFxPv0zpNuCJqvoM8H3AjyeZTvKpJMcMJ1RJkiRJktR1A3dUJJkEzgeOBl4OvCDJ24DvAp6pqpOANcANz7H/yqYzY3pqTW/QMCRJkiRJUoe0mfpxJvBgVT0GkOQ24FRgK3Brs83twIf2tHNV9YAewDM7qRZxSJIkSZI0MqaoGK42VT+2AKckWZokwBnAJuCPgDc227wB+EK7ECVJkiRJ0rgYeERFVd2d5BZgA7AT+Bz9ERKHAB9JcjXwdeBnhhGoJEmSJEnqvlSNftaFUz8kSdJiMbli1bzbzK5fvQCRSN+5+T6/fnYXl3G6Xksmuj07Ymbr1zt5T3v8Ed89kuvWZuqHJEmSJEnSULVJpilJkiRJkjo9XmThtRpRkeTqJBuTzCRZm2RJkr9Ick+zPJLkj4YVrCRJkiRJ6raBR1QkORy4Anh1VT2d5Cbg4qp6/ZxtbgU+1j5MSZIkSZI0DtrmqJgADkkyASwFHtm1IskL6ZcpdUSFJEmSJEnaJwN3VFTVw8D1wBZgG/BEVX1mziYXAHdU1ZN72j/JyiTTSaan1vQGDUOSJEmSpJFKR79Gpc3Uj0ngfOBo4GvAzUneVlW/32xyCfDB59q/qnpADyxPKkmSJEmS+tpM/TgTeLCqHquqHcBtwKkASZYBJwOfaB+iJEmSJEkaF206KrYApyRZmiTAGcCmZt1FwJ9U1TNtA5QkSZIkSeNj4KkfVXV3kluADcBO4HM0UzmAi4Fr24cnSZIkSdLiltGlc+ikVI0+PcR8OSomV6za6/6z61cPNR4dGPxc6Ds132cG5v/c+LmTDnzD+Dlue4xh/D46UPh7c3HZl8/efMblmvnZHa4lEyPMzLgA/uaRfxj9jfV+8OqXv2Ak161teVJJkiRJkqShadVRkeTqJBuTzCRZm2RJkjOSbEhyT5L/luT7hxWsJEmSJEnqtoE7KpIcDlwBnFRVxwMH0c9N8bvAW6vqROAPgH8/jEAlSZIkSVqM0tFlVNpO/ZgADkkyASwFHgEKeFGz/sVNmyRJkiRJ0rwG7qioqoeB6+mXKd0GPFFVnwF+Bvhkkq3AT/Ic1T+SrEwynWR6ak1vT5tIkiRJkqQxM3B50iSTwPnA0cDXgJuTvA24EDi3KV/6TuA/0u+8+CeqqkdTznS+qh+SJEmSJGk8tJn6cSbwYFU9VlU7gNuAfwGcUFV3N9t8FDi1ZYySJEmSJGlMtOmo2AKckmRpkgBnAH8DvDjJq5ptzgI2tYxRkiRJkqTFa9RZLzuWTXPgqR/N1I5bgA3ATuBz9KdybAVuTfIsMAu8fRiBSpIkSZKk7kvV6NNDHPLDq/YaxOz61QsViiRJ0tiYXLFqr+v9G2xhLYbrMV8MiyWOcYmhS5ZMjLTa5X63ads/jP7Gej847rAXjOS6tS1PKkmSJEmSNDStOiqSXJ1kY5KZJGuTLEnyxiQbmrYbkww8vUSSJEmSpMUuHf0alYE7KpIcDlwBnFRVxwMHAT8B3Ahc3LQ9BFw6jEAlSZIkSVL3tZ36MQEc0oyaWAr8A/CNqvpCs34d8JaW55AkSZIkSWNi4I6KqnoYuJ5+mdJtwBPATcDzk5zUbPZjwJF72j/JyiTTSaZ3fnXjoGFIkiRJkqQOaTP1YxI4HzgaeDnwAuCtwMXAB5L8FfD39EuX/jNV1auqk6rqpIlDf3DQMCRJkiRJGqmkm8uotEl0eSbwYFU9BpDkNuDUqvp94PVN248Cr2odpSRJkiRJGgttclRsAU5JsjRJgDOATUmWAyT5LuBdwO+1D1OSJEmSJI2DNjkq7gZuATYAn2+O1QPemWQTcB/wx1X1Z8MIVJIkSZIkdV+qatQx8MxORh+Ehmpyxaq9rp9dv3qBIpHGkz+DGmdtP//z7b8vxxiGxfB9LJb3oq2ufB+LxUJ8NuczjOvVlX8rD5TvY8kEI8x4sP994StPdfKe9lXfu3Qk161teVJJkiSNqcVyAyRJ6pZWHRVJrkwyk2RjkquatpcmWZdkc/M4OZxQJUmSJElS17UpT3o8cBlwMnAC8KYkxwDvBu6oqmOAO5rXkiRJkiRJ82ozouI44K6qeqqqdgJ3AhcA5wM3NtvcCLy5XYiSJEmSJGlctOmomAFOS7IsyVLgXOBI4GVVtQ2geVzePkxJkiRJkhapdHQZkTblSTcB1wHrgE8D9wI793X/JCuTTCeZnlrTGzQMSZIkSZLUIRNtdq6qKWAKIMmvA1uBR5McVlXbkhwGbH+OfXtADyxPKkmSJEmS+tpW/VjePB4FXAisBT4OXNpscinwsTbnkCRJkiRJ46PViArg1iTLgB3A5VU1m+Ra4KYk7wC2ABe1DVKSJEmSpMUqo0zo0EGpGv2sC6d+aLGaXLFqr+tn169eoEgkSZIODP79NDzzvZdw4LyfSya6fSe/+dGnO3lPe8zLDhnJdWs19UOSJEmSJGmY7KiQJEmSJEmLRttkmlcmmUmyMclVTdtFzetnk5w0nDAlSZIkSVqckm4uozJwR0WS44HLgJOBE4A3JTkGmKFfAeSzQ4lQkiRJkiSNjTYjKo4D7qqqp6pqJ3AncEFVbaqqB4YTniRJkiRJGidtOipmgNOSLEuyFDgXOHJfd06yMsl0kumpNb0WYUiSJEmSpK6YGHTHqtqU5DpgHfB14F5g53ewfw/ogeVJJUmSJEkHrk7XXh2BVsk0q2qqql5bVacBjwObhxOWJEmSJEkaRwOPqABIsryqtic5in4CzdcNJyxJkiRJkjSOWnVUALcmWQbsAC6vqtkkFwC/DXwP8Ikk91TVv24bqCRJkiRJ6r5UjT49hDkqNM4mV6za6/rZ9fgqqggAACAASURBVKsXKBJJ+8ti+DmfL4aFiqOtrnwf0jg7UH6OF8Pv7i5ZMtHtNA5f2v50J+9pv2/5ISO5bm1HVEiSJEmSNN463Q2z8Fol00xyZZKZJBuTXNW0/UaS+5Pcl+T2JC8ZTqiSJEmSJKnrBu6oSHI8cBlwMnAC8KYkx9AvV3p8Vb0G+ALwnmEEKkmSJEmSuq/NiIrjgLuq6qmq2gncCVxQVZ9pXgPcBRzRNkhJkiRJkjQe2nRUzACnJVmWZClwLnDkbtu8HfjUnnZOsjLJdJLpqTW9FmFIkiRJkjQ66ejXqAycTLOqNiW5jv5Uj68D9wK7RlKQ5L3N6488x/49oAdW/ZAkSZIkSX2tkmlW1VRVvbaqTgMeBzYDJLkUeBPw1loM9U8lSZIkSdIBoVV50iTLq2p7kqOAC4HXJTkbeBfwhqp6ahhBSpIkSZKk8dCqowK4NckyYAdweVXNJlkNfBewLgn0E27+XMvzSJIkSZK0KGV06Rw6KYthZsYhP7xqr0HMrl+9UKEc8CZXrNrret9LSZKk8bUY/lacL4Z94d+0B54lEyPMzLgAHvzqM6O/sd4Pjj50yUiuW6scFZIkSZIkScPUqqMiyZVJZpJsTHJV0/arSe5Lck+SzyR5+XBClSRJkiRJXTdwR0WS44HLgJOBE4A3JTkG+I2qek1VnQj8CfC+oUQqSZIkSdIilI4uo9JmRMVx9BNlPlVVO4E7gQuq6sk527wA6ORcHUmSJEmSNHxtOipmgNOSLEuyFDgXOBIgya8l+VvgrTzHiIokK5NMJ5ne+dWNLcKQJEmSJEldMXBHRVVtAq4D1gGfBu4Fdjbr3ltVRwIfAfaY1reqelV1UlWdNHHoDw4ahiRJkiRJ6pBWyTSraqqqXltVpwGPA5t32+QPgLe0OYckSZIkSRofE212TrK8qrYnOQq4EHhdkmOqaleHxXnA/W2DlCRJkiRp0Rpl5skOatVRAdyaZBmwA7i8qmaTfDDJscCzwEPAz7UNUpIkSZIkjYdUjb4oxzM7rQwiSZIkad9MrthjGrzvyOz61UOIRPtqyUS3xxx8+e+e6eQ97SuWLRnJdWuVo0KSJEmSJGmYWnVUJLkyyUySjUmu2m3dLyapJIe2C1GSJEmSpMUrHf0alYE7KpIcD1wGnAycALwpyTHNuiOBs4AtwwhSkiRJkiSNhzYjKo4D7qqqp6pqJ3AncEGz7gPAL4G5JyRJkiRJ0r5r01ExA5yWZFmSpcC5wJFJzgMerqp7hxKhJEmSJEkaGwN3VFTVJuA6YB3waeBeYCfwXuB98+2fZGWS6STTU2t6g4YhSZIkSdJIJd1cRmWizc5VNQVMAST5deBR4K3Avel/V0cAG5KcXFVf2W3fHtADy5NKkiRJkqS+tlU/ljePRwEXAv+lqpZX1Suq6hXAVuC1u3dSSJIkSZIk7UmrERXArUmWATuAy6tqdggxSZIkSZKkMdV26sfr51n/ijbHlyRJkqTdza5fvdf1kytWLVAkUt8I0zl0UqupH5IkSZIkScNkR4UkSZIkSVo02ibTvDLJTJKNSa5q2v7PJA8nuadZzh1OqJIkSZIkqesGzlGR5HjgMuBk4JvAp5N8oln9gaq6fgjxSZIkSZKkMdImmeZxwF1V9RRAkjuBC4YSlSRJkiRJB4iYTXOo2kz9mAFOS7IsyVLgXODIZt2qJPcluSHJ5J52TrIyyXSS6ak1vRZhSJIkSZKkrhh4REVVbUpyHbAO+DpwL7AT+F3gV4FqHn8TePse9u8BPYBndlKDxiFJkiRJkrqjVTLNqpqqqtdW1WnA48Dmqnq0qr5VVc8Ca+jnsJAkSZIkSZpXmxwVJFleVduTHAVcCLwuyWFVta3Z5AL6U0QkSZIkSeook1QMU6uOCuDWJMuAHcDlVTWb5P9NciL9qR9fBn625TkkSZIkSdKYSNXo00OYo0KSJEldN7li1V7Xz65fvUCRCLweC23JRLeHHGyd/WYn72mPmDx4JNetVY4KSZIkSZKkYWrVUZHkyiQzSTYmuWpO+y8keaBpf3/7MCVJkiRJWpySbi6jMnCOiiTHA5fRr+rxTeDTST4BHAGcD7ymqr6RZPlQIpUkSZIkSZ3XJpnmccBdVfUUQJI76Vf5OAm4tqq+AVBV21tHKUmSJEmSxkKbqR8zwGlJliVZCpwLHAm8Cnh9kruT3JlkxZ52TrIyyXSS6ak1vRZhSJIkSZKkrhh4REVVbUpyHbAO+DpwL7CzOeYkcAqwArgpyStrt/IiVdUDemDVD0mSJEnSgavTJU1GoFUyzaqaqqrXVtVpwOPAZmArcFv1/RXwLHBo+1AlSZIkSVLXtclRQZLlVbU9yVHAhcDr6HdMvBH48ySvAg4Gvto6UkmSJEmS1HmtOiqAW5MsA3YAl1fVbJIbgBuSzNCvBnLp7tM+JEmSJEmS9iSLoQ/BHBU6UE2uWLXX9bPrVy9QJJJGaTH8LlgMMSyE+b5P6M73uhj4fi8uw7ge+3KMtufoinH5vbpQlkx0O43DI1/7ZifvaV/+koNHct3ajqiQJEnSmPJGTZL60ulumIXXKplmkiuTzCTZmOSqpu2jSe5pli8nuWc4oUqSJEmSpK4beERFkuOBy4CT6eei+HSST1TVj8/Z5jeBJ1pHKUmSJEmSxkKbERXHAXdV1VNVtRO4E7hg18okAf4dsLZdiJIkSZIkaVy06aiYAU5LsizJUuBc4Mg5618PPFpVm/e0c5KVSaaTTE+t6bUIQ5IkSZKk0UlHv0Zl4KkfVbUpyXXAOuDrwL3AzjmbXMJeRlNUVQ/ogVU/JEmSJElSX6tkmlU1VVWvrarTgMeBzQBJJoALgY+2D1GSJEmSJI2LVuVJkyyvqu1JjqLfMfG6ZtWZwP1VtbVtgJIkSZIkaXy06qgAbk2yDNgBXF5Vs037xZhEU5IkSZI0DkaXzqGTUjX69BDmqNh3kytW7XX97PrVCxSJJElSe/P9bQP+fTNM/i05PH52vzNLJrp9K/+VJ3d08p72e1/0/JFct1Y5KiRJkiRJkoapVUdFkiuTzCTZmOSqpu3EJHcluacpP3rycEKVJEmSJEldN3COiiTHA5cBJwPfBD6d5BPA+4FfrqpPJTm3eX36EGKVJEmSJGnR6fS8lhFok0zzOOCuqnoKIMmdwAVAAS9qtnkx8EirCCVJkiRJ0tho01ExA/xaU/XjaeBcYBq4CvjTJNfTn1pyausoJUmSJEnSWBg4R0VVbQKuA9YBnwbuBXYCPw9cXVVHAlcDU3vaP8nKJofF9NSa3qBhSJIkSZKkDmkzooKqmqLpiEjy68BW4P8Crmw2uRn44HPs2wN6YHlSSZIkSZLU17bqx/Lm8SjgQmAt/ZwUb2g2eSOwuc05JEmSJElazJJuLqPSakQFcGuTo2IHcHlVzSa5DPitJBPAM8DKtkFKkiRJkqTxkKrRz7pw6sfCmlyxaq/rZ9evXqBIJO2JP6P7br73Cny/JHWHv/MOPP6b/m1LJrpdwXP73+/o5D3t8hc+fyTXrdXUDx149uUfOEmSJEmSRqXt1A9JkiRJksZauj1gZMG1TaZ5ZZKZJBuTXNW0nZDkL5N8PskfJ3nRcEKVJEmSJEldN3BHRZLjgcuAk4ETgDclOYZ+OdJ3V9UPAbcD7xxGoJIkSZIkqfvajKg4Drirqp6qqp3AncAFwLHAZ5tt1gFvaReiJEmSJEkaF206KmaA05IsS7IUOBc4smk/r9nmoqbtn0myMsl0kumpNb0WYUiSJEmSNELp6DIiAyfTrKpNSa6jP2ri68C9wE7g7cB/TvI+4OPAN59j/x7QA8uTSpIkSZKkvlbJNKtqqqpeW1WnAY8Dm6vq/qr60ar6EWAt8KVhBCpJkiRJkrqvbdWP5c3jUcCFwNo5bc8D/j3we22DlCRJkiRJ42HgqR+NW5MsA3YAl1fVbFOy9PJm/W3Ah1qeQ5IkSZKkRWuE6Rw6KVWjTw9hjgpJkiSN0uSKVXtdP7t+9QJFonEyTp+7JRPdvpf/6td3dvKe9tDvnhjJdWs19UOSJEmSJGmY5u2oSHJDku1JZua0vTTJuiSbm8fJpj1J/nOSLya5L8lr92fwkiRJkiSpW/ZlRMWHgbN3a3s3cEdVHQPc0bwGOAc4pllWAr87nDAlSZIkSdI4mLejoqo+S7/06FznAzc2z28E3jyn/b9U313AS5IcNqxgJUmSJElabJJuLqMyaI6Kl1XVNoDmcXnTfjjwt3O229q0/TNJViaZTjI9taY3YBiSJEmSJKlL2pYn3d2e+lz2mP20qnpAD6z6IUmSJEmS+gYdUfHorikdzeP2pn0rcOSc7Y4AHhk8PEmSJEmSNE4G7aj4OHBp8/xS4GNz2n+qqf5xCvDErikikiRJkiR1UTr6NSrzTv1IshY4HTg0yVbgGuBa4KYk7wC2ABc1m38SOBf4IvAU8NP7IWZJkiRJktRRqRp9eghzVEiSJC28yRWr9rp+dv3qBYpE0p7M9zMKB87P6ZKJEf73/AJ4/B++1cl72pe+4KCRXLdBp35IkiRJkiQN3bwdFUluSLI9ycyctpcmWZdkc/M42bT/QJK/TPKNJL+4PwOXJEmSJGkxSLq5jMq+jKj4MHD2bm3vBu6oqmOAO5rXAI8DVwDXDytASZIkSZI0PubtqKiqz9LvgJjrfODG5vmNwJubbbdX1XpgxzCDlCRJkiRJ42HQHBUv21V2tHlcPryQJEmSJEnSuBpZMs0kK5NMJ5meWtMbVRiSJEmSJGkRmRhwv0eTHFZV25IcBmz/Tg9QVT2gB5YnlSRJkiRJfYOOqPg4cGnz/FLgY8MJR5IkSZIkjbN5R1QkWQucDhyaZCtwDXAtcFOSdwBbgIuabb8XmAZeBDyb5Crg1VX15P4JX5IkSZIkdUmqRj/rwqkfkiRpnEyuWDXvNrPrV7c6xnz7azz5uemeA+WaLpkgo45hf5p96ludvKedXHrQSK7boDkqJEmSJEkSkE53wyy8eXNUJLkhyfYkM3PaXppkXZLNzeNk0/7WJPc1y39PcsL+DF6SJEmSJHXLviTT/DBw9m5t7wbuqKpjgDua1wAPAm+oqtcAv0pT1UOSJEmSJGlfzNtRUVWfBR7frfl84Mbm+Y3Am5tt/3tVzTbtdwFHDClOSZIkSZI0BgbNUfGyqtoGUFXbkizfwzbvAD41cGSSJEmSJB0A0u1coQtuX6Z+fMeS/Cv6HRXv2ss2K5NMJ5meWuMMEUmSJEmSNPiIikeTHNaMpjgM2L5rRZLXAB8Ezqmqv3uuA1RVjyaHheVJJUmSJEkSDD6i4uPApc3zS4GPASQ5CrgN+Mmq+kL78CRJkiRJ0jiZd0RFkrXA6cChSbYC1wDXAjcleQewBbio2fx9wDLgd9IvJLuzqk7aD3FLkiRJkrQoxBQVQ5Wq0c+6cOqHJElSN02uWLXX9bPrVy9QJO105fvQ+JnvswsL8/ldMtHtbJNPPvNsJ+9pX7TkeSO5bvslmaYkSZIkafGzk02LkR0VkiRJkiRp0Zi3oyLJDUm2J5mZ0/bSJOuSbG4eJ5v285Pcl+SepvTov9yfwUuSJEmSNGrp6DIq+zKi4sPA2bu1vRu4o6qOAe5oXtM8P6GqTgTeTr9MqSRJkiRJ0j6Zt6Oiqj4LPL5b8/nAjc3zG4E3N9t+vb6dnfMFYJJMSZIkSZK07wbNUfGyqtoG0Dwu37UiyQVJ7gc+QX9UxR4lWdlMD5meWtMbMAxJkiRJktQlE8M+YFXdDtye5DTgV4Ezn2O7HtADy5NKkiRJkqS+QUdUPJrkMIDmcfvuGzRTRr4vyaEt4pMkSZIkaXEbddbLjmXTHLSj4uPApc3zS4GPAST5/iRpnr8WOBj4u7ZBSpIkSZKk8TDv1I8ka4HTgUOTbAWuAa4FbkryDmALcFGz+VuAn0qyA3ga+PE5yTUlSZIkSZL2KouhH8EcFZIkSZK08CZXrJp3m9n1q1ufZ8nEKCcS7H9//41nO3lP+8Lvet5IrtvQk2lKkiRJkjRO0u1+mAU3b46KJDck2Z5kZk7bS5OsS7K5eZzcbZ8VSb6V5Mf2R9CSJEmSJKmb9iWZ5oeBs3drezdwR1UdA9zRvAYgyUHAdcCfDilGSZIkSZI0JubtqGjKjD6+W/P5wI3N8xuBN89Z9wvAreyhZKkkSZIkSdLeDFqe9GVVtQ2geVwOkORw4ALg9+Y7QJKVSaaTTE+t6Q0YhiRJkiRJo5V0cxmVYSfT/E/Au6rqW5nnu6qqHtADq35IkiRJkqS+QTsqHk1yWFVtS3IY357mcRLwh00nxaHAuUl2VtUfDSFWSZIkSZLUcYNO/fg4cGnz/FLgYwBVdXRVvaKqXgHcAvyvdlJIkiRJkqR9Ne+IiiRrgdOBQ5NsBa4BrgVuSvIOYAtw0f4MUpIkSZKkxWqE6Rw6KVWjTw9hjgpJkiRp7yZXrJp3m9n1qxcgEo2b+T57+/K5WzLR7Xv5p765CG6s94OlB48mpeagUz8kSZIkSZKGbt6OiiQ3JNmeZGZO20uTrEuyuXmcbNpPT/JEknua5X37M3hJkiRJktQt+zKi4sPA2bu1vRu4o6qOAe5oXu/yF1V1YrP8ynDClCRJkiRJ42Dejoqq+izw+G7N5wM3Ns9vBN485LgkSZIkSTowpKPLiAyao+JlVbUNoHlcPmfd65Lcm+RTSX6wdYSSJEmSJGlsDDuZ5gbgf6qqE4DfBv7ouTZMsjLJdJLpqTW9IYchSZIkSZIORBMD7vdoksOqaluSw4DtAFX15K4NquqTSX4nyaFV9dXdD1BVPaAHlieVJEmSJEl9g46o+DhwafP8UuBjAEm+N+nXWU1ycnP8v2sbpCRJkiRJi1U6+jUq846oSLIWOB04NMlW4BrgWuCmJO8AtgAXNZv/GPDzSXYCTwMXV5WjJSRJkiRJ0j7JYuhHcOqHJEmSJB2YJlesmnebpz+3eoQ1JPa/p3d08572kOePZljFsJNpSpIkSZIkDWzejookNyTZnmRmTttLk6xLsrl5nJyz7vQk9yTZmOTO/RW4JEmSJEmLQdLNZd++9/z/7Z15tFTFncc/X0AZEAXEBBc0uHFcjoMaREdjRHEyMI5LjB6VGRfikolxSc7kRGcyo0kmYyDJUROJc44bKo6omLiO4q5MXHBfnoIbChKjiQbRaBaR3/xR1cO15z1e3b73vS6ev885dbq6bv++/b3Vdbvvra5bNUnS85JeknR6J9sHSro6bp8vaXR3mikjKi4FJjWVnQ7cZWZbA3fF50gaBpwPHGBm27Nq7grHcRzHcRzHcRzHcfoQkvoDPwMmA9sBR0jarullxwLLzGwr4Bxgene63XZUmNk84HdNxQcCl8X8ZcBBMT8F+IWZLYmxv+lO33Ecx3Ecx3Ecx3GcNZLxwEtmtsjM/gxcRegvKFLsP7gWmNhYLbQrWp2jYqSZ/RogPn46lo8Bhku6V9Jjko5qUd9xHMdxHMdxHMdxnLzZBHit8HxpLOv0NWa2AlgOjFitqpl1m4DRQEfh+TtN25fFxxnAQ8A6wAbAi8CYLjRPAB6N6YTmbSm+uvHcdo0cPPh+eF3k6sH3Iy8PuWjk4MH3Iy8Pvh95echFIwcPvh95efD9yMtDXRqe2p+arts7u3Y/FLio8PxI4Lym1zwLjCo8fxkYsbr3bXVExZuSNgKIj41bPJYCc83sfTN7C5gHjO1MwMwuMLNxMV3QtPmEFn3lppGDhzo0cvCQi0YOHurQyMFDHRo5eKhDIwcPuWjk4KEOjRw81KGRg4c6NHLwUIdGDh5y0cjBQx0aOXioQyMHD3Vo5OChDo0cPNSl4bSZpuv2zq7dlwKbFp6PAl7v6jWSBgBD+f/TS3yMVjsqbgSOjvmjgRti/gZgT0kDJA0GdgUWtPgejuM4juM4juM4juPkyyPA1pI2l7Q2cDihv6BIsf/gEOBui0MrumJAd+8qaTYwAdhA0lLgTGAacI2kY4ElxNU9zGyBpLnA08BKwhCQjk6FHcdxHMdxHMdxHMdZYzGzFZJOAm4D+gOXmNmzkr4HPGpmNwIXA7MkvUQYSXF4d7rddlSY2RFdbJrYxet/BPyoO91uaB5OsqZq5OChDo0cPOSikYOHOjRy8FCHRg4e6tDIwUMuGjl4qEMjBw91aOTgoQ6NHDzUoZGDh1w0cvBQh0YOHurQyMFDHRo5eKhDIwcPdWk4awBmdgtwS1PZGYX8H4mDG1JRNyMuHMdxHMdxHMdxHMdxeo1W56hwHMdxHMdxHMdxHMepHe+ocBzHcRzHcRzHcRwnG7yjwnEcx3Ecx3Ecx3GcbMiio0LSNpJOk/RTST+J+W1b0JgoaUhT+aQWPV1e8vW7Slov5gdJ+q6kmyRNlzQ0UWNtSUdJ2jc+nyJphqSvSVqr/F44fRFJn65BY0QdXhzH6TmqHut+nK/CvzfrxdumkyN+nDtO36LtHRWSTgOuAgQ8TFiHVcBsSacnapwC3ACcDHRIOrCw+ayE+Bub0k3AwY3nibtyCfBBzP8EGApMj2UzEzVmAvsBp0qaRZgZdT6wC3BRokaWrIk/HpKGSpomaaGkt2NaEMuGJWqsJ+kHkmZJmtK07fyE+PWb0gjgYUnDJa2f6GGapA1ifpykRcB8SYsl7ZWoMU7SPZKukLSppDskLZf0iKSdEuIHSPqKpLmSnpb0lKRbJf1jaiecpP5R498l7dG07V9TNLrQfaHEa08q1OVWkuZJekfSfEk7JGpsIekSSd+XNETShZI6JM2RNDpRo1J95lCX8fWV6rOmumz7cR5fV+lYz+E4jxqV6jOHuowaba9Pb5sf06ijbfr3Jn3ne7OvHOdNWiMl7SxpJ0kjy8T2BJIOqBi/laQvSdquLk/OJwwza2sCXgDW6qR8beDFRI1ngCExPxp4FDg1Pn8iIf5x4ApgArBXfPx1zO+V6GFBUa9p25OJGk/HxwHAm0D/+FyNbQkaQ4FpwELg7ZgWxLJhiRrrAT8AZgFTmradnxC/flMaAbwKDAfWT/QwDdgg5scBi4CXgMUlPpNxwD3xs90UuANYTugM2ykh/jbgNGDDQtmGseyORA8/j/tyEHBjfD6ws3bSRfxK4JWm9GF8XJR6fBTy9wC7xPwYwtrGKRoPA5OBI4DXgENi+UTgwYT42cB/ArsBo2LaLZZdnejhIuBK4OvAY8DZhW3d1mV83XvAuzG9F9NHjfKE+GcL+f8GvhjzE4D7Ez3MA74KnA50AP8U2+exwN2JGpXqM4e6rKM+a6rLth/n8XWVjnUyOM7rqM8c6jKX+vS2WXvb9O9N6zvfm1XbZWZtc0fgIcL5+p0xLYxlOydq7BBf/xphSdDhRY8J8Qc3pS8BbzSeJ3q4h1Xn7kcSrvEuIlynnZyi4clTMbXfQDgQP9NJ+WeA5xM1nmt6PgSYC5xNQicBYWTJNwgXsjvGsqQvuYLGHGBqzM8ExsX8GOCRRI0OQgfNcMKP1vqx/C8odIR0o+E/Hqviqp4kdtn+SrTNJ5uefxu4n9B5k1KX34xteYdC2Ssl2+ZCYEDMP9RVPXej8UQhv6SrbS3W5QuJHp4u5AcQfoh/AQxM8RDjzgMuB0a2Up/F/Wg+rknvTKxUl3XUZw51WUd99kJd9spxHmMqHes5HOd11GcOdZlLfXrb7NW26d+b9dWlnx+Vr88ngV07Kd8NeCpR45fAJGBYrJtngS1TfQArgJsJI8RnxvRefLwk0UNHsW0CI2J+cErb9OSpObX91g9Cz/RdcfjdBTHNBe4CTk3UeEPSjo0nZvZ74O+ADQg9jKvFzFaa2TnAVODbkmYQfoTKcBywl6SXge2AB+MQsgvjthQuJnxpPkn4wp4j6ULCwX5VosZoM5tuZm80CszsDTObDmyWqLGlmZ1uZteb2QGEESd3K/3Wi28BzwMHmNnmZrY5sDTmt0jUWEtS4zMYZGaPxH15gXBSkKRhZrea2ewQatdGjbsInT/dsVjSt4rD7+KwvNMIHR8pDJT0f8eZmf0H4eRmHuHHeLWY2Y8J7ecMSWdLWhewxPdu8DPgFkn7AHMlnSvp85K+S2hrKfxR0hckHQqYpIMA4tDIjxLil0k6tFgXkvpJOgxYluhh7UbGzFaY2QnAU8DdhM7JbjGzkwm3Zs2WdEr0U6Y+r5V0qaQtgOskfV3SZpKmAksSNVZKGiNpPDBY0jgIQySB/okaVeszh7qE6vXZqMtdaL0u236cx5iqx3oOxzlUr88c6hLyqE9vm6uoo23692agq9+grVmDvjeb2uU5a/BxDrCOmc1vLjSzh4B1EjWGmNlcM3sn1s1JhH3ajbR6+StgEOGa48tmNhV4y8ymmtmXEz18KGmTmP898H7M/4n0tuU4q2h3T4mZQRjRsBthmNEhMd+/RPwoCiMImrbt0YKf/YCzWtyXdYGxwGcp9JiXiN8Y2Djmh8X6GF8i/nZCR0Gxt34kYUTFnYkaC4B+TWVHE3pnF5f4TOYQRrWsS/kRKifHfdkH+A5wLvB54LvArESNB4EvEOb6WAwcFMv3ImFUBmFky3RC59Ey4HexbqaTfgvLD4F9OymfROKtTYWY/QnD+t5ooV1NAK4GniAMwbsFOIFObrvqIn4sYbTOrcA2hBOtd2Kb2D0hfnR8/98QhgK+EPNXA5snergCmNRJ+XHAhyXrox9wCvA/wOslY48hzB3zFuHfhucIc+EMTYyfSOjIWwB8jjBi6cVYHwcmajTq87exLhvxSfWZS13G+Kmt1mc3dXlQ4vtndZzHuJaO9RqO8x07Oc6XxeM86be0an3mUpc11Wdn35vJ9dnH2ubePVCXyb9BUWM0FX6HyOt78xh65nsz9Teo0TYXxHbZF86PHi+0za/01nEeNX5KuI3nMGD3mA6LZTMSNZ5q/vyBv4yf7dsl2uWphJHM4yl/7j4h7vf3gBnAA8AZhBHr3yz72XjyJLOynY9OzkgaTrjv8ECgFaQl6wAAB1RJREFUMYHlm4RbOKaZWbf/Gkj6IXC7md3ZVD4JOM/Mti7hZ3/C6JDRZrZhalyMnUC4j3IMYYTLa8D1hCFoKxLixxJ+CFcSbu35KqHD5VfA8Wb2QILGNoROl4csjNRplE8ys7mJ+7ENsAkwv0ljspndWiae0Du/pZl11OShjMa2hI60ljQk7Uro1X8Z2JbQIfmcmd2S8v5RYzxhdMwjCpMzTQIWVtDYk3AC/WiqRlP89tHDgpIedgVWFjQmU7IuClojCPPYnGtm/1A2vqBzuZkdVSF+I8Kwz0qT3kqaZWZHVoi/mTCaa2WL8XsSTtCeMbPbW9T4XNToqKCxJ6FT9eFWNGraj9IasW0vNLPlkgYTfo92Jpy8nmVmy0vEDwL+GdiJeCHWXXxBY4GZvRs9fCd6eKyCRqn9WM2+lKmLU4DrzCz1H+pe0Yj7sqWZdbTLQ4saaxNuBX2dcFE6mXBB+CxwgZl92E38QMLF4+tmdqfCBJC7Ey7QLzSzPyd6OLygcSThD5gfR43VeuhC4++jj+cS96NRD78qxO9BYj0UdLYCvkiY32IFoeNndsqxUdDYsknjxTIaTfEfEW4xvrwFDwdX3I9GXYyKGi8BV5bUmEw4d9+E8Ju+FLixxLnJFELHwkNN5ZsB/2Zmx5fwsglwDuE29tSR0I3YocAUVp27LwVuMLOFZXQcB/COik8Skqaa2cze1iie1LTLQysa8cToa4STkB0JE7TeELc9bmY7J7zPyYThdy1p1OShLo0TCf/stbIfZxJOCgcQetbHA/cB+wK3WRjy2Z2HZo1dgXsrapTy0UP70YpGZ6sR7UMYgoyFW7bKxIvQYZMUX4eHOjRq8vCwmY2P+eMIx8r1hNFYN5nZtJIax0eN6ypqnFjGRw/tRykPMe5ZYKyZrZB0AWHo788J/+KONbODS8Z/AFybGl+Hh1w0JC2PMS8TJnGcY2Zvdfe+q9GYHTV+W1HjmjI+MtqP/yJ87w4iTKy9DuE4nUg4Hz46MX4wYTTHEMIcFRMBzOyYEh4aGqU81KFRtR6iximEW6znAX9LuE1iGeFi/UQzu7enNWL8/oTfz1Y9nEoYRd22/XAcZzVYBsM6PPVOommSn3Zo5OAhVYOKq8nUoZGDhxr3oz/hxOpdYL1YPoj0SSjbrpGDh/jaSisVEYZfV13pqI7VknLYj+JEaI8An4r5dWhtMrW2aOTgIb620gpYVeP7kkZs3/0IHUUXE271mksYGbhu6mfabo0cPESNSiurVY3PRaMmD88U4gYD98b8ZpQ7r2hZIwcPNWo0VuxbQOsr9lVa9a9qfF0anjwVUw6TaTo1orA2eGfpGcJcFT2ukYOHmjT6W7zFwcxeJVwETZZ0NuEHPYWqGjl4qENjhZl9ZGYfAC+b2btR6w+EW3NSyEEjBw8Qlt59jHBb1XIL/9j8wczuM7P7EuI/WzG+Dg+57Ec/ScMVb6Gx+C+tmb1PGMK7pmjk4AGgQ2FSP4CntGqivjGE1Z96Or4vaZiFyb5vN7NjCbfenU+43WxRooccNHLwAKF9r02YN2sw4aIKwgTda/VCfC4adXiAVZPOD4xamNmSXtbIwUMdGtcQRmHsbWYjLNxCuTdhxMyckhoTmjSWJWpUjV+dRpn9cJxVtLunxFO9idA7viNheddiGk3ihE1VNXLwUNN+3E1crrZQNoCwtNhHiR4qaeTgoab9mA8Mjvl+hfKhpC+R13aNHDw0aTUmrZ1BCyONqsbnolElHniVcKHzSnzcMJYPIf2f87Zr5OCh0I4vJQzRn0+4IF9EGJ49tqfj+5IGq/k3lrAaVoqHtmvk4CG+7hux/hcTJrK8i7Ay2zPAmT0dn4tGTR5OBZ4mrNKxEJgayz8FzOsNjRw81KhRx3KvVZeGbrsHT56ak89R0ceQdDEw08x+2cm2K81sSk9r5OChDg1Jowj/fr/RybY9zOz+BA+VNHLwUIeGpIFm9qdOyjcANjKzZxI8tF0jBw9daO5HmF38X8rG1hGfi0YdHgpagwmrJ72yJmu0y4PCUoFbECdTM7M3S75npfi+oCFpjIUluVsmB40cPBR0NgYws9clDSPMDbTEzB7ujfhcNGrysD1hYuwOa3GixKoaOXioQ0PS7cCdwGWN7weFpV+PAf7azPbtaY0cPDhOM95R4TiO4ziO4ziO0wZUz4p9lTRy8OA4zXhHheM4juM4juM4TmYog9XycvDgfDLxjgrHcRzHcRzHcZzMkLTEzDZrp0YOHpxPJgO6f4njOI7jOI7jOI5TN5Ke7moTJVa6q6KRgwfHacY7KhzHcRzHcRzHcdrDSOBvCEt7FhHwQC9p5ODBcT6Gd1Q4juM4juM4juO0h5uBIWb2ZPMGSff2kkYOHhznY/gcFY7jOI7jOI7jOI7jZEO/dhtwHMdxHMdxHMdxHMdp4B0VjuM4juM4juM4juNkg3dUOI7jOI7jOI7jOI6TDd5R4TiO4ziO4ziO4zhONnhHheM4juM4juM4juM42fC/0a91cvrJvssAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 1440x1440 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Spectral Color Map used: Violet means 1, Red means -1 and Yellow means 0\n",
    "cfs1 = confusion_matrix(y_true, y_pred)\n",
    "plt.figure(figsize=(20,20))\n",
    "sn.heatmap(cfs1, cmap=\"Blues\", robust=True)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Multiclass Precision, Recall and Accuracy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 264,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.80      1.00      0.89        16\n",
      "           1       0.80      0.67      0.73         6\n",
      "           2       0.91      0.77      0.83        13\n",
      "           3       0.83      0.83      0.83         6\n",
      "           4       1.00      0.60      0.75         5\n",
