NeuralNetwork.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "fce2f77d",
   "metadata": {},
   "source": [
    "# Neural Network from scratch  "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8e19c905",
   "metadata": {},
   "source": [
    "In this notebook, we will implement a fully parameterized neural network from scratch. We will be using only few libraries like numpy, pandas, matplotlib and sklearn for scale and show the accuracy (No API for neural network are used).\n",
    "\n",
    "We will fit and test this neural netwrok to classify images from FASHION-MNIST dataset."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "031120ac",
   "metadata": {},
   "source": [
    "## import libraries and load the dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "b17ebddd",
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.metrics import accuracy_score\n",
    "import random\n",
    "from statistics import mean"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "8e22958f",
   "metadata": {},
   "outputs": [],
   "source": [
    "path = \"./data/\" \n",
    "\n",
    "filename_read_train = os.path.join(path, \"fashion-mnist_train.csv\")\n",
    "filename_read_test = os.path.join(path, \"fashion-mnist_test.csv\")\n",
    "\n",
    "train_data = pd.read_csv(filename_read_train, na_values=['NA', '?'])\n",
    "test_data = pd.read_csv(filename_read_test, na_values=['NA', '?'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "9e21bea4",
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train = np.array(train_data.loc[:, train_data.columns!='label'])\n",
    "y_train = pd.get_dummies(train_data[\"label\"]).values.astype(np.float32)\n",
    "\n",
    "X_test = np.array(test_data.loc[:, test_data.columns!='label'])\n",
    "y_test = pd.get_dummies(test_data[\"label\"]).values.astype(np.float32)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "3d6de920",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(60000, 784) (60000, 10)\n",
      "(10000, 784) (10000, 10)\n"
     ]
    }
   ],
   "source": [
    "print(X_train.shape, y_train.shape)\n",
    "print(X_test.shape, y_test.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "ddd7cfd0",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Scale the data using MinMaxScaler\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "scaler = MinMaxScaler()\n",
    "X_train = scaler.fit_transform(X_train)\n",
    "X_test = scaler.fit_transform(X_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ab51ea12",
   "metadata": {},
   "source": [
    "## Implement NeuralNetwork Class"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "78ccd708",
   "metadata": {},
   "outputs": [],
   "source": [
    "class NeuralNetwork:\n",
    "    \n",
    "    # constructer takes as input: the network size(list of nodes for each layer), list of activation function for each layer\n",
    "    # learning rate\n",
    "    def __init__(self,\n",
    "                network_size,\n",
    "                activation_fct,\n",
    "                learning_rate=0.01):\n",
    "        self.network_size = network_size\n",
    "        self.activation_fct = activation_fct\n",
    "        self.learning_rate = learning_rate\n",
    "        # L: lenght of the network\n",
    "        self.L = len(network_size)\n",
    "        self.weight_params = {}\n",
    "        # list to keep track of costs\n",
    "        self.costs = []\n",
    "        # dic for store the value and derivatives and use it inside the class\n",
    "        self.cache = {}\n",
    "        # method to initialize the weigths and biases\n",
    "        self.initialize_weigths()\n",
    "        \n",
    "    \n",
    "    # initialize weights and biases\n",
    "    def initialize_weigths(self, init_method='xavier'):\n",
    "        if init_method == 'xavier':\n",
    "            for layer in range(self.L - 1):\n",
    "                self.weight_params[f'b{layer + 1}'] = np.ones((1, self.network_size[layer + 1]))\n",
    "                self.weight_params[f'w{layer + 1}'] = np.random.randn(self.network_size[layer], self.network_size[layer + 1]) / np.sqrt(self.network_size[layer])\n",
    "\n",
    "        elif init_method == 'He':\n",
    "            for layer in range(self.L - 1):\n",
    "                self.weight_params[f'b{layer + 1}'] = np.ones((1, self.network_size[layer + 1]))\n",
    "                self.weight_params[f'w{layer + 1}'] = np.random.randn(self.network_size[layer], self.network_size[layer + 1]) / np.sqrt(self.network_size[layer]/2)\n",
    "        \n",
    "        else:\n",
    "            for layer in range(self.L - 1):\n",
    "                self.weight_params[f'b{layer + 1}'] = np.ones((1, self.network_size[layer + 1]))\n",
    "                self.weight_params[f'w{layer + 1}'] = 2*np.random.random((self.network_size[layer], self.network_size[layer + 1])) - 1\n",
    "                \n",
    "    \n",
    "    # activation functions and there derivatives\n",
    "    def sigmoid(self, x):\n",
    "        return 1 / (1 + np.exp(-x))\n",
    "\n",
    "    def sigmoid_derivation(self, x):\n",
    "        return x * (1-x)\n",
    "\n",
    "    def ReLU(self, x):\n",
    "        return np.maximum(0,x)\n",
    "\n",
    "    def ReLU_derivation(self, x):\n",
    "        return 1.0*(x>0)\n",
    "\n",
    "    def softmax(self, s):\n",
    "        exps = np.exp(s - np.max(s, axis=1, keepdims=True))\n",
    "        return exps/np.sum(exps, axis=1, keepdims=True)\n",
    "    \n",
    "    # cross entropy and error to use it with softmax in output layer\n",
    "    def cross_entropy(self, pred, real):\n",
    "        n_samples = real.shape[0]\n",
    "        res = pred - real\n",
    "        return res/n_samples\n",
    "    \n",
    "    def error(self, yHat, y):\n",
    "        cost = -np.mean(y * np.log(yHat + 1e-8))\n",
    "        return cost\n",
    "    \n",
    "    # method takes as input X, y, batch_size and seeds, to return mini_batchs according to the size of batchs, and use the seeds\n",
    "    # to get every time diffrenet batchs\n",
    "    def get_minibatches_train(self, X, y, batch_size, seeds):\n",
    "        size_split = int(len(X)/batch_size)\n",
    "        mini_batches = []\n",
    "        indexes = np.arange(len(X))\n",
    "        np.random.seed(seeds)\n",
    "        np.random.shuffle(indexes)\n",
    "\n",
    "        for i in range(size_split):\n",
    "\n",
    "            X_mini = X[indexes[i * batch_size: (i+1) * batch_size]]\n",
    "            y_mini = y[indexes[i * batch_size: (i+1) * batch_size]]\n",
    "            mini_batch = (X_mini, y_mini)\n",
    "            mini_batches.append(mini_batch)\n",
    "\n",
    "        return mini_batches\n",
    "    \n",
    "    # Feedforward method, takes as input the values of X\n",
    "    def _feedforward(self, X):\n",
    "        \n",
    "        # initialize value small x as big X, so we can use it in the loop and update it to calculate z and a\n",
    "        x = X\n",
    "        \n",
    "        # test the loop and make sure it creates the correct value of z and a\n",
    "        \"\"\"\n",
    "        for layer in range(1, self.L - 1):\n",
    "            print(f'z{layer} = {x} * w{layer}')\n",
    "            if self.activation_fct[layer - 1] == 'ReLU':\n",
    "                print(f'a{layer} = ReLU(z{layer})')\n",
    "                \n",
    "            elif self.activation_fct[layer - 1] == 'sigmoid':\n",
    "                print(f'a{layer} = sigmoid(z{layer})')\n",
    "            \n",
    "            \n",
    "            x = f'a{layer}'\n",
    "                \n",
    "        \"\"\"\n",
    "        # Loop over the number of layer, except the last layer, because we're using softmax since it is a classification\n",
    "        for layer in range(1, self.L - 1):\n",
    "            # z = x.w + b\n",
    "            self.cache[f'z{layer}'] = np.dot(x, self.weight_params[f'w{layer}']) + self.weight_params[f'b{layer}']\n",
    "            \n",
    "            # check the activation function\n",
    "            # a = activation(z)\n",
    "            \n",
    "            if self.activation_fct[layer - 1] == 'ReLU':\n",
    "                self.cache[f'a{layer}'] = self.ReLU(self.cache[f'z{layer}'])\n",
    "                \n",
    "            elif self.activation_fct[layer - 1] == 'sigmoid':\n",
    "                self.cache[f'a{layer}'] = self.sigmoid(self.cache[f'z{layer}'])\n",
    "            \n",
    "            # update the value of x to be x = a\n",
    "            x = self.cache[f'a{layer}']\n",
    "            \n",
    "        # Now calculate the z and a to the last layer using softmax, so we will get propability distribution between 0 and 1\n",
    "        self.cache[f'z{self.L - 1}'] = np.dot(x, self.weight_params[f'w{self.L - 1}']) + self.weight_params[f'b{self.L - 1}']\n",
    "        self.cache[f'a{self.L - 1}'] = self.softmax(self.cache[f'z{self.L - 1}'])\n",
    "        \n",
    "        \n",
    "        # return the last value of a, so it's the output of softmax, in other words, yHat or predicted y\n",
    "        return self.cache[f'a{self.L - 1}']\n",
    "    \n",
    "    # backward function takes as inputs X and y\n",
    "    def _backward(self, X, y):\n",
    "        \n",
    "        # test the loop and make sure it creates the correct value of z_delta and a_delta\n",
    "        \"\"\"\n",
    "        for layer in range(self.L - 2, 0, -1):\n",
    "            print(f'z{layer}_delta = a{layer + 1}_delta * w{layer+1}.T')\n",
    "            \n",
    "            print(self.activation_fct[layer-1])\n",
    "            if self.activation_fct[layer-1] == 'ReLU':\n",
    "                print(f'a{layer}_delta = z{layer}_delta * dev_ReLU(a{layer})')\n",
    "                \n",
    "            elif self.activation_fct[layer-1] == 'sigmoid':\n",
    "                print(f'a{layer}_delta = z{layer}_delta * dev_sigmoid(a{layer})')\n",
    "        \n",
    "        \n",
    "        \"\"\"\n",
    "        #print(cache[f'a{self.L - 1}'])\n",
    "        #loss = self.error(self.cache[f'a{self.L - 1}'], y)\n",
    "        \n",
    "        self.cache[f'a{self.L - 1}_delta'] = self.cross_entropy(self.cache[f'a{self.L - 1}'], y)\n",
    "        self.cache[f'z{self.L - 1}_delta'] = self.cross_entropy(self.cache[f'z{self.L - 1}'], y)\n",
    "        \n",
    "        for layer in range(self.L - 2, 0, -1):\n",
    "            #print(layer)\n",
    "            self.cache[f'z{layer}_delta'] = np.dot( self.cache[f'a{layer + 1}_delta'], self.weight_params[f'w{layer+1}'].T)\n",
    "            \n",
    "            #print(self.activation_fct[layer - 1])\n",
    "            if self.activation_fct[layer - 1] == 'ReLU':\n",
    "                self.cache[f'a{layer}_delta'] = self.cache[f'z{layer}_delta'] * self.ReLU_derivation(self.cache[f'a{layer}'])\n",
    "                \n",
    "            elif self.activation_fct[layer - 1] == 'sigmoid':\n",
    "                self.cache[f'a{layer}_delta'] = self.cache[f'z{layer}_delta'] * self.sigmoid_derivation(self.cache[f'a{layer}'])\n",
    "\n",
    "        #print(self.cache.keys())\n",
    "        \n",
    "    # Method to fit the data, takes as input X, y, number of epochs, print epochs every loss_every, optimizer method, and batch_size\n",