      "           5       0.80      0.67      0.73         6\n",
      "           6       0.92      0.92      0.92        12\n",
      "           7       1.00      0.83      0.91         6\n",
      "           8       0.67      1.00      0.80        10\n",
      "           9       0.73      0.80      0.76        10\n",
      "          10       0.89      0.80      0.84        10\n",
      "          11       0.91      0.91      0.91        11\n",
      "          12       0.73      0.92      0.81        12\n",
      "          13       0.83      0.50      0.62        10\n",
      "          14       0.57      0.50      0.53         8\n",
      "          15       0.80      0.57      0.67         7\n",
      "          16       1.00      0.57      0.73         7\n",
      "          17       0.78      0.88      0.82         8\n",
      "          18       0.50      0.50      0.50         6\n",
      "          19       1.00      0.50      0.67         6\n",
      "          20       0.50      0.17      0.25         6\n",
      "          21       0.50      0.50      0.50         6\n",
      "          22       0.50      0.33      0.40         6\n",
      "          23       0.40      0.40      0.40         5\n",
      "          24       1.00      0.50      0.67         6\n",
      "          25       1.00      0.50      0.67         6\n",
      "          26       0.62      0.62      0.62        13\n",
      "          27       0.83      0.83      0.83        12\n",
      "          28       1.00      0.75      0.86        12\n",
      "          29       0.90      0.75      0.82        12\n",
      "          30       1.00      0.33      0.50         6\n",
      "          31       0.67      0.75      0.71         8\n",
      "          32       0.89      0.80      0.84        10\n",
      "          33       0.86      0.86      0.86         7\n",
      "          34       0.62      0.67      0.64        12\n",
      "          35       0.56      0.56      0.56         9\n",
      "          36       0.89      0.93      0.91       112\n",
      "          37       0.88      0.78      0.82         9\n",
      "          38       1.00      1.00      1.00         6\n",
      "          39       0.60      0.43      0.50         7\n",
      "          40       1.00      1.00      1.00         6\n",
      "          41       0.50      0.33      0.40         6\n",
      "          42       1.00      0.57      0.73         7\n",
      "          43       0.40      0.33      0.36         6\n",
      "          44       0.75      0.50      0.60         6\n",
      "          45       1.00      1.00      1.00         6\n",
      "          46       1.00      0.80      0.89         5\n",
      "          47       1.00      1.00      1.00         6\n",
      "          48       0.33      0.33      0.33         6\n",
      "          49       0.50      0.50      0.50         6\n",
      "          50       0.67      0.33      0.44         6\n",
      "          51       0.50      0.50      0.50         6\n",
      "          52       0.50      0.80      0.62         5\n",
      "          53       0.50      0.43      0.46         7\n",
      "          54       0.80      0.67      0.73         6\n",
      "          55       0.00      0.00      0.00         6\n",
      "          56       0.71      0.83      0.77         6\n",
      "          57       0.67      0.33      0.44         6\n",
      "          58       0.67      0.33      0.44         6\n",
      "          59       1.00      0.14      0.25         7\n",
      "          60       0.57      0.67      0.62         6\n",
      "          61       0.00      0.00      0.00         6\n",
      "          62       0.97      0.95      0.96       191\n",
      "          63       0.87      0.90      0.89       114\n",
      "          64       0.89      0.91      0.90       124\n",
      "          65       0.94      0.98      0.96       100\n",
      "          66       0.90      0.93      0.91        69\n",
      "          67       0.89      0.88      0.88        65\n",
      "          68       0.91      0.76      0.83        54\n",
      "          69       0.91      0.94      0.93       100\n",
      "          70       0.95      0.95      0.95       115\n",
      "          71       0.99      0.96      0.97        80\n",
      "          72       0.86      0.93      0.90        69\n",
      "          73       0.87      0.99      0.93        69\n",
      "          74       0.87      0.86      0.86        63\n",
      "          75       1.00      0.99      0.99        75\n",
      "          76       0.92      0.95      0.93       105\n",
      "          77       0.93      0.98      0.96        87\n",
      "          78       0.77      0.91      0.83        11\n",
      "          79       0.82      0.90      0.86        10\n",
      "          80       0.88      0.97      0.93        77\n",
      "          81       0.97      0.95      0.96        41\n",
      "          82       0.96      0.99      0.98       157\n",
      "          83       1.00      1.00      1.00        12\n",
      "          84       0.90      1.00      0.95         9\n",
      "          85       0.86      0.77      0.81        39\n",
      "          86       0.92      0.91      0.92       114\n",
      "          87       0.88      0.91      0.89        32\n",
      "          88       0.88      0.90      0.89       129\n",
      "          89       0.94      0.88      0.91       107\n",
      "          90       0.89      0.82      0.85        89\n",
      "          91       0.94      0.98      0.96        92\n",
      "          92       0.93      0.97      0.95       151\n",
      "          93       0.82      0.91      0.86        56\n",
      "          94       0.97      0.97      0.97        33\n",
      "          95       0.97      1.00      0.99       110\n",
      "          96       0.96      0.99      0.97       110\n",
      "          97       0.95      0.96      0.95       214\n",
      "          98       0.98      0.98      0.98       140\n",
      "          99       0.95      0.96      0.95        92\n",
      "         100       0.93      0.96      0.94        26\n",
      "         101       0.92      0.97      0.95        37\n",
      "         102       0.96      0.99      0.98       291\n",
      "         103       0.94      0.94      0.94        33\n",
      "         104       1.00      1.00      1.00         2\n",
      "         105       1.00      0.50      0.67         2\n",
      "         106       1.00      1.00      1.00         2\n",
      "\n",
      "    accuracy                           0.91      4276\n",
      "   macro avg       0.82      0.75      0.77      4276\n",
      "weighted avg       0.91      0.91      0.91      4276\n",
      "\n"
     ]
    }
   ],
   "source": [
    "cr1 = classification_report(y_true, y_pred, digits=2)\n",
    "print(cr1)\n",
    "cr1 = classification_report(y_true, y_pred, digits=4, output_dict=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 265,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Use this line before each model training to save memory\n",
    "del model1, history1, tensorboard_callback1, res"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model 2: CNN 2 (based on AlexNet) without Batch Normalization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 169,
   "metadata": {},
   "outputs": [],
   "source": [
    "K.clear_session()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 170,
   "metadata": {},
   "outputs": [],
   "source": [
    "# optimizer\n",
    "lr = 0.01\n",
    "mtn=0.9\n",
    "optimizer = SGD(lr=lr, momentum=mtn)\n",
    "# optimizer = RMSprop(lr=lr)\n",
    "\n",
    "# loss function\n",
    "loss = 'categorical_crossentropy'\n",
    "\n",
    "batch_size = 512\n",
    "epochs = 30"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 171,
   "metadata": {},
   "outputs": [],
   "source": [
    "# create network\n",
    "model2 = create_network2(input_shape, data_format, num_classes)\n",
    "# compile network\n",
    "model2.compile(loss=loss, optimizer=optimizer, metrics = ['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 172,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model 2 summary\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d_1 (Conv2D)            (None, 25, 32, 32)        650       \n",
      "_________________________________________________________________\n",
      "activation_1 (Activation)    (None, 25, 32, 32)        0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_1 (MaxPooling2 (None, 25, 16, 16)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_2 (Conv2D)            (None, 50, 16, 16)        11300     \n",
      "_________________________________________________________________\n",
      "activation_2 (Activation)    (None, 50, 16, 16)        0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_2 (MaxPooling2 (None, 50, 8, 8)          0         \n",
      "_________________________________________________________________\n",
      "conv2d_3 (Conv2D)            (None, 100, 8, 8)         20100     \n",
      "_________________________________________________________________\n",
      "activation_3 (Activation)    (None, 100, 8, 8)         0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_3 (MaxPooling2 (None, 100, 4, 4)         0         \n",
      "_________________________________________________________________\n",
      "conv2d_4 (Conv2D)            (None, 200, 4, 4)         80200     \n",
      "_________________________________________________________________\n",
      "activation_4 (Activation)    (None, 200, 4, 4)         0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_4 (MaxPooling2 (None, 200, 2, 2)         0         \n",
      "_________________________________________________________________\n",
      "flatten_1 (Flatten)          (None, 800)               0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 3200)              2563200   \n",
      "_________________________________________________________________\n",
      "activation_5 (Activation)    (None, 3200)              0         \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 1600)              5121600   \n",
      "_________________________________________________________________\n",
      "activation_6 (Activation)    (None, 1600)              0         \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 107)               171307    \n",
      "_________________________________________________________________\n",
      "activation_7 (Activation)    (None, 107)               0         \n",
      "=================================================================\n",
      "Total params: 7,968,357\n",
      "Trainable params: 7,968,357\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "# Model 2 Summary\n",
    "print('Model 2 summary')\n",
    "print(model2.summary())\n",
    "# plot_model(model2, to_file=model_dir+'model2.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 173,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 38484 samples, validate on 4276 samples\n",
      "Epoch 1/30\n",
      "38484/38484 [==============================] - 5s 141us/step - loss: 4.1694 - acc: 0.0592 - val_loss: 3.9533 - val_acc: 0.0716\n",
      "Epoch 2/30\n",
      "38484/38484 [==============================] - 4s 116us/step - loss: 3.8114 - acc: 0.1109 - val_loss: 3.1337 - val_acc: 0.2993\n",
      "Epoch 3/30\n",
      "38484/38484 [==============================] - 4s 116us/step - loss: 2.1652 - acc: 0.5048 - val_loss: 1.3114 - val_acc: 0.6819\n",
      "Epoch 4/30\n",
      "38484/38484 [==============================] - 4s 116us/step - loss: 1.0234 - acc: 0.7465 - val_loss: 0.7771 - val_acc: 0.8066\n",
      "Epoch 5/30\n",
      "38484/38484 [==============================] - 4s 116us/step - loss: 0.6659 - acc: 0.8297 - val_loss: 0.5823 - val_acc: 0.8454\n",
      "Epoch 6/30\n",
      "38484/38484 [==============================] - 4s 116us/step - loss: 0.5132 - acc: 0.8681 - val_loss: 0.4959 - val_acc: 0.8739\n",
      "Epoch 7/30\n",
      "38484/38484 [==============================] - 4s 116us/step - loss: 0.4061 - acc: 0.8939 - val_loss: 0.4230 - val_acc: 0.8896\n",
      "Epoch 8/30\n",
      "38484/38484 [==============================] - 4s 116us/step - loss: 0.3396 - acc: 0.9096 - val_loss: 0.4383 - val_acc: 0.8875\n",
      "Epoch 9/30\n",
      "38484/38484 [==============================] - 4s 116us/step - loss: 0.3004 - acc: 0.9190 - val_loss: 0.3637 - val_acc: 0.9100\n",
      "Epoch 10/30\n",
      "38484/38484 [==============================] - 4s 117us/step - loss: 0.2663 - acc: 0.9269 - val_loss: 0.3268 - val_acc: 0.9165\n",
      "Epoch 11/30\n",
      "38484/38484 [==============================] - 4s 116us/step - loss: 0.2250 - acc: 0.9383 - val_loss: 0.3307 - val_acc: 0.9149\n",
      "Epoch 12/30\n",
      "38484/38484 [==============================] - 4s 117us/step - loss: 0.2030 - acc: 0.9430 - val_loss: 0.3213 - val_acc: 0.9200\n",
      "Epoch 13/30\n",
      "38484/38484 [==============================] - 4s 117us/step - loss: 0.1813 - acc: 0.9490 - val_loss: 0.3278 - val_acc: 0.9203\n",
      "Epoch 14/30\n",
      "38484/38484 [==============================] - 5s 118us/step - loss: 0.1574 - acc: 0.9557 - val_loss: 0.3154 - val_acc: 0.9256\n",
      "Epoch 15/30\n",
      "38484/38484 [==============================] - 5s 118us/step - loss: 0.1377 - acc: 0.9596 - val_loss: 0.3030 - val_acc: 0.9254\n",
      "Epoch 16/30\n",
      "38484/38484 [==============================] - 4s 116us/step - loss: 0.1219 - acc: 0.9642 - val_loss: 0.3030 - val_acc: 0.9298\n",
      "Epoch 17/30\n",
      "38484/38484 [==============================] - 4s 117us/step - loss: 0.1080 - acc: 0.9688 - val_loss: 0.3008 - val_acc: 0.9331\n",
      "Epoch 18/30\n",
      "38484/38484 [==============================] - 4s 117us/step - loss: 0.1016 - acc: 0.9703 - val_loss: 0.3178 - val_acc: 0.9268\n",
      "Epoch 19/30\n",
      "38484/38484 [==============================] - 4s 117us/step - loss: 0.0813 - acc: 0.9760 - val_loss: 0.3306 - val_acc: 0.9200\n",
      "Epoch 20/30\n",
      "38484/38484 [==============================] - 5s 117us/step - loss: 0.0798 - acc: 0.9760 - val_loss: 0.3740 - val_acc: 0.9036\n",
      "Epoch 21/30\n",
      "38484/38484 [==============================] - 4s 117us/step - loss: 0.0778 - acc: 0.9780 - val_loss: 0.3247 - val_acc: 0.9333\n",
      "Epoch 22/30\n",
      "38484/38484 [==============================] - 4s 117us/step - loss: 0.0621 - acc: 0.9811 - val_loss: 0.3276 - val_acc: 0.9319\n",
      "Epoch 23/30\n",
      "38484/38484 [==============================] - 4s 117us/step - loss: 0.0569 - acc: 0.9835 - val_loss: 0.3173 - val_acc: 0.9373\n",
      "Epoch 24/30\n",
      "38484/38484 [==============================] - 4s 117us/step - loss: 0.0551 - acc: 0.9841 - val_loss: 0.3359 - val_acc: 0.9324\n",
      "Epoch 25/30\n",
      "38484/38484 [==============================] - 4s 117us/step - loss: 0.0484 - acc: 0.9855 - val_loss: 0.3403 - val_acc: 0.9249\n",
      "Epoch 26/30\n",
      "38484/38484 [==============================] - 5s 118us/step - loss: 0.0394 - acc: 0.9880 - val_loss: 0.3525 - val_acc: 0.9275\n",
      "Epoch 27/30\n",
      "38484/38484 [==============================] - 5s 117us/step - loss: 0.0399 - acc: 0.9876 - val_loss: 0.3415 - val_acc: 0.9308\n",
      "Epoch 28/30\n",
      "38484/38484 [==============================] - 5s 117us/step - loss: 0.0393 - acc: 0.9888 - val_loss: 0.3475 - val_acc: 0.9364\n",
      "Epoch 29/30\n",
      "38484/38484 [==============================] - 5s 117us/step - loss: 0.0381 - acc: 0.9888 - val_loss: 0.3409 - val_acc: 0.9336\n",
      "Epoch 30/30\n",
      "38484/38484 [==============================] - 5s 117us/step - loss: 0.0308 - acc: 0.9911 - val_loss: 0.3392 - val_acc: 0.9315\n",
      "Training completed on 2020-04-23, 19-42\n"
     ]
    }
   ],
   "source": [
    "# Fit dataset\n",
    "tensorboard_callback2 = keras.callbacks.TensorBoard(os.path.join(model_dir+'logs','model2'))\n",
    "history2 = model2.fit(X_train, \n",
    "                      y_train, \n",
    "                      validation_data=(X_test, y_test), \n",
    "                      batch_size=batch_size, \n",
    "                      epochs=epochs, \n",
    "                      verbose=1, \n",
    "                      callbacks=[tensorboard_callback2])\n",
    "model2.save(model_dir+'model2.h5')\n",
    "print(f\"Training completed on {datetime.datetime.now().strftime('%Y-%m-%d, %H-%M')}\")\n",
    "\n",
    "# store logs to csv file\n",
    "df_history2 = pd.DataFrame(data=history2.history)\n",
    "df_history2.to_csv(model_logs+'logs_model2.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Metrics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 174,
   "metadata": {},
   "outputs": [],
   "source": [
    "model2 = load_model(model_dir+'model2.h5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 175,
   "metadata": {},
   "outputs": [],
   "source": [
    "res = model2.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 176,
   "metadata": {},
   "outputs": [],
   "source": [
    "# convert predictions from one hot encoded to label value\n",
    "y_pred = convert_pred_list_ohe_to_labels(res, threshold=0.1, get_max=True)\n",
    "# convert test values from one hot encoded to label value\n",
    "y_true = convert_pred_list_ohe_to_labels(y_test, threshold=0.9, get_max=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Confusion Matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 177,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x1440 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Spectral Color Map used: Violet means 1, Red means -1 and Yellow means 0\n",
    "cfs2 = confusion_matrix(y_true, y_pred)\n",
    "plt.figure(figsize=(20,20))\n",
    "sn.heatmap(cfs2, cmap=\"Blues\", robust=True)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Multiclass Precision, Recall and Accuracy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 178,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.89      1.00      0.94        16\n",
      "           1       0.75      1.00      0.86         6\n",
      "           2       0.57      0.92      0.71        13\n",
      "           3       1.00      0.83      0.91         6\n",
      "           4       0.80      0.80      0.80         5\n",
      "           5       1.00      0.67      0.80         6\n",
      "           6       1.00      1.00      1.00        12\n",
      "           7       0.86      1.00      0.92         6\n",
      "           8       0.89      0.80      0.84        10\n",
      "           9       0.80      0.80      0.80        10\n",
      "          10       0.91      1.00      0.95        10\n",
      "          11       0.92      1.00      0.96        11\n",
      "          12       0.85      0.92      0.88        12\n",
      "          13       0.90      0.90      0.90        10\n",
      "          14       0.29      0.62      0.40         8\n",
      "          15       0.42      0.71      0.53         7\n",
      "          16       0.83      0.71      0.77         7\n",
      "          17       1.00      1.00      1.00         8\n",
      "          18       0.43      0.50      0.46         6\n",
      "          19       0.80      0.67      0.73         6\n",
      "          20       0.11      0.17      0.13         6\n",
      "          21       0.56      0.83      0.67         6\n",
      "          22       0.62      0.83      0.71         6\n",
      "          23       1.00      0.40      0.57         5\n",
      "          24       1.00      0.67      0.80         6\n",
      "          25       1.00      0.67      0.80         6\n",
      "          26       0.64      0.54      0.58        13\n",
      "          27       0.92      0.92      0.92        12\n",
      "          28       0.85      0.92      0.88        12\n",
      "          29       1.00      0.75      0.86        12\n",
      "          30       0.83      0.83      0.83         6\n",
      "          31       0.64      0.88      0.74         8\n",
      "          32       0.70      0.70      0.70        10\n",
      "          33       0.75      0.86      0.80         7\n",
      "          34       0.82      0.75      0.78        12\n",
      "          35       0.50      0.44      0.47         9\n",
      "          36       0.93      0.93      0.93       112\n",
      "          37       0.80      0.89      0.84         9\n",
      "          38       1.00      1.00      1.00         6\n",
      "          39       0.25      0.14      0.18         7\n",
      "          40       0.86      1.00      0.92         6\n",
      "          41       0.57      0.67      0.62         6\n",
      "          42       0.50      0.71      0.59         7\n",
      "          43       0.30      0.50      0.37         6\n",
      "          44       1.00      0.67      0.80         6\n",
      "          45       1.00      1.00      1.00         6\n",
      "          46       1.00      0.60      0.75         5\n",
      "          47       0.86      1.00      0.92         6\n",
      "          48       0.75      0.50      0.60         6\n",
      "          49       0.83      0.83      0.83         6\n",
      "          50       0.71      0.83      0.77         6\n",
      "          51       0.00      0.00      0.00         6\n",
      "          52       0.50      0.80      0.62         5\n",
      "          53       1.00      0.43      0.60         7\n",
      "          54       0.80      0.67      0.73         6\n",
      "          55       0.33      0.17      0.22         6\n",
      "          56       0.50      0.67      0.57         6\n",
      "          57       0.75      0.50      0.60         6\n",
      "          58       0.50      0.33      0.40         6\n",
      "          59       0.50      0.14      0.22         7\n",
      "          60       0.50      0.50      0.50         6\n",
      "          61       0.50      0.17      0.25         6\n",
      "          62       0.96      0.97      0.97       191\n",
      "          63       0.94      0.91      0.92       114\n",
      "          64       0.86      0.91      0.89       124\n",
      "          65       1.00      0.98      0.99       100\n",
      "          66       0.99      0.96      0.97        69\n",
      "          67       0.94      0.95      0.95        65\n",
      "          68       0.92      0.81      0.86        54\n",
      "          69       0.97      0.97      0.97       100\n",
      "          70       0.96      0.97      0.96       115\n",
      "          71       0.96      0.97      0.97        80\n",
      "          72       0.97      0.87      0.92        69\n",
      "          73       0.99      0.99      0.99        69\n",
      "          74       0.87      0.94      0.90        63\n",
      "          75       1.00      0.99      0.99        75\n",
      "          76       0.96      0.97      0.97       105\n",
      "          77       0.99      0.95      0.97        87\n",
      "          78       0.91      0.91      0.91        11\n",
      "          79       0.71      1.00      0.83        10\n",
      "          80       0.96      0.92      0.94        77\n",
      "          81       0.95      0.93      0.94        41\n",
      "          82       0.98      1.00      0.99       157\n",
      "          83       1.00      0.92      0.96        12\n",
      "          84       1.00      0.89      0.94         9\n",
      "          85       1.00      0.82      0.90        39\n",
      "          86       0.96      0.95      0.96       114\n",
      "          87       0.94      0.91      0.92        32\n",
      "          88       0.92      0.95      0.93       129\n",
      "          89       0.99      0.96      0.98       107\n",
      "          90       0.90      0.87      0.88        89\n",
      "          91       0.96      0.92      0.94        92\n",
      "          92       0.93      0.99      0.96       151\n",
      "          93       0.96      0.96      0.96        56\n",
      "          94       0.97      0.94      0.95        33\n",
      "          95       0.99      1.00      1.00       110\n",
      "          96       0.95      0.99      0.97       110\n",
      "          97       0.99      0.97      0.98       214\n",
      "          98       0.99      0.99      0.99       140\n",
      "          99       0.96      0.96      0.96        92\n",
      "         100       0.93      0.96      0.94        26\n",
      "         101       0.90      1.00      0.95        37\n",
      "         102       0.99      0.99      0.99       291\n",
      "         103       1.00      1.00      1.00        33\n",
      "         104       1.00      0.50      0.67         2\n",
      "         105       1.00      0.50      0.67         2\n",
      "         106       1.00      0.50      0.67         2\n",
      "\n",
      "    accuracy                           0.93      4276\n",
      "   macro avg       0.82      0.79      0.80      4276\n",
      "weighted avg       0.93      0.93      0.93      4276\n",
      "\n"
     ]
    }
   ],
   "source": [
    "cr2 = classification_report(y_true, y_pred, digits=2)\n",
    "print(cr2)\n",
    "cr2 = classification_report(y_true, y_pred, digits=4, output_dict=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 179,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Use this line before each model training to save memory\n",
    "del model2, history2, tensorboard_callback2, res"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model 3: CNN 2 (based on AlexNet) with Batch Normalization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 180,
   "metadata": {},
   "outputs": [],
   "source": [
    "K.clear_session()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 181,
   "metadata": {},
   "outputs": [],
   "source": [
    "# optimizer\n",
    "lr = 0.01\n",
    "mtn=0.9\n",
    "optimizer = SGD(lr=lr, momentum=mtn)\n",
    "# optimizer = RMSprop(lr=lr)\n",
    "\n",
    "# loss function\n",
    "loss = 'categorical_crossentropy'\n",
    "\n",
    "batch_size = 512\n",
    "epochs = 30"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 182,
   "metadata": {},
   "outputs": [],
   "source": [
    "# create network\n",
    "model3 = create_network3(input_shape, data_format, num_classes)\n",
    "# compile network\n",
    "model3.compile(loss=loss, optimizer=optimizer, metrics = ['accuracy'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Model 3 Summary"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 183,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d_1 (Conv2D)            (None, 25, 32, 32)        650       \n",
      "_________________________________________________________________\n",
      "batch_normalization_1 (Batch (None, 25, 32, 32)        128       \n",
      "_________________________________________________________________\n",
      "activation_1 (Activation)    (None, 25, 32, 32)        0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_1 (MaxPooling2 (None, 25, 16, 16)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_2 (Conv2D)            (None, 50, 16, 16)        11300     \n",
      "_________________________________________________________________\n",
      "batch_normalization_2 (Batch (None, 50, 16, 16)        64        \n",
      "_________________________________________________________________\n",
      "activation_2 (Activation)    (None, 50, 16, 16)        0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_2 (MaxPooling2 (None, 50, 8, 8)          0         \n",
      "_________________________________________________________________\n",
      "conv2d_3 (Conv2D)            (None, 100, 8, 8)         20100     \n",
      "_________________________________________________________________\n",
      "batch_normalization_3 (Batch (None, 100, 8, 8)         32        \n",
      "_________________________________________________________________\n",
      "activation_3 (Activation)    (None, 100, 8, 8)         0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_3 (MaxPooling2 (None, 100, 4, 4)         0         \n",
      "_________________________________________________________________\n",
      "conv2d_4 (Conv2D)            (None, 200, 4, 4)         80200     \n",
      "_________________________________________________________________\n",
      "batch_normalization_4 (Batch (None, 200, 4, 4)         16        \n",
      "_________________________________________________________________\n",
      "activation_4 (Activation)    (None, 200, 4, 4)         0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_4 (MaxPooling2 (None, 200, 2, 2)         0         \n",
      "_________________________________________________________________\n",
      "flatten_1 (Flatten)          (None, 800)               0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 3200)              2563200   \n",
      "_________________________________________________________________\n",
      "activation_5 (Activation)    (None, 3200)              0         \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 1600)              5121600   \n",
      "_________________________________________________________________\n",
      "activation_6 (Activation)    (None, 1600)              0         \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 107)               171307    \n",
      "_________________________________________________________________\n",
      "activation_7 (Activation)    (None, 107)               0         \n",
      "=================================================================\n",
      "Total params: 7,968,597\n",
      "Trainable params: 7,968,477\n",
      "Non-trainable params: 120\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "print(model3.summary())\n",
    "# plot_model(model3, to_file=model_dir+'model3.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 184,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 38484 samples, validate on 4276 samples\n",
      "Epoch 1/30\n",
      "38484/38484 [==============================] - 7s 194us/step - loss: 2.3785 - acc: 0.4783 - val_loss: 1.2326 - val_acc: 0.7142\n",
      "Epoch 2/30\n",
      "38484/38484 [==============================] - 6s 168us/step - loss: 0.6291 - acc: 0.8448 - val_loss: 0.4481 - val_acc: 0.8840\n",
      "Epoch 3/30\n",
      "38484/38484 [==============================] - 6s 168us/step - loss: 0.3645 - acc: 0.9068 - val_loss: 0.3577 - val_acc: 0.9095\n",
      "Epoch 4/30\n",
      "38484/38484 [==============================] - 6s 168us/step - loss: 0.2764 - acc: 0.9281 - val_loss: 0.2989 - val_acc: 0.9266\n",
      "Epoch 5/30\n",
      "38484/38484 [==============================] - 6s 166us/step - loss: 0.2342 - acc: 0.9372 - val_loss: 0.2804 - val_acc: 0.9315\n",
      "Epoch 6/30\n",
      "38484/38484 [==============================] - 6s 167us/step - loss: 0.1954 - acc: 0.9476 - val_loss: 0.2474 - val_acc: 0.9394\n",
      "Epoch 7/30\n",
      "38484/38484 [==============================] - 6s 167us/step - loss: 0.1630 - acc: 0.9559 - val_loss: 0.2718 - val_acc: 0.9352\n",
      "Epoch 8/30\n",
      "38484/38484 [==============================] - 6s 168us/step - loss: 0.1419 - acc: 0.9608 - val_loss: 0.2441 - val_acc: 0.9390\n",
      "Epoch 9/30\n",
      "38484/38484 [==============================] - 6s 167us/step - loss: 0.1255 - acc: 0.9663 - val_loss: 0.2410 - val_acc: 0.9411\n",
      "Epoch 10/30\n",
      "38484/38484 [==============================] - 6s 166us/step - loss: 0.1041 - acc: 0.9721 - val_loss: 0.2578 - val_acc: 0.9350\n",
      "Epoch 11/30\n",
      "38484/38484 [==============================] - 6s 167us/step - loss: 0.0900 - acc: 0.9755 - val_loss: 0.2321 - val_acc: 0.9443\n",
      "Epoch 12/30\n",
      "38484/38484 [==============================] - 6s 167us/step - loss: 0.0766 - acc: 0.9795 - val_loss: 0.2355 - val_acc: 0.9439\n",
      "Epoch 13/30\n",
      "38484/38484 [==============================] - 6s 168us/step - loss: 0.0685 - acc: 0.9820 - val_loss: 0.2349 - val_acc: 0.9394\n",
      "Epoch 14/30\n",
      "38484/38484 [==============================] - 6s 168us/step - loss: 0.0610 - acc: 0.9833 - val_loss: 0.2322 - val_acc: 0.9448\n",
      "Epoch 15/30\n",
      "38484/38484 [==============================] - 7s 170us/step - loss: 0.0545 - acc: 0.9852 - val_loss: 0.2439 - val_acc: 0.9397\n",
      "Epoch 16/30\n",
      "38484/38484 [==============================] - 7s 170us/step - loss: 0.0519 - acc: 0.9865 - val_loss: 0.2398 - val_acc: 0.9457\n",
      "Epoch 17/30\n",
      "38484/38484 [==============================] - 7s 171us/step - loss: 0.0382 - acc: 0.9902 - val_loss: 0.2410 - val_acc: 0.9446\n",
      "Epoch 18/30\n",
      "38484/38484 [==============================] - 7s 169us/step - loss: 0.0317 - acc: 0.9924 - val_loss: 0.2444 - val_acc: 0.9443\n",
      "Epoch 19/30\n",
      "38484/38484 [==============================] - 7s 169us/step - loss: 0.0278 - acc: 0.9932 - val_loss: 0.2450 - val_acc: 0.9476\n",
      "Epoch 20/30\n",
      "38484/38484 [==============================] - 7s 170us/step - loss: 0.0239 - acc: 0.9942 - val_loss: 0.2393 - val_acc: 0.9493\n",
      "Epoch 21/30\n",
      "38484/38484 [==============================] - 7s 171us/step - loss: 0.0196 - acc: 0.9952 - val_loss: 0.2502 - val_acc: 0.9464\n",
      "Epoch 22/30\n",
      "38484/38484 [==============================] - 7s 171us/step - loss: 0.0245 - acc: 0.9941 - val_loss: 0.2465 - val_acc: 0.9464\n",
      "Epoch 23/30\n",
      "38484/38484 [==============================] - 7s 171us/step - loss: 0.0188 - acc: 0.9953 - val_loss: 0.2380 - val_acc: 0.9481\n",
      "Epoch 24/30\n",
      "38484/38484 [==============================] - 7s 171us/step - loss: 0.0159 - acc: 0.9961 - val_loss: 0.2431 - val_acc: 0.9467\n",
      "Epoch 25/30\n",