    "    def fit(self, X, y, nb_epochs = 500, loss_every=50, optimizer='GD', stop=0.02,batch_size=None):\n",
    "        \n",
    "        # if batch_size not given, then it's a normal gradient descent\n",
    "        # if it is > len(X) so batch_size = len(X) and if < 0 fo it's = 1,(SGD)\n",
    "        if batch_size == None:\n",
    "            batch_size = len(X)\n",
    "        \n",
    "        if batch_size > len(X):\n",
    "            batch_size = len(X)\n",
    "        \n",
    "        if batch_size <= 1:\n",
    "            batch_size = 1\n",
    "        \n",
    "        if optimizer == 'GD':\n",
    "            for epoch in range(nb_epochs):\n",
    "                \n",
    "                mini_batches = self.get_minibatches_train(X, y, batch_size, epoch)\n",
    "                mini_batch_cost = []\n",
    "                \n",
    "                # for every batch in mini_batchs run forward and backward and update the weights and biases\n",
    "                for mini_batch in mini_batches:\n",
    "                    X_batch, y_batch = mini_batch\n",
    "                    yHat = self._feedforward(X_batch)\n",
    "                    self._backward(X_batch, y_batch) \n",
    "                    loss = self.error(yHat, y_batch)\n",
    "                    a = X_batch.T\n",
    "                    \n",
    "                    # m is the number of samples\n",
    "                    m = X_batch.shape[0]\n",
    "                    \n",
    "                    # int(len(self.weight_params)/2) is the number of w and b to update\n",
    "                    for layer in range(1, int(len(self.weight_params)/2)):\n",
    "                        \n",
    "                        self.weight_params[f'w{layer}'] -= self.learning_rate  * np.dot(a, self.cache[f'a{layer}_delta'])\n",
    "                        self.weight_params[f'b{layer}'] -= self.learning_rate  * np.sum(self.cache[f'z{layer}_delta'], axis = 0, keepdims = True)\n",
    "                        a = self.cache[f'a{layer}'].T\n",
    "                    \n",
    "                    # append the loss to the cost list and to mini_batchs_cost to stop it accourding to stoping creteria\n",
    "                    mini_batch_cost.append(loss)\n",
    "                    self.costs.append(loss)\n",
    "                \n",
    "                \n",
    "                #self.costs.append(mean(mini_batch_cost))\n",
    "                \n",
    "                # print loss \n",
    "                if epoch % loss_every == 0:\n",
    "                        #print(f'epoch {epoch}: |||| Loss = {round(mean(mini_batch_cost), 5)}')\n",
    "                        print(\"Epoch: {} - Loss:{:.4f}\".format(epoch, mean(mini_batch_cost)))\n",
    "                        print('=======================================================================')\n",
    "                \n",
    "                # check the loss and stop teh network if loss less then stop, default = 0.02\n",
    "                if mean(mini_batch_cost) < stop:\n",
    "                    break\n",
    "        \n",
    "        \"\"\"\n",
    "        for epoch in range(nb_epochs):\n",
    "            \n",
    "            yHat = self._feedforward(X)\n",
    "            #print(self.error(yHat, y))\n",
    "            self._backward(X, y) \n",
    "            \n",
    "            loss = self.error(yHat, y)\n",
    "            \n",
    "            a = X.T\n",
    "            for layer in range(1, len(self.weight_params)):\n",
    "                self.weight_params[f'w{layer}'] -= self.learning_rate * np.dot(a, self.cache[f'a{layer}_delta'])\n",
    "                a = self.cache[f'a{layer}'].T\n",
    "            \n",
    "            if epoch % loss_every == 0:\n",
    "                print(f'epoch {epoch}: {loss}')\n",
    "                \n",
    "            self.costs.append(loss)\n",
    "         \"\"\"       \n",
    "        \n",
    "    # predict method return yHat\n",
    "    def predict(self, X):\n",
    "        yHat = self._feedforward(X)\n",
    "        yHat = np.argmax(yHat, axis=1)\n",
    "        return yHat\n",
    "    \n",
    "    # method to check accuracy\n",
    "    def check_accuracy(self, y, yHat):\n",
    "        y_true = np.argmax(y, axis=1)\n",
    "        score = accuracy_score(y_true, yHat)\n",
    "        return score * 100\n",
    "    \n",
    "    # plot the costs\n",
    "    def plot_cost(self, title, xlabel='epochs'):\n",
    "        plt.figure()\n",
    "        plt.plot(np.arange(len(self.costs)), self.costs)\n",
    "        plt.xlabel(xlabel)\n",
    "        plt.ylabel(\"cost\")\n",
    "        plt.title(title)\n",
    "        plt.show()\n",
    "        \n",
    "    \n",
    "    def model_summary(self):\n",
    "        print(f\"Input features = {self.network_size[0]} \\nOutput labels = {self.network_size[-1]}\\n\")\n",
    "        for layer in range(len(self.activation_fct)):\n",
    "            \n",
    "            if layer == len(self.activation_fct):\n",
    "                print(f\"Output Layer {layer} \\nNumber of nodes = {self.network_size[layer + 1]} \\nActivation function: {self.activation_fct[layer]}\")\n",
    "            \n",
    "            else:\n",
    "                print(f\"Hidden Layer {layer} \\nNumber of nodes = {self.network_size[layer + 1]} \\nweight shape: {self.weight_params[f'w{layer + 1}'].shape} \\nBias Shape {self.weight_params[f'b{layer + 1}'].shape} \\nActivation function: {self.activation_fct[layer]}\\n\")\n",
    "    \n",
    "        \n",
    "    # method to print the images with the prediction and if its true or false\n",