      "38484/38484 [==============================] - 7s 171us/step - loss: 0.0125 - acc: 0.9975 - val_loss: 0.2450 - val_acc: 0.9474\n",
      "Epoch 26/30\n",
      "38484/38484 [==============================] - 7s 171us/step - loss: 0.0312 - acc: 0.9925 - val_loss: 0.2543 - val_acc: 0.9486\n",
      "Epoch 27/30\n",
      "38484/38484 [==============================] - 7s 170us/step - loss: 0.0125 - acc: 0.9975 - val_loss: 0.2482 - val_acc: 0.9469\n",
      "Epoch 28/30\n",
      "38484/38484 [==============================] - 7s 170us/step - loss: 0.0099 - acc: 0.9982 - val_loss: 0.2551 - val_acc: 0.9500\n",
      "Epoch 29/30\n",
      "38484/38484 [==============================] - 7s 170us/step - loss: 0.0079 - acc: 0.9989 - val_loss: 0.2549 - val_acc: 0.9495\n",
      "Epoch 30/30\n",
      "38484/38484 [==============================] - 7s 171us/step - loss: 0.0067 - acc: 0.9992 - val_loss: 0.2520 - val_acc: 0.9471\n",
      "Training completed on 2020-04-23, 19-47\n"
     ]
    }
   ],
   "source": [
    "# Fit dataset\n",
    "tensorboard_callback3 = keras.callbacks.TensorBoard(os.path.join(model_dir+'logs','model3'))\n",
    "history3 = model3.fit(X_train, \n",
    "                      y_train, \n",
    "                      validation_data=(X_test, y_test), \n",
    "                      batch_size=batch_size, \n",
    "                      epochs=epochs, \n",
    "                      verbose=1, \n",
    "                      callbacks=[tensorboard_callback3])\n",
    "model3.save(model_dir+'model3.h5')\n",
    "print(f\"Training completed on {datetime.datetime.now().strftime('%Y-%m-%d, %H-%M')}\")\n",
    "\n",
    "# store logs to csv file\n",
    "df_history3 = pd.DataFrame(data=history3.history)\n",
    "df_history3.to_csv(model_logs+'logs_model3.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Metrics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 185,
   "metadata": {},
   "outputs": [],
   "source": [
    "# to load model use\n",
    "model3 = load_model(model_dir+'model3.h5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 186,
   "metadata": {},
   "outputs": [],
   "source": [
    "res = model3.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 187,
   "metadata": {},
   "outputs": [],
   "source": [
    "# convert predictions from one hot encoded to label value\n",
    "y_pred = convert_pred_list_ohe_to_labels(res, threshold=0.1, get_max=True)\n",
    "# convert test values from one hot encoded to label value\n",
    "y_true = convert_pred_list_ohe_to_labels(y_test, threshold=0.9, get_max=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Confusion Matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 188,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x24f3e711470>"
      ]
     },
     "execution_count": 188,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x1440 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Spectral Color Map used: Violet means 1, Red means -1 and Yellow means 0\n",
    "cfs3 = confusion_matrix(y_true, y_pred)\n",
    "plt.figure(figsize=(20,20))\n",
    "sn.heatmap(cfs3, cmap=\"Blues\", robust=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Muticlass Precision, Recall and Accuracy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 189,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       1.00      1.00      1.00        16\n",
      "           1       1.00      1.00      1.00         6\n",
      "           2       0.86      0.92      0.89        13\n",
      "           3       0.86      1.00      0.92         6\n",
      "           4       1.00      0.80      0.89         5\n",
      "           5       1.00      0.83      0.91         6\n",
      "           6       0.86      1.00      0.92        12\n",
      "           7       0.86      1.00      0.92         6\n",
      "           8       0.73      0.80      0.76        10\n",
      "           9       0.90      0.90      0.90        10\n",
      "          10       0.91      1.00      0.95        10\n",
      "          11       1.00      1.00      1.00        11\n",
      "          12       0.86      1.00      0.92        12\n",
      "          13       0.89      0.80      0.84        10\n",
      "          14       0.43      0.38      0.40         8\n",
      "          15       1.00      0.71      0.83         7\n",
      "          16       1.00      0.71      0.83         7\n",
      "          17       1.00      1.00      1.00         8\n",
      "          18       0.44      0.67      0.53         6\n",
      "          19       1.00      0.67      0.80         6\n",
      "          20       0.50      0.17      0.25         6\n",
      "          21       0.62      0.83      0.71         6\n",
      "          22       0.67      0.67      0.67         6\n",
      "          23       0.50      0.40      0.44         5\n",
      "          24       0.71      0.83      0.77         6\n",
      "          25       0.80      0.67      0.73         6\n",
      "          26       0.64      0.69      0.67        13\n",
      "          27       0.92      0.92      0.92        12\n",
      "          28       0.85      0.92      0.88        12\n",
      "          29       1.00      0.92      0.96        12\n",
      "          30       1.00      1.00      1.00         6\n",
      "          31       0.64      0.88      0.74         8\n",
      "          32       0.91      1.00      0.95        10\n",
      "          33       0.86      0.86      0.86         7\n",
      "          34       0.80      0.67      0.73        12\n",
      "          35       0.50      0.33      0.40         9\n",
      "          36       0.96      0.96      0.96       112\n",
      "          37       0.90      1.00      0.95         9\n",
      "          38       1.00      1.00      1.00         6\n",
      "          39       1.00      0.29      0.44         7\n",
      "          40       0.86      1.00      0.92         6\n",
      "          41       0.80      0.67      0.73         6\n",
      "          42       0.75      0.86      0.80         7\n",
      "          43       0.43      0.50      0.46         6\n",
      "          44       1.00      0.83      0.91         6\n",
      "          45       1.00      1.00      1.00         6\n",
      "          46       1.00      0.60      0.75         5\n",
      "          47       1.00      1.00      1.00         6\n",
      "          48       0.33      0.33      0.33         6\n",
      "          49       0.80      0.67      0.73         6\n",
      "          50       0.83      0.83      0.83         6\n",
      "          51       0.75      0.50      0.60         6\n",
      "          52       0.83      1.00      0.91         5\n",
      "          53       0.80      0.57      0.67         7\n",
      "          54       0.83      0.83      0.83         6\n",
      "          55       0.50      0.17      0.25         6\n",
      "          56       0.83      0.83      0.83         6\n",
      "          57       0.50      0.33      0.40         6\n",
      "          58       0.50      0.33      0.40         6\n",
      "          59       0.00      0.00      0.00         7\n",
      "          60       0.50      0.33      0.40         6\n",
      "          61       1.00      0.67      0.80         6\n",
      "          62       0.98      0.98      0.98       191\n",
      "          63       0.95      0.96      0.96       114\n",
      "          64       0.94      0.95      0.94       124\n",
      "          65       1.00      0.99      0.99       100\n",
      "          66       0.94      0.99      0.96        69\n",
      "          67       0.94      0.94      0.94        65\n",
      "          68       0.98      0.83      0.90        54\n",
      "          69       0.96      0.97      0.97       100\n",
      "          70       0.98      0.97      0.98       115\n",
      "          71       0.94      0.99      0.96        80\n",
      "          72       0.93      0.99      0.96        69\n",
      "          73       0.99      0.99      0.99        69\n",
      "          74       0.92      0.95      0.94        63\n",
      "          75       1.00      0.99      0.99        75\n",
      "          76       0.93      0.98      0.95       105\n",
      "          77       0.99      0.97      0.98        87\n",
      "          78       0.91      0.91      0.91        11\n",
      "          79       0.91      1.00      0.95        10\n",
      "          80       0.94      0.99      0.96        77\n",
      "          81       0.98      0.98      0.98        41\n",
      "          82       0.96      0.99      0.98       157\n",
      "          83       1.00      1.00      1.00        12\n",
      "          84       0.90      1.00      0.95         9\n",
      "          85       0.92      0.90      0.91        39\n",
      "          86       0.95      0.93      0.94       114\n",
      "          87       0.97      0.88      0.92        32\n",
      "          88       0.95      0.94      0.94       129\n",
      "          89       0.98      0.95      0.97       107\n",
      "          90       0.93      0.89      0.91        89\n",
      "          91       0.96      1.00      0.98        92\n",
      "          92       0.95      0.95      0.95       151\n",
      "          93       0.95      0.96      0.96        56\n",
      "          94       0.94      0.97      0.96        33\n",
      "          95       0.99      1.00      1.00       110\n",
      "          96       0.98      1.00      0.99       110\n",
      "          97       0.98      0.98      0.98       214\n",
      "          98       0.98      0.99      0.98       140\n",
      "          99       0.97      0.98      0.97        92\n",
      "         100       0.86      0.96      0.91        26\n",
      "         101       0.97      1.00      0.99        37\n",
      "         102       0.99      0.99      0.99       291\n",
      "         103       0.94      1.00      0.97        33\n",
      "         104       1.00      1.00      1.00         2\n",
      "         105       1.00      1.00      1.00         2\n",
      "         106       1.00      0.50      0.67         2\n",
      "\n",
      "    accuracy                           0.95      4276\n",
      "   macro avg       0.86      0.83      0.84      4276\n",
      "weighted avg       0.95      0.95      0.95      4276\n",
      "\n"
     ]
    }
   ],
   "source": [
    "cr3 = classification_report(y_true, y_pred, digits=2)\n",
    "print(cr3)\n",
    "cr3 = classification_report(y_true, y_pred, digits=4, output_dict=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 190,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Use this line before each model training to save memory\n",
    "del model3, history3, tensorboard_callback3, res"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# ANN"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 207,
   "metadata": {},
   "outputs": [],
   "source": [
    "config = tf.ConfigProto()\n",
    "config.gpu_options.allow_growth=True\n",
    "config.gpu_options.per_process_gpu_memory_fraction=0.5\n",
    "sess = tf.Session(config=config)\n",
    "K.set_session(sess)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Split Set for ANN models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 208,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CNN Input Shape is (1024,)\n"
     ]
    }
   ],
   "source": [
    "# input shape of dataset\n",
    "input_shape = (32 * 32,)\n",
    "print(\"CNN Input Shape is\", input_shape)\n",
    "\n",
    "# define the channels location\n",
    "data_format = 'channels_first'\n",
    "\n",
    "X_train, X_test, y_train, y_test = process_x_y_train_test_stratified_ann(train, test, input_shape=input_shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model 4: ANN without regularization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 244,
   "metadata": {},
   "outputs": [],
   "source": [
    "K.clear_session()\n",
    "# optimizer\n",
    "lr = 0.01\n",
    "mtn=0.95\n",
    "optimizer = SGD(lr=lr, momentum=mtn)\n",
    "# optimizer = RMSprop(lr=lr)\n",
    "\n",
    "# loss function\n",
    "loss = 'categorical_crossentropy'\n",
    "\n",
    "batch_size = 512\n",
    "epochs = 50"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 245,
   "metadata": {},
   "outputs": [],
   "source": [
    "# create network\n",
    "model4 = create_network_ann1(input_shape, num_classes)\n",
    "# compile network\n",
    "model4.compile(loss=loss, optimizer=optimizer, metrics = ['accuracy'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Model Summary"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 246,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "dense_1 (Dense)              (None, 1024)              1049600   \n",
      "_________________________________________________________________\n",
      "activation_1 (Activation)    (None, 1024)              0         \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 1024)              0         \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 1024)              1049600   \n",
      "_________________________________________________________________\n",
      "activation_2 (Activation)    (None, 1024)              0         \n",
      "_________________________________________________________________\n",
      "dropout_2 (Dropout)          (None, 1024)              0         \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 2048)              2099200   \n",
      "_________________________________________________________________\n",
      "activation_3 (Activation)    (None, 2048)              0         \n",
      "_________________________________________________________________\n",
      "dropout_3 (Dropout)          (None, 2048)              0         \n",
      "_________________________________________________________________\n",
      "dense_4 (Dense)              (None, 107)               219243    \n",
      "_________________________________________________________________\n",
      "activation_4 (Activation)    (None, 107)               0         \n",
      "=================================================================\n",
      "Total params: 4,417,643\n",
      "Trainable params: 4,417,643\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "print(model4.summary())\n",
    "#plot_model(model4, to_file=dir_+'model4.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 247,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 38484 samples, validate on 4276 samples\n",
      "Epoch 1/50\n",
      "38484/38484 [==============================] - 2s 48us/step - loss: 3.3530 - acc: 0.2708 - val_loss: 2.0885 - val_acc: 0.5463\n",
      "Epoch 2/50\n",
      "38484/38484 [==============================] - 1s 34us/step - loss: 1.7043 - acc: 0.6236 - val_loss: 1.2426 - val_acc: 0.7142\n",
      "Epoch 3/50\n",
      "38484/38484 [==============================] - 1s 34us/step - loss: 1.2398 - acc: 0.7104 - val_loss: 0.9923 - val_acc: 0.7629\n",
      "Epoch 4/50\n",
      "38484/38484 [==============================] - 1s 34us/step - loss: 1.0355 - acc: 0.7470 - val_loss: 0.8715 - val_acc: 0.7895\n",
      "Epoch 5/50\n",
      "38484/38484 [==============================] - 1s 34us/step - loss: 0.8914 - acc: 0.7765 - val_loss: 0.8018 - val_acc: 0.8057\n",
      "Epoch 6/50\n",
      "38484/38484 [==============================] - 1s 34us/step - loss: 0.7888 - acc: 0.8026 - val_loss: 0.7245 - val_acc: 0.8260\n",
      "Epoch 7/50\n",
      "38484/38484 [==============================] - 1s 34us/step - loss: 0.7329 - acc: 0.8132 - val_loss: 0.6655 - val_acc: 0.8400\n",
      "Epoch 8/50\n",
      "38484/38484 [==============================] - 1s 34us/step - loss: 0.6758 - acc: 0.8296 - val_loss: 0.6224 - val_acc: 0.8517\n",
      "Epoch 9/50\n",
      "38484/38484 [==============================] - 1s 35us/step - loss: 0.6281 - acc: 0.8409 - val_loss: 0.6012 - val_acc: 0.8585\n",
      "Epoch 10/50\n",
      "38484/38484 [==============================] - 1s 34us/step - loss: 0.5878 - acc: 0.8489 - val_loss: 0.5690 - val_acc: 0.8637\n",
      "Epoch 11/50\n",
      "38484/38484 [==============================] - 1s 34us/step - loss: 0.5444 - acc: 0.8593 - val_loss: 0.5539 - val_acc: 0.8674\n",
      "Epoch 12/50\n",
      "38484/38484 [==============================] - 1s 34us/step - loss: 0.5258 - acc: 0.8639 - val_loss: 0.5349 - val_acc: 0.8718\n",
      "Epoch 13/50\n",
      "38484/38484 [==============================] - 1s 34us/step - loss: 0.4994 - acc: 0.8693 - val_loss: 0.5140 - val_acc: 0.8779\n",
      "Epoch 14/50\n",
      "38484/38484 [==============================] - 1s 34us/step - loss: 0.4688 - acc: 0.8770 - val_loss: 0.5029 - val_acc: 0.8831\n",
      "Epoch 15/50\n",
      "38484/38484 [==============================] - 1s 34us/step - loss: 0.4479 - acc: 0.8824 - val_loss: 0.4826 - val_acc: 0.8875\n",
      "Epoch 16/50\n",