    "    def plot_images_with_prediction(self, test_data, y_test, yHat):\n",
    "        images = test_data.drop(\"label\", axis=1).values\n",
    "        images = np.reshape(images,(10000,28,28))\n",
    "        y_real = np.argmax(y_test, axis=1)\n",
    "        label_names = ['T-shirt/top', 'Trouser', 'Pullover', 'Dress', 'Coat','Sandal', 'Shirt', 'Sneaker', 'Bag', 'Ankle boot']\n",
    "        ytest = test_data.label.values\n",
    "        plt.figure(figsize=(15,15))\n",
    "        random = random.randint(0, test_data.shape[0] - 26)\n",
    "        for i in range(random, random+24):\n",
    "            plt.subplot(5,5,i+1) \n",
    "            plt.xticks([]) \n",
    "            plt.yticks([]) \n",
    "            plt.grid(False) \n",
    "            \n",
    "            plt.imshow(images[i],cmap='gray')\n",
    "            if yHat[i] == y_real[i]: \n",
    "                plt.xlabel(f'TRUE: {label_names[yHat[i]]}', c='b')\n",
    "            else:\n",
    "                plt.xlabel(f'False: {label_names[ytest[i]]}-{label_names[yHat[i]]}', c='r')\n",
    "\n",
    "        plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "af5326d4",
   "metadata": {},
   "source": [
    "## Fit the Model with 1 hidden layer with 500 nodes and Sigmoid activation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "97a959a9",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Input features = 784 \n",
      "Output labels = 10\n",
      "\n",
      "Hidden Layer 0 \n",
      "Number of nodes = 500 \n",
      "weight shape: (784, 500) \n",
      "Bias Shape (1, 500) \n",
      "Activation function: sigmoid\n",
      "\n",
      "Hidden Layer 1 \n",
      "Number of nodes = 10 \n",
      "weight shape: (500, 10) \n",
      "Bias Shape (1, 10) \n",
      "Activation function: softmax\n",
      "\n"
     ]
    }
   ],
   "source": [
    "network_size = [784, 500, 10]\n",
    "activation_fct = ['sigmoid', 'softmax']\n",
    "nn0 = NeuralNetwork(network_size, activation_fct)\n",
    "nn0.model_summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "c7cf3352",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 0 - Loss:0.2404\n",
      "=======================================================================\n",
      "Epoch: 5 - Loss:0.2398\n",
      "=======================================================================\n",
      "Epoch: 10 - Loss:0.2392\n",
      "=======================================================================\n",
      "Epoch: 15 - Loss:0.2386\n",
      "=======================================================================\n",
      "Epoch: 20 - Loss:0.2380\n",
      "=======================================================================\n",
      "Epoch: 25 - Loss:0.2374\n",
      "=======================================================================\n",
      "Epoch: 30 - Loss:0.2368\n",
      "=======================================================================\n",
      "Epoch: 35 - Loss:0.2362\n",
      "=======================================================================\n",
      "Epoch: 40 - Loss:0.2356\n",
      "=======================================================================\n",
      "Epoch: 45 - Loss:0.2351\n",
      "=======================================================================\n",
      "Epoch: 50 - Loss:0.2345\n",
      "=======================================================================\n",
      "Epoch: 55 - Loss:0.2340\n",
      "=======================================================================\n",
      "Epoch: 60 - Loss:0.2335\n",
      "=======================================================================\n",
      "Epoch: 65 - Loss:0.2329\n",
      "=======================================================================\n",
      "Epoch: 70 - Loss:0.2324\n",
      "=======================================================================\n",
      "Epoch: 75 - Loss:0.2319\n",
      "=======================================================================\n",
      "Epoch: 80 - Loss:0.2314\n",
      "=======================================================================\n",
      "Epoch: 85 - Loss:0.2309\n",
      "=======================================================================\n",
      "Epoch: 90 - Loss:0.2304\n",
      "=======================================================================\n",
      "Epoch: 95 - Loss:0.2299\n",
      "=======================================================================\n"
     ]
    }
   ],
   "source": [
    "nn0.fit(X_train, y_train, nb_epochs = 100 ,loss_every=5, stop=0.025)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "33c2a07d",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy Score:  11.559999999999999 %\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "yHat = nn0.predict(X_test)\n",
    "print('Accuracy Score: ', nn0.check_accuracy(y_test, yHat), '%')\n",
    "nn0.plot_cost(title='Model with 1 hidden layer (500 nodes) and Sigmoid activation')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a23e56f3",
   "metadata": {},
   "source": [
    "## Fit the Model with 1 hidden layer with 500 nodes and ReLU activation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "64b17bdd",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Input features = 784 \n",
      "Output labels = 10\n",
      "\n",
      "Hidden Layer 0 \n",
      "Number of nodes = 500 \n",
      "weight shape: (784, 500) \n",
      "Bias Shape (1, 500) \n",
      "Activation function: ReLU\n",
      "\n",
      "Hidden Layer 1 \n",
      "Number of nodes = 10 \n",
      "weight shape: (500, 10) \n",
      "Bias Shape (1, 10) \n",