      "38484/38484 [==============================] - 1s 34us/step - loss: 0.4244 - acc: 0.8894 - val_loss: 0.4877 - val_acc: 0.8863\n",
      "Epoch 17/50\n",
      "38484/38484 [==============================] - 1s 34us/step - loss: 0.4051 - acc: 0.8936 - val_loss: 0.4824 - val_acc: 0.8866\n",
      "Epoch 18/50\n",
      "38484/38484 [==============================] - 1s 34us/step - loss: 0.3918 - acc: 0.8963 - val_loss: 0.4652 - val_acc: 0.8868\n",
      "Epoch 19/50\n",
      "38484/38484 [==============================] - 1s 34us/step - loss: 0.3769 - acc: 0.9001 - val_loss: 0.4658 - val_acc: 0.8859\n",
      "Epoch 20/50\n",
      "38484/38484 [==============================] - 1s 35us/step - loss: 0.3667 - acc: 0.9010 - val_loss: 0.4730 - val_acc: 0.8842\n",
      "Epoch 21/50\n",
      "38484/38484 [==============================] - 1s 34us/step - loss: 0.3429 - acc: 0.9083 - val_loss: 0.4604 - val_acc: 0.8884\n",
      "Epoch 22/50\n",
      "38484/38484 [==============================] - 1s 34us/step - loss: 0.3411 - acc: 0.9081 - val_loss: 0.4492 - val_acc: 0.8884\n",
      "Epoch 23/50\n",
      "38484/38484 [==============================] - 1s 34us/step - loss: 0.3303 - acc: 0.9096 - val_loss: 0.4631 - val_acc: 0.8915\n",
      "Epoch 24/50\n",
      "38484/38484 [==============================] - 1s 35us/step - loss: 0.3190 - acc: 0.9137 - val_loss: 0.4379 - val_acc: 0.8938\n",
      "Epoch 25/50\n",
      "38484/38484 [==============================] - 1s 34us/step - loss: 0.3139 - acc: 0.9136 - val_loss: 0.4268 - val_acc: 0.8969\n",
      "Epoch 26/50\n",
      "38484/38484 [==============================] - 1s 34us/step - loss: 0.2874 - acc: 0.9218 - val_loss: 0.4269 - val_acc: 0.8992\n",
      "Epoch 27/50\n",
      "38484/38484 [==============================] - 1s 35us/step - loss: 0.2810 - acc: 0.9227 - val_loss: 0.4194 - val_acc: 0.9008\n",
      "Epoch 28/50\n",
      "38484/38484 [==============================] - 1s 36us/step - loss: 0.2702 - acc: 0.9266 - val_loss: 0.4289 - val_acc: 0.8950\n",
      "Epoch 29/50\n",
      "38484/38484 [==============================] - 1s 35us/step - loss: 0.2639 - acc: 0.9277 - val_loss: 0.4231 - val_acc: 0.8964\n",
      "Epoch 30/50\n",
      "38484/38484 [==============================] - 1s 35us/step - loss: 0.2571 - acc: 0.9289 - val_loss: 0.4220 - val_acc: 0.8985\n",
      "Epoch 31/50\n",
      "38484/38484 [==============================] - 1s 35us/step - loss: 0.2412 - acc: 0.9324 - val_loss: 0.4251 - val_acc: 0.9018\n",
      "Epoch 32/50\n",
      "38484/38484 [==============================] - 1s 35us/step - loss: 0.2358 - acc: 0.9338 - val_loss: 0.4162 - val_acc: 0.9001\n",
      "Epoch 33/50\n",
      "38484/38484 [==============================] - 1s 35us/step - loss: 0.2235 - acc: 0.9366 - val_loss: 0.4156 - val_acc: 0.9011\n",
      "Epoch 34/50\n",
      "38484/38484 [==============================] - 1s 35us/step - loss: 0.2239 - acc: 0.9371 - val_loss: 0.4282 - val_acc: 0.8971\n",
      "Epoch 35/50\n",
      "38484/38484 [==============================] - 1s 35us/step - loss: 0.2180 - acc: 0.9372 - val_loss: 0.4245 - val_acc: 0.9022\n",
      "Epoch 36/50\n",
      "38484/38484 [==============================] - 1s 36us/step - loss: 0.2081 - acc: 0.9413 - val_loss: 0.4151 - val_acc: 0.9032\n",
      "Epoch 37/50\n",
      "38484/38484 [==============================] - 1s 37us/step - loss: 0.1998 - acc: 0.9423 - val_loss: 0.4270 - val_acc: 0.9013\n",
      "Epoch 38/50\n",
      "38484/38484 [==============================] - 1s 36us/step - loss: 0.1981 - acc: 0.9423 - val_loss: 0.4002 - val_acc: 0.9102\n",
      "Epoch 39/50\n",
      "38484/38484 [==============================] - 1s 36us/step - loss: 0.1880 - acc: 0.9464 - val_loss: 0.4115 - val_acc: 0.9090\n",
      "Epoch 40/50\n",
      "38484/38484 [==============================] - 1s 36us/step - loss: 0.1783 - acc: 0.9479 - val_loss: 0.4032 - val_acc: 0.9097\n",
      "Epoch 41/50\n",
      "38484/38484 [==============================] - 1s 36us/step - loss: 0.1762 - acc: 0.9481 - val_loss: 0.3985 - val_acc: 0.9079\n",
      "Epoch 42/50\n",
      "38484/38484 [==============================] - 1s 35us/step - loss: 0.1770 - acc: 0.9478 - val_loss: 0.4098 - val_acc: 0.9067\n",
      "Epoch 43/50\n",
      "38484/38484 [==============================] - 1s 37us/step - loss: 0.1672 - acc: 0.9493 - val_loss: 0.4158 - val_acc: 0.9067\n",
      "Epoch 44/50\n",
      "38484/38484 [==============================] - 1s 36us/step - loss: 0.1666 - acc: 0.9508 - val_loss: 0.4211 - val_acc: 0.9046\n",
      "Epoch 45/50\n",
      "38484/38484 [==============================] - 1s 36us/step - loss: 0.1630 - acc: 0.9518 - val_loss: 0.4153 - val_acc: 0.9062\n",
      "Epoch 46/50\n",
      "38484/38484 [==============================] - 1s 36us/step - loss: 0.1568 - acc: 0.9537 - val_loss: 0.3994 - val_acc: 0.9097\n",
      "Epoch 47/50\n",
      "38484/38484 [==============================] - 1s 35us/step - loss: 0.1502 - acc: 0.9555 - val_loss: 0.4164 - val_acc: 0.9072\n",
      "Epoch 48/50\n",
      "38484/38484 [==============================] - 1s 35us/step - loss: 0.1448 - acc: 0.9570 - val_loss: 0.4053 - val_acc: 0.9093\n",
      "Epoch 49/50\n",
      "38484/38484 [==============================] - 1s 36us/step - loss: 0.1429 - acc: 0.9564 - val_loss: 0.4061 - val_acc: 0.9104\n",
      "Epoch 50/50\n",
      "38484/38484 [==============================] - 1s 35us/step - loss: 0.1371 - acc: 0.9588 - val_loss: 0.4122 - val_acc: 0.9100\n",
      "Training completed on 2020-04-23, 20-03\n"
     ]
    }
   ],
   "source": [
    "# Fit dataset\n",
    "tensorboard_callback4 = keras.callbacks.TensorBoard(os.path.join(model_dir+'logs','model4'))\n",
    "history4 = model4.fit(X_train, \n",
    "                      y_train, \n",
    "                      validation_data=(X_test, y_test), \n",
    "                      batch_size=batch_size, \n",
    "                      epochs=epochs, \n",
    "                      verbose=1, \n",
    "                      callbacks=[tensorboard_callback4])\n",
    "model4.save(model_dir+'model4.h5')\n",
    "print(f\"Training completed on {datetime.datetime.now().strftime('%Y-%m-%d, %H-%M')}\")\n",
    "\n",
    "# store logs to csv file\n",
    "df_history4 = pd.DataFrame(data=history4.history)\n",
    "df_history4.to_csv(model_logs+'logs_model4.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Metrics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 248,
   "metadata": {},
   "outputs": [],
   "source": [
    "# to load model use\n",
    "model4 = load_model(model_dir+'model4.h5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 249,
   "metadata": {},
   "outputs": [],
   "source": [
    "res = model4.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 250,
   "metadata": {},
   "outputs": [],
   "source": [
    "# convert predictions from one hot encoded to label value\n",
    "y_pred = convert_pred_list_ohe_to_labels(res, threshold=0.001, get_max=True)\n",
    "# convert test values from one hot encoded to label value\n",
    "y_true = convert_pred_list_ohe_to_labels(y_test, threshold=0.9, get_max=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Confusion Matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 251,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x1440 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Spectral Color Map used: Violet means 1, Red means -1 and Yellow means 0\n",
    "cfs4 = confusion_matrix(y_true, y_pred)\n",
    "plt.figure(figsize=(20,20))\n",
    "sn.heatmap(cfs4, cmap=\"Blues\", robust=True, square=True)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Multiclass Precision, Recall and Accuracy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 252,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       1.00      1.00      1.00        16\n",
      "           1       1.00      0.33      0.50         6\n",
      "           2       0.79      0.85      0.81        13\n",
      "           3       1.00      1.00      1.00         6\n",
      "           4       1.00      0.80      0.89         5\n",
      "           5       1.00      0.50      0.67         6\n",
      "           6       0.91      0.83      0.87        12\n",
      "           7       1.00      1.00      1.00         6\n",
      "           8       0.80      0.80      0.80        10\n",
      "           9       0.86      0.60      0.71        10\n",
      "          10       0.90      0.90      0.90        10\n",
      "          11       0.92      1.00      0.96        11\n",
      "          12       0.79      0.92      0.85        12\n",
      "          13       1.00      0.80      0.89        10\n",
      "          14       0.33      0.25      0.29         8\n",
      "          15       1.00      0.57      0.73         7\n",
      "          16       0.83      0.71      0.77         7\n",
      "          17       0.73      1.00      0.84         8\n",
      "          18       0.50      0.50      0.50         6\n",
      "          19       0.80      0.67      0.73         6\n",
      "          20       1.00      0.17      0.29         6\n",
      "          21       0.50      0.50      0.50         6\n",
      "          22       0.40      0.33      0.36         6\n",
      "          23       1.00      0.40      0.57         5\n",
      "          24       0.80      0.67      0.73         6\n",
      "          25       1.00      0.50      0.67         6\n",
      "          26       0.56      0.69      0.62        13\n",
      "          27       0.79      0.92      0.85        12\n",
      "          28       0.89      0.67      0.76        12\n",
      "          29       0.88      0.58      0.70        12\n",
      "          30       0.83      0.83      0.83         6\n",
      "          31       0.64      0.88      0.74         8\n",
      "          32       0.75      0.60      0.67        10\n",
      "          33       0.86      0.86      0.86         7\n",
      "          34       0.75      0.75      0.75        12\n",
      "          35       0.50      0.56      0.53         9\n",
      "          36       0.90      0.92      0.91       112\n",
      "          37       0.78      0.78      0.78         9\n",
      "          38       1.00      1.00      1.00         6\n",
      "          39       0.00      0.00      0.00         7\n",
      "          40       1.00      1.00      1.00         6\n",
      "          41       0.50      0.67      0.57         6\n",
      "          42       0.56      0.71      0.63         7\n",
      "          43       0.00      0.00      0.00         6\n",
      "          44       1.00      0.67      0.80         6\n",
      "          45       1.00      0.83      0.91         6\n",
      "          46       1.00      0.80      0.89         5\n",
      "          47       0.75      1.00      0.86         6\n",
      "          48       0.50      0.50      0.50         6\n",
      "          49       0.67      0.67      0.67         6\n",
      "          50       1.00      0.33      0.50         6\n",
      "          51       0.67      0.33      0.44         6\n",
      "          52       0.83      1.00      0.91         5\n",
      "          53       0.57      0.57      0.57         7\n",
      "          54       0.80      0.67      0.73         6\n",
      "          55       0.33      0.17      0.22         6\n",
      "          56       0.71      0.83      0.77         6\n",
      "          57       0.60      0.50      0.55         6\n",
      "          58       0.57      0.67      0.62         6\n",
      "          59       0.00      0.00      0.00         7\n",
      "          60       0.50      0.50      0.50         6\n",
      "          61       1.00      0.17      0.29         6\n",
      "          62       0.90      0.95      0.92       191\n",
      "          63       0.89      0.90      0.90       114\n",
      "          64       0.85      0.90      0.87       124\n",
      "          65       0.97      0.99      0.98       100\n",
      "          66       0.86      0.94      0.90        69\n",
      "          67       0.91      0.89      0.90        65\n",
      "          68       0.88      0.70      0.78        54\n",
      "          69       0.91      0.96      0.94       100\n",
      "          70       0.92      0.95      0.94       115\n",
      "          71       0.94      0.94      0.94        80\n",
      "          72       0.88      0.93      0.90        69\n",
      "          73       0.89      0.97      0.93        69\n",
      "          74       0.91      0.84      0.88        63\n",
      "          75       0.99      0.99      0.99        75\n",
      "          76       0.93      0.93      0.93       105\n",
      "          77       0.93      0.92      0.92        87\n",
      "          78       0.61      1.00      0.76        11\n",
      "          79       0.77      1.00      0.87        10\n",
      "          80       0.94      0.99      0.96        77\n",
      "          81       0.98      0.98      0.98        41\n",
      "          82       0.98      0.98      0.98       157\n",
      "          83       1.00      1.00      1.00        12\n",
      "          84       0.75      1.00      0.86         9\n",
      "          85       0.91      0.82      0.86        39\n",
      "          86       0.91      0.93      0.92       114\n",
      "          87       0.94      0.91      0.92        32\n",
      "          88       0.90      0.88      0.89       129\n",
      "          89       0.90      0.89      0.90       107\n",
      "          90       0.93      0.75      0.83        89\n",
      "          91       0.92      0.99      0.95        92\n",
      "          92       0.90      0.97      0.93       151\n",
      "          93       0.88      0.89      0.88        56\n",
      "          94       0.94      0.94      0.94        33\n",
      "          95       0.98      1.00      0.99       110\n",
      "          96       0.98      0.99      0.99       110\n",
      "          97       0.96      0.96      0.96       214\n",
      "          98       0.97      0.99      0.98       140\n",
      "          99       0.94      0.95      0.94        92\n",
      "         100       0.93      0.96      0.94        26\n",
      "         101       0.84      0.97      0.90        37\n",
      "         102       0.97      0.98      0.98       291\n",
      "         103       0.94      0.94      0.94        33\n",
      "         104       0.67      1.00      0.80         2\n",
      "         105       1.00      0.50      0.67         2\n",
      "         106       1.00      1.00      1.00         2\n",
      "\n",
      "    accuracy                           0.91      4276\n",
      "   macro avg       0.82      0.77      0.77      4276\n",
      "weighted avg       0.91      0.91      0.91      4276\n",
      "\n"
     ]
    }
   ],
   "source": [
    "cr4 = classification_report(y_true, y_pred, digits=2)\n",
    "print(cr4)\n",
    "cr4 = classification_report(y_true, y_pred, digits=4, output_dict=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 253,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Use this line before each model training to save memory\n",
    "del model4, history4, tensorboard_callback4, res"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model 5: ANN with regularization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 234,
   "metadata": {},
   "outputs": [],
   "source": [
    "K.clear_session()\n",
    "# optimizer\n",
    "lr = 0.01\n",
    "mtn=0.9\n",
    "optimizer = SGD(lr=lr, momentum=mtn)\n",
    "# optimizer = RMSprop(lr=lr)\n",
    "\n",
    "# loss function\n",
    "loss = 'categorical_crossentropy'\n",
    "\n",
    "batch_size = 128\n",
    "epochs = 50"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 235,
   "metadata": {},
   "outputs": [],
   "source": [
    "# create network\n",
    "model5 = create_network_ann2(input_shape, num_classes)\n",
    "# compile network\n",
    "model5.compile(loss=loss, optimizer=optimizer, metrics = ['accuracy'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Model Summary"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 236,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "dense_1 (Dense)              (None, 1024)              1049600   \n",
      "_________________________________________________________________\n",
      "activation_1 (Activation)    (None, 1024)              0         \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 1024)              0         \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 1024)              1049600   \n",
      "_________________________________________________________________\n",
      "activation_2 (Activation)    (None, 1024)              0         \n",
      "_________________________________________________________________\n",
      "dropout_2 (Dropout)          (None, 1024)              0         \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 2048)              2099200   \n",