      "Activation function: softmax\n",
      "\n"
     ]
    }
   ],
   "source": [
    "network_size = [784, 500, 10]\n",
    "activation_fct = ['ReLU', 'softmax']\n",
    "nn1 = NeuralNetwork(network_size, activation_fct, learning_rate=0.1)\n",
    "nn1.model_summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "d7a3c1e0",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 0 - Loss:0.2623\n",
      "=======================================================================\n",
      "Epoch: 5 - Loss:0.1658\n",
      "=======================================================================\n",
      "Epoch: 10 - Loss:0.1351\n",
      "=======================================================================\n",
      "Epoch: 15 - Loss:0.1187\n",
      "=======================================================================\n",
      "Epoch: 20 - Loss:0.1086\n",
      "=======================================================================\n",
      "Epoch: 25 - Loss:0.1016\n",
      "=======================================================================\n",
      "Epoch: 30 - Loss:0.0964\n",
      "=======================================================================\n",
      "Epoch: 35 - Loss:0.0924\n",
      "=======================================================================\n",
      "Epoch: 40 - Loss:0.0892\n",
      "=======================================================================\n",
      "Epoch: 45 - Loss:0.0865\n",
      "=======================================================================\n",
      "Epoch: 50 - Loss:0.0842\n",
      "=======================================================================\n",
      "Epoch: 55 - Loss:0.0822\n",
      "=======================================================================\n",
      "Epoch: 60 - Loss:0.0804\n",
      "=======================================================================\n",
      "Epoch: 65 - Loss:0.0789\n",
      "=======================================================================\n",
      "Epoch: 70 - Loss:0.0775\n",
      "=======================================================================\n",
      "Epoch: 75 - Loss:0.0762\n",
      "=======================================================================\n",
      "Epoch: 80 - Loss:0.0751\n",
      "=======================================================================\n",
      "Epoch: 85 - Loss:0.0740\n",
      "=======================================================================\n",
      "Epoch: 90 - Loss:0.0731\n",
      "=======================================================================\n",
      "Epoch: 95 - Loss:0.0722\n",
      "=======================================================================\n"
     ]
    }
   ],
   "source": [
    "nn1.fit(X_train, y_train, nb_epochs = 100 ,loss_every=5, stop=0.025)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "a35f3992",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy Score:  77.64999999999999\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "yHat = nn1.predict(X_test)\n",
    "print('Accuracy Score: ', nn1.check_accuracy(y_test, yHat))\n",
    "nn1.plot_cost(title='Model with 1 hidden layer (500 nodes) and ReLU activation')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "56e3199f",
   "metadata": {},
   "source": [
    "Lets increase the number of epochs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "370fb6b2",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 0 - Loss:0.0714\n",
      "=======================================================================\n",
      "Epoch: 5 - Loss:0.0706\n",
      "=======================================================================\n",
      "Epoch: 10 - Loss:0.0699\n",
      "=======================================================================\n",
      "Epoch: 15 - Loss:0.0692\n",
      "=======================================================================\n",
      "Epoch: 20 - Loss:0.0685\n",
      "=======================================================================\n",
      "Epoch: 25 - Loss:0.0679\n",
      "=======================================================================\n",
      "Epoch: 30 - Loss:0.0674\n",
      "=======================================================================\n",
      "Epoch: 35 - Loss:0.0668\n",
      "=======================================================================\n",
      "Epoch: 40 - Loss:0.0663\n",
      "=======================================================================\n",
      "Epoch: 45 - Loss:0.0659\n",
      "=======================================================================\n",
      "Epoch: 50 - Loss:0.0654\n",
      "=======================================================================\n",
      "Epoch: 55 - Loss:0.0650\n",
      "=======================================================================\n",
      "Epoch: 60 - Loss:0.0645\n",
      "=======================================================================\n",
      "Epoch: 65 - Loss:0.0641\n",
      "=======================================================================\n",
      "Epoch: 70 - Loss:0.0637\n",
      "=======================================================================\n",
      "Epoch: 75 - Loss:0.0634\n",
      "=======================================================================\n",
      "Epoch: 80 - Loss:0.0630\n",
      "=======================================================================\n",
      "Epoch: 85 - Loss:0.0627\n",
      "=======================================================================\n",
      "Epoch: 90 - Loss:0.0623\n",
      "=======================================================================\n",
      "Epoch: 95 - Loss:0.0620\n",
      "=======================================================================\n",
      "Epoch: 100 - Loss:0.0617\n",
      "=======================================================================\n",
      "Epoch: 105 - Loss:0.0614\n",
      "=======================================================================\n",
      "Epoch: 110 - Loss:0.0611\n",
      "=======================================================================\n",
      "Epoch: 115 - Loss:0.0609\n",
      "=======================================================================\n",
      "Epoch: 120 - Loss:0.0606\n",
      "=======================================================================\n",
      "Epoch: 125 - Loss:0.0603\n",
      "=======================================================================\n",
      "Epoch: 130 - Loss:0.0601\n",
      "=======================================================================\n",
      "Epoch: 135 - Loss:0.0598\n",
      "=======================================================================\n",
      "Epoch: 140 - Loss:0.0596\n",
      "=======================================================================\n",
      "Epoch: 145 - Loss:0.0594\n",
      "=======================================================================\n",
      "Epoch: 150 - Loss:0.0592\n",
      "=======================================================================\n",
      "Epoch: 155 - Loss:0.0589\n",
      "=======================================================================\n",
      "Epoch: 160 - Loss:0.0587\n",
      "=======================================================================\n",
      "Epoch: 165 - Loss:0.0585\n",
      "=======================================================================\n",
      "Epoch: 170 - Loss:0.0583\n",
      "=======================================================================\n",
      "Epoch: 175 - Loss:0.0581\n",
      "=======================================================================\n",
      "Epoch: 180 - Loss:0.0580\n",
      "=======================================================================\n",
      "Epoch: 185 - Loss:0.0578\n",
      "=======================================================================\n",
      "Epoch: 190 - Loss:0.0576\n",
      "=======================================================================\n",
      "Epoch: 195 - Loss:0.0574\n",
      "=======================================================================\n",
      "Epoch: 200 - Loss:0.0573\n",
      "=======================================================================\n",
      "Epoch: 205 - Loss:0.0571\n",
      "=======================================================================\n",
      "Epoch: 210 - Loss:0.0569\n",
      "=======================================================================\n",
      "Epoch: 215 - Loss:0.0568\n",
      "=======================================================================\n",
      "Epoch: 220 - Loss:0.0566\n",
      "=======================================================================\n",
      "Epoch: 225 - Loss:0.0565\n",
      "=======================================================================\n",
      "Epoch: 230 - Loss:0.0563\n",
      "=======================================================================\n",
      "Epoch: 235 - Loss:0.0562\n",
      "=======================================================================\n",
      "Epoch: 240 - Loss:0.0560\n",
      "=======================================================================\n",
      "Epoch: 245 - Loss:0.0559\n",
      "=======================================================================\n",
      "Epoch: 250 - Loss:0.0558\n",
      "=======================================================================\n",
      "Epoch: 255 - Loss:0.0556\n",
      "=======================================================================\n",
      "Epoch: 260 - Loss:0.0555\n",
      "=======================================================================\n",
      "Epoch: 265 - Loss:0.0554\n",
      "=======================================================================\n",
      "Epoch: 270 - Loss:0.0553\n",
      "=======================================================================\n",
      "Epoch: 275 - Loss:0.0551\n",
      "=======================================================================\n",
      "Epoch: 280 - Loss:0.0550\n",
      "=======================================================================\n",
      "Epoch: 285 - Loss:0.0549\n",
      "=======================================================================\n",
      "Epoch: 290 - Loss:0.0548\n",
      "=======================================================================\n",
      "Epoch: 295 - Loss:0.0547\n",
      "=======================================================================\n",
      "Epoch: 300 - Loss:0.0546\n",
      "=======================================================================\n",
      "Epoch: 305 - Loss:0.0545\n",
      "=======================================================================\n",
      "Epoch: 310 - Loss:0.0543\n",
      "=======================================================================\n",
      "Epoch: 315 - Loss:0.0542\n",
      "=======================================================================\n",
      "Epoch: 320 - Loss:0.0541\n",
      "=======================================================================\n",
      "Epoch: 325 - Loss:0.0540\n",
      "=======================================================================\n",
      "Epoch: 330 - Loss:0.0539\n",
      "=======================================================================\n",