      "_________________________________________________________________\n",
      "activation_3 (Activation)    (None, 2048)              0         \n",
      "_________________________________________________________________\n",
      "dropout_3 (Dropout)          (None, 2048)              0         \n",
      "_________________________________________________________________\n",
      "dense_4 (Dense)              (None, 107)               219243    \n",
      "_________________________________________________________________\n",
      "activation_4 (Activation)    (None, 107)               0         \n",
      "=================================================================\n",
      "Total params: 4,417,643\n",
      "Trainable params: 4,417,643\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "print(model5.summary())\n",
    "#plot_model(model5, to_file=dir_+'model5.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 237,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 38484 samples, validate on 4276 samples\n",
      "Epoch 1/50\n",
      "38484/38484 [==============================] - 4s 116us/step - loss: 27.0259 - acc: 0.5062 - val_loss: 14.3673 - val_acc: 0.6763\n",
      "Epoch 2/50\n",
      "38484/38484 [==============================] - 3s 85us/step - loss: 9.0958 - acc: 0.6883 - val_loss: 5.5372 - val_acc: 0.7025\n",
      "Epoch 3/50\n",
      "38484/38484 [==============================] - 3s 85us/step - loss: 3.9557 - acc: 0.7077 - val_loss: 2.8371 - val_acc: 0.7299\n",
      "Epoch 4/50\n",
      "38484/38484 [==============================] - 3s 86us/step - loss: 2.4254 - acc: 0.7153 - val_loss: 2.0086 - val_acc: 0.7378\n",
      "Epoch 5/50\n",
      "38484/38484 [==============================] - 3s 85us/step - loss: 1.9412 - acc: 0.7227 - val_loss: 1.7677 - val_acc: 0.7526\n",
      "Epoch 6/50\n",
      "38484/38484 [==============================] - 3s 85us/step - loss: 1.7744 - acc: 0.7295 - val_loss: 1.7197 - val_acc: 0.7339\n",
      "Epoch 7/50\n",
      "38484/38484 [==============================] - 3s 85us/step - loss: 1.6964 - acc: 0.7366 - val_loss: 1.8780 - val_acc: 0.6941\n",
      "Epoch 8/50\n",
      "38484/38484 [==============================] - 3s 85us/step - loss: 1.6706 - acc: 0.7389 - val_loss: 1.5913 - val_acc: 0.7626\n",
      "Epoch 9/50\n",
      "38484/38484 [==============================] - 3s 85us/step - loss: 1.6341 - acc: 0.7438 - val_loss: 1.5873 - val_acc: 0.7582\n",
      "Epoch 10/50\n",
      "38484/38484 [==============================] - 3s 85us/step - loss: 1.5944 - acc: 0.7497 - val_loss: 1.5736 - val_acc: 0.7615\n",
      "Epoch 11/50\n",
      "38484/38484 [==============================] - 3s 85us/step - loss: 1.6017 - acc: 0.7503 - val_loss: 1.5832 - val_acc: 0.7491\n",
      "Epoch 12/50\n",
      "38484/38484 [==============================] - 3s 85us/step - loss: 1.5664 - acc: 0.7556 - val_loss: 1.6281 - val_acc: 0.7420\n",
      "Epoch 13/50\n",
      "38484/38484 [==============================] - 3s 86us/step - loss: 1.5450 - acc: 0.7603 - val_loss: 1.5745 - val_acc: 0.7481\n",
      "Epoch 14/50\n",
      "38484/38484 [==============================] - 3s 85us/step - loss: 1.5435 - acc: 0.7584 - val_loss: 1.5358 - val_acc: 0.7636\n",
      "Epoch 15/50\n",
      "38484/38484 [==============================] - 3s 85us/step - loss: 1.5400 - acc: 0.7615 - val_loss: 1.4439 - val_acc: 0.7862\n",
      "Epoch 16/50\n",
      "38484/38484 [==============================] - 3s 85us/step - loss: 1.5409 - acc: 0.7566 - val_loss: 1.5536 - val_acc: 0.7626\n",
      "Epoch 17/50\n",
      "38484/38484 [==============================] - 3s 86us/step - loss: 1.5240 - acc: 0.7591 - val_loss: 1.5772 - val_acc: 0.7362\n",
      "Epoch 18/50\n",
      "38484/38484 [==============================] - 3s 87us/step - loss: 1.5257 - acc: 0.7590 - val_loss: 1.5761 - val_acc: 0.7575\n",
      "Epoch 19/50\n",
      "38484/38484 [==============================] - 3s 85us/step - loss: 1.5255 - acc: 0.7575 - val_loss: 1.4347 - val_acc: 0.7919\n",
      "Epoch 20/50\n",
      "38484/38484 [==============================] - 3s 85us/step - loss: 1.4909 - acc: 0.7669 - val_loss: 1.5520 - val_acc: 0.7523\n",
      "Epoch 21/50\n",
      "38484/38484 [==============================] - 3s 85us/step - loss: 1.4871 - acc: 0.7657 - val_loss: 1.5335 - val_acc: 0.7685\n",
      "Epoch 22/50\n",
      "38484/38484 [==============================] - 3s 85us/step - loss: 1.4800 - acc: 0.7684 - val_loss: 1.3858 - val_acc: 0.7949\n",
      "Epoch 23/50\n",
      "38484/38484 [==============================] - 3s 85us/step - loss: 1.4799 - acc: 0.7658 - val_loss: 1.4688 - val_acc: 0.7802\n",
      "Epoch 24/50\n",
      "38484/38484 [==============================] - 3s 86us/step - loss: 1.4856 - acc: 0.7640 - val_loss: 1.4902 - val_acc: 0.7753\n",
      "Epoch 25/50\n",
      "38484/38484 [==============================] - 3s 85us/step - loss: 1.4434 - acc: 0.7732 - val_loss: 1.4828 - val_acc: 0.7685\n",
      "Epoch 26/50\n",
      "38484/38484 [==============================] - 3s 86us/step - loss: 1.4516 - acc: 0.7723 - val_loss: 1.4543 - val_acc: 0.7755\n",
      "Epoch 27/50\n",
      "38484/38484 [==============================] - 3s 86us/step - loss: 1.4540 - acc: 0.7689 - val_loss: 1.6648 - val_acc: 0.7145\n",
      "Epoch 28/50\n",
      "38484/38484 [==============================] - 3s 85us/step - loss: 1.4534 - acc: 0.7684 - val_loss: 1.4058 - val_acc: 0.7872\n",
      "Epoch 29/50\n",
      "38484/38484 [==============================] - 3s 85us/step - loss: 1.4324 - acc: 0.7740 - val_loss: 1.4262 - val_acc: 0.7841\n",
      "Epoch 30/50\n",
      "38484/38484 [==============================] - 3s 85us/step - loss: 1.4342 - acc: 0.7762 - val_loss: 1.4917 - val_acc: 0.7575\n",
      "Epoch 31/50\n",
      "38484/38484 [==============================] - 3s 86us/step - loss: 1.4360 - acc: 0.7727 - val_loss: 1.4074 - val_acc: 0.7827\n",
      "Epoch 32/50\n",
      "38484/38484 [==============================] - 3s 85us/step - loss: 1.4136 - acc: 0.7780 - val_loss: 1.3641 - val_acc: 0.7991\n",
      "Epoch 33/50\n",
      "38484/38484 [==============================] - 3s 86us/step - loss: 1.4173 - acc: 0.7762 - val_loss: 1.3982 - val_acc: 0.7916\n",
      "Epoch 34/50\n",
      "38484/38484 [==============================] - 3s 85us/step - loss: 1.4275 - acc: 0.7741 - val_loss: 1.5144 - val_acc: 0.7533\n",
      "Epoch 35/50\n",
      "38484/38484 [==============================] - 3s 86us/step - loss: 1.4162 - acc: 0.7743 - val_loss: 1.4269 - val_acc: 0.7886\n",
      "Epoch 36/50\n",
      "38484/38484 [==============================] - 3s 86us/step - loss: 1.4068 - acc: 0.7780 - val_loss: 1.4452 - val_acc: 0.7778\n",
      "Epoch 37/50\n",
      "38484/38484 [==============================] - 3s 85us/step - loss: 1.3989 - acc: 0.7813 - val_loss: 1.4457 - val_acc: 0.7729\n",
      "Epoch 38/50\n",
      "38484/38484 [==============================] - 3s 86us/step - loss: 1.4063 - acc: 0.7762 - val_loss: 1.4492 - val_acc: 0.7741\n",
      "Epoch 39/50\n",
      "38484/38484 [==============================] - 3s 85us/step - loss: 1.3946 - acc: 0.7807 - val_loss: 1.4018 - val_acc: 0.7827\n",
      "Epoch 40/50\n",
      "38484/38484 [==============================] - 3s 86us/step - loss: 1.4225 - acc: 0.7728 - val_loss: 1.3600 - val_acc: 0.7972\n",
      "Epoch 41/50\n",
      "38484/38484 [==============================] - 3s 85us/step - loss: 1.3841 - acc: 0.7823 - val_loss: 1.3977 - val_acc: 0.7881\n",
      "Epoch 42/50\n",
      "38484/38484 [==============================] - 3s 85us/step - loss: 1.4005 - acc: 0.7753 - val_loss: 1.4583 - val_acc: 0.7699\n",
      "Epoch 43/50\n",
      "38484/38484 [==============================] - 3s 86us/step - loss: 1.3872 - acc: 0.7827 - val_loss: 1.4891 - val_acc: 0.7596\n",
      "Epoch 44/50\n",
      "38484/38484 [==============================] - 3s 86us/step - loss: 1.3669 - acc: 0.7844 - val_loss: 1.4864 - val_acc: 0.7493\n",
      "Epoch 45/50\n",
      "38484/38484 [==============================] - 3s 90us/step - loss: 1.4130 - acc: 0.7739 - val_loss: 1.3735 - val_acc: 0.7907\n",
      "Epoch 46/50\n",
      "38484/38484 [==============================] - 3s 91us/step - loss: 1.3752 - acc: 0.7815 - val_loss: 1.4672 - val_acc: 0.7671\n",
      "Epoch 47/50\n",
      "38484/38484 [==============================] - 3s 87us/step - loss: 1.3522 - acc: 0.7873 - val_loss: 1.2674 - val_acc: 0.8190\n",
      "Epoch 48/50\n",
      "38484/38484 [==============================] - 3s 86us/step - loss: 1.3883 - acc: 0.7763 - val_loss: 1.3682 - val_acc: 0.7888\n",
      "Epoch 49/50\n",
      "38484/38484 [==============================] - 3s 87us/step - loss: 1.3927 - acc: 0.7772 - val_loss: 1.3338 - val_acc: 0.8057\n",
      "Epoch 50/50\n",
      "38484/38484 [==============================] - 3s 85us/step - loss: 1.3789 - acc: 0.7808 - val_loss: 1.4135 - val_acc: 0.7839\n",
      "Training completed on 2020-04-23, 20-01\n"
     ]
    }
   ],
   "source": [
    "# Fit dataset\n",
    "tensorboard_callback5 = keras.callbacks.TensorBoard(os.path.join(model_dir+'logs','model5'))\n",
    "history5 = model5.fit(X_train, \n",
    "                      y_train, \n",
    "                      validation_data=(X_test, y_test), \n",
    "                      batch_size=batch_size, \n",
    "                      epochs=epochs, \n",
    "                      verbose=1, \n",
    "                      callbacks=[tensorboard_callback5])\n",
    "model5.save(model_dir+'model5.h5')\n",
    "print(f\"Training completed on {datetime.datetime.now().strftime('%Y-%m-%d, %H-%M')}\")\n",
    "\n",
    "# store logs to csv file\n",
    "df_history5 = pd.DataFrame(data=history5.history)\n",
    "df_history5.to_csv(model_logs+'logs_model5.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Metrics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 238,
   "metadata": {},
   "outputs": [],
   "source": [
    "# to load model use\n",
    "model5 = load_model(model_dir+'model5.h5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 239,
   "metadata": {},
   "outputs": [],
   "source": [
    "res = model5.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 240,
   "metadata": {},
   "outputs": [],
   "source": [
    "# convert predictions from one hot encoded to label value\n",
    "y_pred = convert_pred_list_ohe_to_labels(res, threshold=0.001, get_max=True)\n",
    "# convert test values from one hot encoded to label value\n",
    "y_true = convert_pred_list_ohe_to_labels(y_test, threshold=0.9, get_max=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Confusion Matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 241,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x1440 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Spectral Color Map used: Violet means 1, Red means -1 and Yellow means 0\n",
    "cfs5 = confusion_matrix(y_true, y_pred)\n",
    "plt.figure(figsize=(20,20))\n",
    "sn.heatmap(cfs5, cmap=\"Blues\", robust=True)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Multiclass Precision, Recall and Accuracy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 242,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.91      0.62      0.74        16\n",
      "           1       0.50      0.17      0.25         6\n",
      "           2       0.80      0.31      0.44        13\n",
      "           3       0.67      0.67      0.67         6\n",
      "           4       0.00      0.00      0.00         5\n",
      "           5       1.00      0.33      0.50         6\n",
      "           6       0.89      0.67      0.76        12\n",
      "           7       0.00      0.00      0.00         6\n",
      "           8       1.00      0.20      0.33        10\n",
      "           9       0.40      0.20      0.27        10\n",
      "          10       0.67      0.40      0.50        10\n",
      "          11       0.82      0.82      0.82        11\n",
      "          12       0.00      0.00      0.00        12\n",
      "          13       1.00      0.20      0.33        10\n",
      "          14       0.00      0.00      0.00         8\n",
      "          15       0.33      0.43      0.38         7\n",
      "          16       1.00      0.14      0.25         7\n",
      "          17       0.55      0.75      0.63         8\n",
      "          18       0.00      0.00      0.00         6\n",
      "          19       1.00      0.67      0.80         6\n",
      "          20       0.00      0.00      0.00         6\n",
      "          21       1.00      0.17      0.29         6\n",
      "          22       0.00      0.00      0.00         6\n",
      "          23       0.00      0.00      0.00         5\n",
      "          24       0.00      0.00      0.00         6\n",
      "          25       1.00      0.17      0.29         6\n",
      "          26       0.62      0.62      0.62        13\n",
      "          27       0.00      0.00      0.00        12\n",
      "          28       0.45      0.42      0.43        12\n",
      "          29       0.50      0.17      0.25        12\n",
      "          30       0.00      0.00      0.00         6\n",
      "          31       0.57      0.50      0.53         8\n",
      "          32       0.00      0.00      0.00        10\n",
      "          33       0.80      0.57      0.67         7\n",
      "          34       0.25      0.08      0.12        12\n",
      "          35       0.50      0.33      0.40         9\n",
      "          36       0.88      0.84      0.86       112\n",
      "          37       1.00      0.11      0.20         9\n",
      "          38       1.00      0.33      0.50         6\n",
      "          39       0.00      0.00      0.00         7\n",
      "          40       1.00      0.50      0.67         6\n",
      "          41       0.00      0.00      0.00         6\n",
      "          42       0.00      0.00      0.00         7\n",
      "          43       0.33      0.17      0.22         6\n",
      "          44       1.00      0.17      0.29         6\n",
      "          45       1.00      0.17      0.29         6\n",
      "          46       0.00      0.00      0.00         5\n",
      "          47       1.00      0.83      0.91         6\n",
      "          48       0.00      0.00      0.00         6\n",
      "          49       1.00      0.17      0.29         6\n",
      "          50       0.50      0.17      0.25         6\n",
      "          51       0.00      0.00      0.00         6\n",
      "          52       0.67      0.40      0.50         5\n",
      "          53       1.00      0.14      0.25         7\n",
      "          54       0.40      0.33      0.36         6\n",
      "          55       0.00      0.00      0.00         6\n",
      "          56       0.00      0.00      0.00         6\n",
      "          57       1.00      0.33      0.50         6\n",
      "          58       0.00      0.00      0.00         6\n",
      "          59       0.00      0.00      0.00         7\n",
      "          60       0.00      0.00      0.00         6\n",
      "          61       0.00      0.00      0.00         6\n",
      "          62       0.77      0.87      0.82       191\n",
      "          63       0.71      0.83      0.77       114\n",
      "          64       0.78      0.64      0.70       124\n",
      "          65       0.92      0.90      0.91       100\n",
      "          66       0.73      0.78      0.76        69\n",
      "          67       0.80      0.80      0.80        65\n",
      "          68       0.62      0.59      0.60        54\n",
      "          69       0.84      0.87      0.86       100\n",
      "          70       0.56      0.95      0.70       115\n",
      "          71       0.77      0.85      0.81        80\n",
      "          72       0.47      0.86      0.61        69\n",
      "          73       0.82      0.71      0.76        69\n",
      "          74       0.88      0.78      0.82        63\n",
      "          75       0.99      0.91      0.94        75\n",
      "          76       0.98      0.50      0.66       105\n",
      "          77       0.79      0.78      0.79        87\n",
      "          78       0.56      0.45      0.50        11\n",
      "          79       0.78      0.70      0.74        10\n",
      "          80       0.80      0.94      0.86        77\n",
      "          81       0.62      0.98      0.76        41\n",
      "          82       0.94      0.88      0.91       157\n",
      "          83       0.92      1.00      0.96        12\n",
      "          84       1.00      0.78      0.88         9\n",
      "          85       0.91      0.51      0.66        39\n",
      "          86       0.90      0.77      0.83       114\n",
      "          87       0.86      0.75      0.80        32\n",
      "          88       0.57      0.88      0.69       129\n",
      "          89       0.72      0.93      0.81       107\n",
      "          90       0.65      0.66      0.66        89\n",
      "          91       0.84      0.99      0.91        92\n",
      "          92       0.75      0.95      0.84       151\n",
      "          93       0.73      0.86      0.79        56\n",
      "          94       1.00      0.64      0.78        33\n",
      "          95       0.83      1.00      0.91       110\n",
      "          96       0.95      0.92      0.94       110\n",
      "          97       0.95      0.91      0.93       214\n",
      "          98       0.92      0.86      0.89       140\n",
      "          99       0.73      0.99      0.84        92\n",
      "         100       0.94      0.58      0.71        26\n",
      "         101       0.65      0.95      0.77        37\n",
      "         102       0.93      0.98      0.96       291\n",
      "         103       0.66      0.70      0.68        33\n",
      "         104       0.00      0.00      0.00         2\n",
      "         105       0.00      0.00      0.00         2\n",
      "         106       0.00      0.00      0.00         2\n",
      "\n",
      "    accuracy                           0.78      4276\n",
      "   macro avg       0.58      0.45      0.47      4276\n",
      "weighted avg       0.78      0.78      0.76      4276\n",
      "\n"
     ]
    }
   ],
   "source": [
    "cr5 = classification_report(y_true, y_pred, digits=2)\n",
    "print(cr5)\n",
    "cr5 = classification_report(y_true, y_pred, digits=4, output_dict=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 243,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Use this line before each model training to save memory\n",
    "del model5, history5, tensorboard_callback5, res"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# CNN (contd)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model 6: LeNet-5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 191,
   "metadata": {},
   "outputs": [],
   "source": [
    "K.clear_session()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 196,
   "metadata": {},
   "outputs": [],
   "source": [
    "# optimizer\n",
    "lr = 0.01\n",
    "mtn=0.9\n",
    "\n",
    "# optimizer = RMSprop(lr=lr)\n",
    "# optimizer = SGD(lr=lr, momentum=mtn)\n",
    "optimizer = Adam()\n",
    "\n",
    "# loss function\n",
    "loss = 'categorical_crossentropy'\n",
    "\n",
    "batch_size = 512\n",
    "epochs = 30"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 197,
   "metadata": {},
   "outputs": [],
   "source": [
    "# create network\n",
    "model6 = create_network6(input_shape, num_classes)\n",
    "# compile network\n",
    "model6.compile(loss=loss, optimizer=optimizer, metrics = ['accuracy'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Model Summary"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 198,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d_4 (Conv2D)            (None, 6, 32, 32)         156       \n",
      "_________________________________________________________________\n",
      "activation_7 (Activation)    (None, 6, 32, 32)         0         \n",
      "_________________________________________________________________\n",
      "average_pooling2d_3 (Average (None, 6, 31, 31)         0         \n",
      "_________________________________________________________________\n",
      "conv2d_5 (Conv2D)            (None, 16, 27, 27)        2416      \n",
      "_________________________________________________________________\n",
      "activation_8 (Activation)    (None, 16, 27, 27)        0         \n",
      "_________________________________________________________________\n",
      "average_pooling2d_4 (Average (None, 16, 13, 13)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_6 (Conv2D)            (None, 120, 9, 9)         48120     \n",
      "_________________________________________________________________\n",
      "activation_9 (Activation)    (None, 120, 9, 9)         0         \n",
      "_________________________________________________________________\n",
      "flatten_2 (Flatten)          (None, 9720)              0         \n",
      "_________________________________________________________________\n",
      "dense_4 (Dense)              (None, 1024)              9954304   \n",
      "_________________________________________________________________\n",
      "activation_10 (Activation)   (None, 1024)              0         \n",
      "_________________________________________________________________\n",
      "dropout_3 (Dropout)          (None, 1024)              0         \n",
      "_________________________________________________________________\n",
      "dense_5 (Dense)              (None, 512)               524800    \n",
      "_________________________________________________________________\n",
      "activation_11 (Activation)   (None, 512)               0         \n",
      "_________________________________________________________________\n",
      "dropout_4 (Dropout)          (None, 512)               0         \n",
      "_________________________________________________________________\n",
      "dense_6 (Dense)              (None, 107)               54891     \n",
      "_________________________________________________________________\n",
      "activation_12 (Activation)   (None, 107)               0         \n",
      "=================================================================\n",
      "Total params: 10,584,687\n",
      "Trainable params: 10,584,687\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "print(model6.summary())\n",
    "#plot_model(model6, to_file=dir_+'model6.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 199,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 38484 samples, validate on 4276 samples\n",
      "Epoch 1/30\n",
      "38484/38484 [==============================] - 7s 179us/step - loss: 2.1216 - acc: 0.5380 - val_loss: 1.0998 - val_acc: 0.7463\n",
      "Epoch 2/30\n",
      "38484/38484 [==============================] - 6s 155us/step - loss: 0.9790 - acc: 0.7691 - val_loss: 0.7583 - val_acc: 0.8246\n",
      "Epoch 3/30\n",
      "38484/38484 [==============================] - 6s 156us/step - loss: 0.7105 - acc: 0.8298 - val_loss: 0.6674 - val_acc: 0.8471\n",
      "Epoch 4/30\n",
      "38484/38484 [==============================] - 6s 157us/step - loss: 0.5865 - acc: 0.8567 - val_loss: 0.5727 - val_acc: 0.8662\n",
      "Epoch 5/30\n",
      "38484/38484 [==============================] - 6s 157us/step - loss: 0.4750 - acc: 0.8818 - val_loss: 0.5167 - val_acc: 0.8756\n",
      "Epoch 6/30\n",
      "38484/38484 [==============================] - 6s 154us/step - loss: 0.4132 - acc: 0.8959 - val_loss: 0.4995 - val_acc: 0.8803\n",
      "Epoch 7/30\n",
      "38484/38484 [==============================] - 6s 156us/step - loss: 0.3506 - acc: 0.9105 - val_loss: 0.4352 - val_acc: 0.8948\n",
      "Epoch 8/30\n",
      "38484/38484 [==============================] - 6s 156us/step - loss: 0.2928 - acc: 0.9268 - val_loss: 0.4554 - val_acc: 0.8906\n",
      "Epoch 9/30\n",
      "38484/38484 [==============================] - 6s 156us/step - loss: 0.2805 - acc: 0.9267 - val_loss: 0.4022 - val_acc: 0.9046\n",
      "Epoch 10/30\n",
      "38484/38484 [==============================] - 6s 157us/step - loss: 0.2320 - acc: 0.9378 - val_loss: 0.3961 - val_acc: 0.9065\n",
      "Epoch 11/30\n",
      "38484/38484 [==============================] - 6s 156us/step - loss: 0.1949 - acc: 0.9488 - val_loss: 0.4119 - val_acc: 0.8990\n",
      "Epoch 12/30\n",
      "38484/38484 [==============================] - 6s 157us/step - loss: 0.1751 - acc: 0.9536 - val_loss: 0.4122 - val_acc: 0.9025\n",
      "Epoch 13/30\n",
      "38484/38484 [==============================] - 6s 157us/step - loss: 0.1520 - acc: 0.9605 - val_loss: 0.4058 - val_acc: 0.9083\n",
      "Epoch 14/30\n",
      "38484/38484 [==============================] - 6s 158us/step - loss: 0.1382 - acc: 0.9620 - val_loss: 0.4094 - val_acc: 0.9081\n",
      "Epoch 15/30\n",
      "38484/38484 [==============================] - 6s 155us/step - loss: 0.1166 - acc: 0.9693 - val_loss: 0.4017 - val_acc: 0.9079\n",
      "Epoch 16/30\n",
      "38484/38484 [==============================] - 6s 159us/step - loss: 0.1024 - acc: 0.9733 - val_loss: 0.4003 - val_acc: 0.9074\n",
      "Epoch 17/30\n",
      "38484/38484 [==============================] - 6s 156us/step - loss: 0.0922 - acc: 0.9756 - val_loss: 0.3928 - val_acc: 0.9111\n",
      "Epoch 18/30\n",
      "38484/38484 [==============================] - 6s 156us/step - loss: 0.0905 - acc: 0.9754 - val_loss: 0.4561 - val_acc: 0.8990\n",
      "Epoch 19/30\n",
      "38484/38484 [==============================] - 6s 156us/step - loss: 0.0952 - acc: 0.9739 - val_loss: 0.4118 - val_acc: 0.9083\n",
      "Epoch 20/30\n",
      "38484/38484 [==============================] - 6s 155us/step - loss: 0.0896 - acc: 0.9757 - val_loss: 0.4178 - val_acc: 0.9137\n",
      "Epoch 21/30\n",
      "38484/38484 [==============================] - 6s 155us/step - loss: 0.0822 - acc: 0.9779 - val_loss: 0.4182 - val_acc: 0.9090\n",
      "Epoch 22/30\n",
      "38484/38484 [==============================] - 6s 156us/step - loss: 0.0718 - acc: 0.9804 - val_loss: 0.4280 - val_acc: 0.9107\n",
      "Epoch 23/30\n",
      "38484/38484 [==============================] - 6s 157us/step - loss: 0.0663 - acc: 0.9821 - val_loss: 0.4292 - val_acc: 0.9095\n",
      "Epoch 24/30\n",
      "38484/38484 [==============================] - 6s 160us/step - loss: 0.0609 - acc: 0.9829 - val_loss: 0.4268 - val_acc: 0.9088\n",
      "Epoch 25/30\n",
      "38484/38484 [==============================] - 6s 156us/step - loss: 0.0581 - acc: 0.9846 - val_loss: 0.4339 - val_acc: 0.9083\n",
      "Epoch 26/30\n",
      "38484/38484 [==============================] - 6s 155us/step - loss: 0.0526 - acc: 0.9857 - val_loss: 0.4244 - val_acc: 0.9123\n",
      "Epoch 27/30\n",
      "38484/38484 [==============================] - 6s 155us/step - loss: 0.0499 - acc: 0.9860 - val_loss: 0.4412 - val_acc: 0.9090\n",
      "Epoch 28/30\n",
      "38484/38484 [==============================] - 6s 157us/step - loss: 0.0570 - acc: 0.9837 - val_loss: 0.4589 - val_acc: 0.9051\n",
      "Epoch 29/30\n",
      "38484/38484 [==============================] - 6s 155us/step - loss: 0.0588 - acc: 0.9828 - val_loss: 0.4548 - val_acc: 0.9058\n",
      "Epoch 30/30\n",
      "38484/38484 [==============================] - 6s 156us/step - loss: 0.0574 - acc: 0.9834 - val_loss: 0.4487 - val_acc: 0.9076\n",
      "Training completed on 2020-04-23, 19-52\n"
     ]
    }
   ],
   "source": [
    "# Fit dataset\n",
    "tensorboard_callback6 = keras.callbacks.TensorBoard(os.path.join(model_dir+'logs','model6'))\n",
    "history6 = model6.fit(X_train, \n",
    "                      y_train, \n",
    "                      validation_data=(X_test, y_test), \n",
    "                      batch_size=batch_size, \n",
    "                      epochs=epochs, \n",
    "                      verbose=1, \n",
    "                      callbacks=[tensorboard_callback6])\n",
    "model6.save(model_dir+'model6.h5')\n",
    "print(f\"Training completed on {datetime.datetime.now().strftime('%Y-%m-%d, %H-%M')}\")\n",
    "\n",
    "# store logs to csv file\n",
    "df_history6 = pd.DataFrame(data=history6.history)\n",
    "df_history6.to_csv(model_logs+'logs_model6.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Metrics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 200,
   "metadata": {},
   "outputs": [],
   "source": [
    "# to load model use\n",
    "model6 = load_model(model_dir+'model6.h5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 201,
   "metadata": {},
   "outputs": [],
   "source": [
    "res = model6.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 202,
   "metadata": {},
   "outputs": [],
   "source": [
    "# convert predictions from one hot encoded to label value\n",
    "y_pred = convert_pred_list_ohe_to_labels(res, threshold=0.001, get_max=True)\n",
    "# convert test values from one hot encoded to label value\n",
    "y_true = convert_pred_list_ohe_to_labels(y_test, threshold=0.9, get_max=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Confusion Matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 203,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x1440 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Spectral Color Map used: Violet means 1, Red means -1 and Yellow means 0\n",
    "cfs6 = confusion_matrix(y_true, y_pred)\n",
    "plt.figure(figsize=(20,20))\n",
    "sn.heatmap(cfs6, cmap=\"Blues\", robust=True)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Multiclass Precision, Recall and Accuracy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 204,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.80      1.00      0.89        16\n",
      "           1       1.00      0.50      0.67         6\n",
      "           2       0.83      0.77      0.80        13\n",
      "           3       1.00      0.83      0.91         6\n",
      "           4       0.57      0.80      0.67         5\n",
      "           5       1.00      0.33      0.50         6\n",
      "           6       0.86      1.00      0.92        12\n",
      "           7       1.00      0.67      0.80         6\n",
      "           8       0.88      0.70      0.78        10\n",
      "           9       0.89      0.80      0.84        10\n",
      "          10       0.90      0.90      0.90        10\n",
      "          11       1.00      0.91      0.95        11\n",
      "          12       0.79      0.92      0.85        12\n",
      "          13       0.88      0.70      0.78        10\n",
      "          14       0.45      0.62      0.53         8\n",
      "          15       0.80      0.57      0.67         7\n",
      "          16       1.00      0.71      0.83         7\n",
      "          17       0.80      1.00      0.89         8\n",
      "          18       0.43      0.50      0.46         6\n",
      "          19       0.80      0.67      0.73         6\n",
      "          20       0.00      0.00      0.00         6\n",
      "          21       0.60      0.50      0.55         6\n",
      "          22       0.75      0.50      0.60         6\n",
      "          23       0.50      0.40      0.44         5\n",
      "          24       1.00      0.67      0.80         6\n",
      "          25       1.00      0.50      0.67         6\n",
      "          26       0.62      0.77      0.69        13\n",
      "          27       0.82      0.75      0.78        12\n",
      "          28       0.75      0.75      0.75        12\n",
      "          29       0.69      0.75      0.72        12\n",
      "          30       0.83      0.83      0.83         6\n",
      "          31       0.67      0.50      0.57         8\n",
      "          32       0.67      0.60      0.63        10\n",
      "          33       0.86      0.86      0.86         7\n",
      "          34       0.57      0.67      0.62        12\n",
      "          35       0.43      0.33      0.38         9\n",
      "          36       0.89      0.92      0.90       112\n",
      "          37       0.83      0.56      0.67         9\n",
      "          38       1.00      0.83      0.91         6\n",
      "          39       0.50      0.29      0.36         7\n",
      "          40       1.00      0.83      0.91         6\n",
      "          41       0.67      0.33      0.44         6\n",
      "          42       0.71      0.71      0.71         7\n",
      "          43       0.25      0.17      0.20         6\n",
      "          44       1.00      0.50      0.67         6\n",
      "          45       1.00      0.83      0.91         6\n",
      "          46       1.00      0.80      0.89         5\n",
      "          47       0.83      0.83      0.83         6\n",
      "          48       0.17      0.17      0.17         6\n",
      "          49       0.50      0.17      0.25         6\n",
      "          50       1.00      0.50      0.67         6\n",
      "          51       1.00      0.17      0.29         6\n",
      "          52       0.67      0.80      0.73         5\n",
      "          53       0.67      0.29      0.40         7\n",
      "          54       0.75      0.50      0.60         6\n",
      "          55       0.00      0.00      0.00         6\n",
      "          56       0.50      0.50      0.50         6\n",
      "          57       0.75      0.50      0.60         6\n",
      "          58       0.67      0.67      0.67         6\n",
      "          59       0.00      0.00      0.00         7\n",
      "          60       0.57      0.67      0.62         6\n",
      "          61       0.50      0.17      0.25         6\n",
      "          62       0.90      0.95      0.92       191\n",