      "Epoch: 335 - Loss:0.0538\n",
      "=======================================================================\n",
      "Epoch: 340 - Loss:0.0537\n",
      "=======================================================================\n",
      "Epoch: 345 - Loss:0.0536\n",
      "=======================================================================\n",
      "Epoch: 350 - Loss:0.0535\n",
      "=======================================================================\n",
      "Epoch: 355 - Loss:0.0534\n",
      "=======================================================================\n",
      "Epoch: 360 - Loss:0.0534\n",
      "=======================================================================\n",
      "Epoch: 365 - Loss:0.0533\n",
      "=======================================================================\n",
      "Epoch: 370 - Loss:0.0532\n",
      "=======================================================================\n",
      "Epoch: 375 - Loss:0.0531\n",
      "=======================================================================\n",
      "Epoch: 380 - Loss:0.0530\n",
      "=======================================================================\n",
      "Epoch: 385 - Loss:0.0529\n",
      "=======================================================================\n",
      "Epoch: 390 - Loss:0.0528\n",
      "=======================================================================\n",
      "Epoch: 395 - Loss:0.0528\n",
      "=======================================================================\n",
      "Epoch: 400 - Loss:0.0527\n",
      "=======================================================================\n",
      "Epoch: 405 - Loss:0.0526\n",
      "=======================================================================\n",
      "Epoch: 410 - Loss:0.0525\n",
      "=======================================================================\n",
      "Epoch: 415 - Loss:0.0524\n",
      "=======================================================================\n",
      "Epoch: 420 - Loss:0.0524\n",
      "=======================================================================\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 425 - Loss:0.0523\n",
      "=======================================================================\n",
      "Epoch: 430 - Loss:0.0522\n",
      "=======================================================================\n",
      "Epoch: 435 - Loss:0.0521\n",
      "=======================================================================\n",
      "Epoch: 440 - Loss:0.0521\n",
      "=======================================================================\n",
      "Epoch: 445 - Loss:0.0520\n",
      "=======================================================================\n",
      "Epoch: 450 - Loss:0.0519\n",
      "=======================================================================\n",
      "Epoch: 455 - Loss:0.0518\n",
      "=======================================================================\n",
      "Epoch: 460 - Loss:0.0518\n",
      "=======================================================================\n",
      "Epoch: 465 - Loss:0.0517\n",
      "=======================================================================\n",
      "Epoch: 470 - Loss:0.0516\n",
      "=======================================================================\n",
      "Epoch: 475 - Loss:0.0516\n",
      "=======================================================================\n",
      "Epoch: 480 - Loss:0.0515\n",
      "=======================================================================\n",
      "Epoch: 485 - Loss:0.0514\n",
      "=======================================================================\n",
      "Epoch: 490 - Loss:0.0514\n",
      "=======================================================================\n",
      "Epoch: 495 - Loss:0.0513\n",
      "=======================================================================\n"
     ]
    }
   ],
   "source": [
    "nn1.fit(X_train, y_train, nb_epochs = 500 ,loss_every=5, stop=0.025)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "54d6e6bf",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy Score:  83.15\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "yHat = nn1.predict(X_test)\n",
    "print('Accuracy Score: ', nn1.check_accuracy(y_test, yHat))\n",
    "nn1.plot_cost(title='Model with 1 hidden layer (500 nodes) and ReLU activation')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "16666ba2",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x1080 with 25 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "nn1.plot_images_with_prediction(test_data, y_test, yHat)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9f1066a5",
   "metadata": {},
   "source": [
    "## Fit the Model with 1 hidden layer with 500, ReLU activation function and mini batch as oprimazation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "22dc8a8d",
   "metadata": {},
   "outputs": [],
   "source": [
    "network_size = [784, 500, 10]\n",
    "activation_fct = ['ReLU', 'softmax']\n",
    "nn2 = NeuralNetwork(network_size, activation_fct, learning_rate=0.1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "a54febb1",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 0 - Loss:0.0681\n",
      "=======================================================================\n",
      "Epoch: 50 - Loss:0.0318\n",
      "=======================================================================\n"
     ]
    }
   ],
   "source": [
    "nn2.fit(X_train, y_train, nb_epochs = 100, loss_every=50, batch_size = 100, stop=0.025)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "22b238e4",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy Score:  88.07000000000001\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "yHat = nn2.predict(X_test)\n",