      "          63       0.87      0.90      0.89       114\n",
      "          64       0.91      0.89      0.90       124\n",
      "          65       0.99      0.97      0.98       100\n",
      "          66       0.86      0.93      0.90        69\n",
      "          67       0.88      0.91      0.89        65\n",
      "          68       0.90      0.67      0.77        54\n",
      "          69       0.95      0.95      0.95       100\n",
      "          70       0.92      0.96      0.94       115\n",
      "          71       0.95      0.94      0.94        80\n",
      "          72       0.84      0.96      0.89        69\n",
      "          73       0.90      0.93      0.91        69\n",
      "          74       0.93      0.84      0.88        63\n",
      "          75       0.96      0.99      0.97        75\n",
      "          76       0.93      0.90      0.91       105\n",
      "          77       0.92      0.98      0.95        87\n",
      "          78       0.65      1.00      0.79        11\n",
      "          79       0.60      0.60      0.60        10\n",
      "          80       0.89      0.99      0.94        77\n",
      "          81       0.91      0.98      0.94        41\n",
      "          82       0.93      1.00      0.97       157\n",
      "          83       1.00      0.92      0.96        12\n",
      "          84       0.82      1.00      0.90         9\n",
      "          85       0.89      0.82      0.85        39\n",
      "          86       0.91      0.91      0.91       114\n",
      "          87       0.97      0.88      0.92        32\n",
      "          88       0.89      0.91      0.90       129\n",
      "          89       0.94      0.94      0.94       107\n",
      "          90       0.91      0.80      0.85        89\n",
      "          91       0.96      0.98      0.97        92\n",
      "          92       0.92      0.96      0.94       151\n",
      "          93       0.91      0.93      0.92        56\n",
      "          94       1.00      0.94      0.97        33\n",
      "          95       0.96      1.00      0.98       110\n",
      "          96       0.98      0.98      0.98       110\n",
      "          97       0.94      0.98      0.96       214\n",
      "          98       0.99      0.99      0.99       140\n",
      "          99       0.97      0.97      0.97        92\n",
      "         100       0.89      0.96      0.93        26\n",
      "         101       0.86      1.00      0.92        37\n",
      "         102       0.95      0.98      0.96       291\n",
      "         103       0.97      0.91      0.94        33\n",
      "         104       1.00      1.00      1.00         2\n",
      "         105       1.00      1.00      1.00         2\n",
      "         106       1.00      1.00      1.00         2\n",
      "\n",
      "    accuracy                           0.91      4276\n",
      "   macro avg       0.80      0.74      0.76      4276\n",
      "weighted avg       0.90      0.91      0.90      4276\n",
      "\n"
     ]
    }
   ],
   "source": [
    "cr6 = classification_report(y_true, y_pred, digits=2)\n",
    "print(cr6)\n",
    "cr6 = classification_report(y_true, y_pred, digits=4, output_dict=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 205,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Use this line before each model training to save memory\n",
    "del model6, history6, tensorboard_callback6, res"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## T Test (Comparison Test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(4276, 1, 32, 32)"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [],
   "source": [
    "sym2latex = {i.Index : i.latex for i in symbols.itertuples()}\n",
    "latex2sym = {v: k for k, v in sym2latex.items()}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "\"\\nms = [('model6.h5', 1)]\\nmodels = [(load_model(model_dir+i_[0]), i_[1]) for i_ in ms]\\n\""
      ]
     },
     "execution_count": 48,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "\n",
    "ms = [\n",
    "    ('model1.h5', 1),\n",
    "    ('model2.h5', 1),\n",
    "    ('model3.h5', 1),\n",
    "    ('model4.h5', 0),\n",
    "    ('model5.h5', 0),\n",
    "    ('model6.h5', 1),\n",
    "]\n",
    "models = [(load_model(model_dir+i_[0]), i_[1]) for i_ in ms]\n",
    "\"\"\"\n",
    "ms = [('model6.h5', 1)]\n",
    "models = [(load_model(model_dir+i_[0]), i_[1]) for i_ in ms]\n",
    "\"\"\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Predictng labels for Model 1\n",
      "Predictng labels for Model 2\n",
      "Predictng labels for Model 3\n",
      "Predictng labels for Model 4\n",
      "Predictng labels for Model 5\n",
      "Predictng labels for Model 6\n",
      "Predictions Done\n",
      "Comparing Results...\n",
      "Done\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>symbol</th>\n",
       "      <th>latex</th>\n",
       "      <th>model1</th>\n",
       "      <th>model2</th>\n",
       "      <th>model3</th>\n",
       "      <th>model4</th>\n",
       "      <th>model5</th>\n",
       "      <th>model6</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>80</td>\n",
       "      <td>\\subset</td>\n",
       "      <td>\\subset</td>\n",
       "      <td>\\subset</td>\n",
       "      <td>\\subset</td>\n",
       "      <td>\\subset</td>\n",
       "      <td>\\subset</td>\n",
       "      <td>\\subset</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>81</td>\n",
       "      <td>\\supset</td>\n",
       "      <td>\\supset</td>\n",
       "      <td>\\supset</td>\n",
       "      <td>\\supset</td>\n",
       "      <td>\\supset</td>\n",
       "      <td>\\supset</td>\n",
       "      <td>\\supset</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>74</td>\n",
       "      <td>\\omega</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>\\theta</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>63</td>\n",
       "      <td>\\gamma</td>\n",
       "      <td>\\gamma</td>\n",
       "      <td>\\gamma</td>\n",
       "      <td>\\gamma</td>\n",
       "      <td>\\gamma</td>\n",
       "      <td>\\gamma</td>\n",
       "      <td>\\gamma</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>101</td>\n",
       "      <td>\\|</td>\n",
       "      <td>\\|</td>\n",
       "      <td>\\|</td>\n",
       "      <td>\\|</td>\n",
       "      <td>\\|</td>\n",
       "      <td>\\|</td>\n",
       "      <td>\\|</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  symbol    latex   model1   model2   model3   model4   model5   model6\n",
       "0     80  \\subset  \\subset  \\subset  \\subset  \\subset  \\subset  \\subset\n",
       "1     81  \\supset  \\supset  \\supset  \\supset  \\supset  \\supset  \\supset\n",
       "2     74   \\omega        1        0        0        0   \\theta        0\n",
       "3     63   \\gamma   \\gamma   \\gamma   \\gamma   \\gamma   \\gamma   \\gamma\n",
       "4    101       \\|       \\|       \\|       \\|       \\|       \\|       \\|"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ttest = t_test(X_test, y_test, models=models, sym2latex_dict=sym2latex)\n",
    "ttest.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>latex</th>\n",
       "      <th>total_count</th>\n",
       "      <th>model1</th>\n",
       "      <th>model2</th>\n",
       "      <th>model3</th>\n",
       "      <th>model4</th>\n",
       "      <th>model5</th>\n",
       "      <th>model6</th>\n",
       "      <th>model1_acc</th>\n",
       "      <th>model2_acc</th>\n",
       "      <th>model3_acc</th>\n",
       "      <th>model4_acc</th>\n",
       "      <th>model5_acc</th>\n",
       "      <th>model6_acc</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>A</td>\n",
       "      <td>16</td>\n",
       "      <td>16</td>\n",
       "      <td>15</td>\n",
       "      <td>16</td>\n",
       "      <td>16</td>\n",
       "      <td>13</td>\n",
       "      <td>16</td>\n",
       "      <td>100.00</td>\n",
       "      <td>93.75</td>\n",
       "      <td>100.00</td>\n",
       "      <td>100.00</td>\n",
       "      <td>81.25</td>\n",
       "      <td>100.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>B</td>\n",
       "      <td>6</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "      <td>5</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>33.33</td>\n",
       "      <td>66.67</td>\n",
       "      <td>83.33</td>\n",
       "      <td>33.33</td>\n",
       "      <td>0.00</td>\n",
       "      <td>50.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>C</td>\n",
       "      <td>13</td>\n",
       "      <td>11</td>\n",
       "      <td>11</td>\n",
       "      <td>11</td>\n",
       "      <td>9</td>\n",
       "      <td>0</td>\n",
       "      <td>11</td>\n",
       "      <td>84.62</td>\n",
       "      <td>84.62</td>\n",
       "      <td>84.62</td>\n",
       "      <td>69.23</td>\n",
       "      <td>0.00</td>\n",
       "      <td>84.62</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>D</td>\n",
       "      <td>6</td>\n",
       "      <td>6</td>\n",
       "      <td>5</td>\n",
       "      <td>6</td>\n",
       "      <td>6</td>\n",
       "      <td>3</td>\n",
       "      <td>5</td>\n",
       "      <td>100.00</td>\n",
       "      <td>83.33</td>\n",
       "      <td>100.00</td>\n",
       "      <td>100.00</td>\n",
       "      <td>50.00</td>\n",
       "      <td>83.33</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>E</td>\n",
       "      <td>5</td>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>80.00</td>\n",
       "      <td>80.00</td>\n",
       "      <td>80.00</td>\n",
       "      <td>80.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>60.00</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  latex total_count model1 model2 model3 model4 model5 model6  model1_acc  \\\n",
       "0     A          16     16     15     16     16     13     16      100.00   \n",
       "1     B           6      2      4      5      2      0      3       33.33   \n",
       "2     C          13     11     11     11      9      0     11       84.62   \n",
       "3     D           6      6      5      6      6      3      5      100.00   \n",
       "4     E           5      4      4      4      4      0      3       80.00   \n",
       "\n",
       "   model2_acc  model3_acc  model4_acc  model5_acc  model6_acc  \n",
       "0       93.75      100.00      100.00       81.25      100.00  \n",
       "1       66.67       83.33       33.33        0.00       50.00  \n",
       "2       84.62       84.62       69.23        0.00       84.62  \n",
       "3       83.33      100.00      100.00       50.00       83.33  \n",
       "4       80.00       80.00       80.00        0.00       60.00  "
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "t_test_report = create_t_test_report(ttest)\n",
    "t_test_report.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Plots"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Train-Test Accuracy plots"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 305,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_history1 = pd.read_csv(model_logs+'logs_model1.csv').drop(['Unnamed: 0'], axis=1).reset_index(drop=True)\n",
    "df_history2 = pd.read_csv(model_logs+'logs_model2.csv').drop(['Unnamed: 0'], axis=1).reset_index(drop=True)\n",
    "df_history3 = pd.read_csv(model_logs+'logs_model3.csv').drop(['Unnamed: 0'], axis=1).reset_index(drop=True)\n",
    "df_history4 = pd.read_csv(model_logs+'logs_model4.csv').drop(['Unnamed: 0'], axis=1).reset_index(drop=True)\n",
    "df_history5 = pd.read_csv(model_logs+'logs_model5.csv').drop(['Unnamed: 0'], axis=1).reset_index(drop=True)\n",
    "df_history6 = pd.read_csv(model_logs+'logs_model6.csv').drop(['Unnamed: 0'], axis=1).reset_index(drop=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 337,
   "metadata": {},
   "outputs": [],
   "source": [
    "def draw_train_test_plot(df_, modelno=0):\n",
    "    plt.figure(figsize=(10, 5))\n",
    "    plt.plot(df_['acc'], label='Train Accuracy')\n",
    "    plt.plot(df_['val_acc'], label='Validation Accuracy')\n",
    "    plt.title(f'Train Validation Accuracy Plot for Model {modelno}')\n",
    "    plt.xlabel('Epoch')\n",
    "    plt.ylabel('Accuracy')\n",
    "    plt.ylim([0.5, 1.05])\n",
    "    plt.legend(loc='lower right')\n",
    "    plt.savefig(plot_dir+f'train_test_plot_model_{modelno}', bbox_inches='tight')\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 338,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "draw_train_test_plot(df_history1, 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 339,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "draw_train_test_plot(df_history2, 2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 340,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "draw_train_test_plot(df_history3, 3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 341,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "draw_train_test_plot(df_history4, 4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 342,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "draw_train_test_plot(df_history5, 5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 343,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "draw_train_test_plot(df_history6, 6)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Find the model which scored highest F1-score for each class"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 298,
   "metadata": {},
   "outputs": [],
   "source": [
    "# find model with highest f1 scores\n",
    "df_f1 =  pd.DataFrame(data = symbols['latex'])\n",
    "df_f1['model'] = [[-1] for i in range(len(df_f1))]\n",
    "df_f1['f1-score'] = [-1 for i in range(len(df_f1))]\n",
    "\n",
    "for i in range(len(df_f1)):\n",
    "    for m, j in enumerate([df_cr1, df_cr2, df_cr3, df_cr4, df_cr5, df_cr6]):\n",
    "        if j.loc[i, 'f1-score'] > df_f1.loc[i, 'f1-score']:\n",
    "            df_f1['model'][i] = [m+1]\n",
    "            df_f1.loc[i, 'f1-score'] = j.loc[i, 'f1-score']\n",
    "        elif j.loc[i, 'f1-score'] == df_f1.loc[i, 'f1-score']:\n",
    "            df_f1['model'][i].append(m+1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 299,
   "metadata": {},
   "outputs": [],
   "source": [
    "m_count = {k:0 for k in range(1, 7)}\n",
    "for i in df_f1['model']:\n",
    "    for j in i:\n",
    "        m_count[j] += 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 300,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{1: 19, 2: 36, 3: 67, 4: 18, 5: 1, 6: 11}\n"
     ]
    }
   ],
   "source": [
    "pprint(m_count)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Comparison of validation accuracy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "log_dir = model_dir + 'logs/'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_history1 = pd.read_csv(log_dir+'logs_model1.csv').drop('Unnamed: 0', axis=1).reset_index(drop=True)\n",
    "df_history2 = pd.read_csv(log_dir+'logs_model2.csv').drop('Unnamed: 0', axis=1).reset_index(drop=True)\n",
    "df_history3 = pd.read_csv(log_dir+'logs_model3.csv').drop('Unnamed: 0', axis=1).reset_index(drop=True)\n",
    "df_history4 = pd.read_csv(log_dir+'logs_model4.csv').drop('Unnamed: 0', axis=1).reset_index(drop=True)\n",
    "df_history5 = pd.read_csv(log_dir+'logs_model5.csv').drop('Unnamed: 0', axis=1).reset_index(drop=True)\n",
    "df_history6 = pd.read_csv(log_dir+\"logs_model6.csv\").drop('Unnamed: 0', axis=1).reset_index(drop=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(10, 5))\n",
    "plt.plot(df_history1['val_acc'], label='Model 1')\n",
    "plt.plot(df_history2['val_acc'], label='Model 2')\n",
    "plt.plot(df_history3['val_acc'], label='Model 3')\n",
    "plt.plot(df_history4['val_acc'], label='Model 4')\n",
    "plt.plot(df_history5['val_acc'], label='Model 5')\n",
    "plt.plot(df_history6['val_acc'], label='Model 6')\n",
    "plt.title('Comparison for Validation/Test Accuracy')\n",
    "plt.xlabel('Epoch')\n",
    "plt.ylabel('Accuracy')\n",
    "plt.ylim([0.5, 1.05])\n",
    "plt.legend(loc='lower right')\n",
    "plt.savefig(plot_dir+'val_acc_comparison_plot.png', bbox_inches='tight')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(10, 5))\n",
    "plt.plot(df_history1['acc'], label='Model 1')\n",
    "plt.plot(df_history2['acc'], label='Model 2')\n",
    "plt.plot(df_history3['acc'], label='Model 3')\n",
    "plt.plot(df_history4['acc'], label='Model 4')\n",
    "plt.plot(df_history5['acc'], label='Model 5')\n",
    "plt.plot(df_history6['acc'], label='Model 6')\n",
    "plt.title('Comparison between Train Accuracy')\n",
    "plt.xlabel('Epoch')\n",
    "plt.ylabel('Accuracy')\n",
    "plt.ylim([0.5, 1.05])\n",
    "plt.legend(loc='lower right')\n",
    "plt.savefig(plot_dir+'train_acc_comparison_plot.png', bbox_inches='tight')\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": "3.6.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}