    "print('Accuracy Score: ', nn2.check_accuracy(y_test, yHat))\n",
    "nn2.plot_cost(title='Model with 1 hidden layer (500 nodes) and ReLU activation and mini-batch')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d7b9665c",
   "metadata": {},
   "source": [
    "## Fit the Model with 1 hidden layer with 500 nodes, ReLU activation function with SGD"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "3c9c6378",
   "metadata": {},
   "outputs": [],
   "source": [
    "network_size = [784, 500, 10]\n",
    "activation_fct = ['ReLU', 'softmax']\n",
    "nn3 = NeuralNetwork(network_size, activation_fct)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "d389e35f",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 0 - Loss:0.0521\n",
      "=======================================================================\n",
      "Epoch: 1 - Loss:0.0415\n",
      "=======================================================================\n",
      "Epoch: 2 - Loss:0.0382\n",
      "=======================================================================\n",
      "Epoch: 3 - Loss:0.0361\n",
      "=======================================================================\n",
      "Epoch: 4 - Loss:0.0344\n",
      "=======================================================================\n",
      "Epoch: 5 - Loss:0.0332\n",
      "=======================================================================\n",
      "Epoch: 6 - Loss:0.0320\n",
      "=======================================================================\n",
      "Epoch: 7 - Loss:0.0312\n",
      "=======================================================================\n",
      "Epoch: 8 - Loss:0.0303\n",
      "=======================================================================\n",
      "Epoch: 9 - Loss:0.0296\n",
      "=======================================================================\n",
      "Epoch: 10 - Loss:0.0290\n",
      "=======================================================================\n",
      "Epoch: 11 - Loss:0.0282\n",
      "=======================================================================\n",
      "Epoch: 12 - Loss:0.0277\n",
      "=======================================================================\n",
      "Epoch: 13 - Loss:0.0272\n",
      "=======================================================================\n",
      "Epoch: 14 - Loss:0.0267\n",
      "=======================================================================\n",
      "Epoch: 15 - Loss:0.0263\n",
      "=======================================================================\n",
      "Epoch: 16 - Loss:0.0258\n",
      "=======================================================================\n",
      "Epoch: 17 - Loss:0.0255\n",
      "=======================================================================\n",
      "Epoch: 18 - Loss:0.0251\n",
      "=======================================================================\n",
      "Epoch: 19 - Loss:0.0248\n",
      "=======================================================================\n"
     ]
    }
   ],
   "source": [
    "nn3.fit(X_train, y_train, nb_epochs = 100, loss_every=1, batch_size = 1, stop=0.025)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "a9934daf",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy Score:  88.96\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAasAAAEWCAYAAADYRbjGAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjQuMywgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/MnkTPAAAACXBIWXMAAAsTAAALEwEAmpwYAAA160lEQVR4nO3dd5wdVf3/8deHREQEASX6U4pBaaICYsSuqPg1gIoFFVCwoIjt69fylXwVEQGVIojUUIx0AtIhIZESUkwC2UA66XVTN70sKbv5/P44czeTm1vm7m1zd9/Px2Mfj7135s6caecz58yZc8zdERERSbPd6p0AERGRYhSsREQk9RSsREQk9RSsREQk9RSsREQk9RSsREQk9eoSrMyst5m5mfVMMO93zGxUmev7uJnNqER6OrHugss2s9+a2W0Ffj/fzE7MM+0EM2uuVFqzlu1mdmg1ll0qM/uhmV1T73SUoxLHyszeYmavmNlrK5WuIuur2vmVRmbW38x+X6VlTzWzE6qx7EqpRF5bTUWDVZRZbjWz/bO+nxBlaL2rlroKcfeR7n5E5nOhAJCEmX3KzIaZ2Tozm19m2v7s7t8vZxldmZntDlwAXBn7zs1sk5ltjP5uy/rNL8xsWXR8BsQzdzN7o5k9Ev1+gZmdWbutKY+7LweGAefWOy2wy3FYbGZXm1mPhL/NeQ3mukkys4vM7O5KpTta5i4Zs7uf5+6XVGDZt5vZpVnLfre7P1/usuvJzM4xs+lmtsHMlpvZIDPbOza9j5k9aWZrzGytmU0zsz+Z2X7R9O+YWXvsup1nZv80s8OTrD9pyWoecEYsUe8FXlfCdnY1m4ABwP/WOyFdSZ6M7lRgursvzvr+GHffK/rrCPZm9jmgH/AZoDfwDuCPsd/dAGwF3gJ8E7jJzN5dua2ounuAH9Y7ETHHuPtewCeBbwDfq3N6pArM7JPAn4Ez3H1v4F3AA7HpHwGeB/4DHOnu+wJ9gTbgmNiixkTnyz7AicCrwHgze0/RRLh7wT9gPuHOdlzsu78CvwMc6B19tw9wJ9ACLIh+s1s0rUf0m5XAXOAn0W97xn77D2ApsBi4FOgRTfsOMCpP2u4AfhX9f0C0zB9Hnw8FVgMGnAA0R9/fBWyPdtJG4DeETM2BbwMLo3T+LsG+ORGYX2SegssGLgLujn0+K9p/q6J9PB84MZr2OuB2YA0wjRAsm2O/fRvwUHQM5gH/nbWeB6JjtAGYCvQpkG4HDo3+PwV4GVgPLAIuis03CPhZ1m8nAV+K/j8SeDo6FjOAr8fmux24CRhMuAE4MUc6BgAX5EtbjvnvBf4c+/wZYFn0/+sJgerw2PS7gMvyLOs7wCjCubsm2qcnZe3vx6Ntmw38IDatnGN1PNAU7e/lwNWxaT2BVuDtedJc6Fj1pvC5WDDNhc6R6PMDwA2xz58HJgBrgdHA0Vn5Sq7jvcuxJesayZr2TuA5wvWykhDM941NPwh4ONrPq4DrCRntZqCdkAesjZ2Pl0b/vwJ8Pmu/rwSOiz7/C1gGrANGAO+Ovj8X2BadZxuBJ7K3F3gtcA2wJPq7BnhtNO0EoBn4FbCCkCd+t8Ax+G6U1g2EvPWHsWkFlwW8iXD+rgdeBC4hf177a+DRAukYBVyXb3r8esrx/ZPAg4V+6+6Jg9WJhIzmXYTAswh4OzsHqzuBx4C9CRfFTOCcaNp5wPToxHkjoSojHqweBW4mZCZvjnbcDwttYDTte7GT4UxgDnB/bNpj8YOW70Jhx0V8K+GCPQbYAryryL4pJVjlXDaxCxE4inCCf4JwQl9NuDPJnOSXASOjfXgQMIUdQXg3YDxwIbA7oUQxF/hcbD2bgZOjY/gXYGySjCjaf++N1nE0IQPNBKOvAy/EfncMIVPYPTqeiwgXVE/gOMIFn7mwbydc7B+Nlr1HjnSMA76WI21LCJnFw0TnYDRtIvCN2Of9o/nfBLwPeDXHRfhEgYtrG/CDaJ/9KFqvRdOHAzcCewDHEjLEz1TgWI0Bzor+3wv4UFa6JgFfzJPmQseqN4XPxbxpTnCOHEnIDH8RfT6OkEF+MNp33yZcd5lMeT6VCVaHAp8lXC+9CIHjmmhaj+h8+BvhXNwD+Fi+fIWdg9WFwD2xaacQSvjxvGdvdgSeCbmWkyvPAS4GxhLyul6EQH5J7Pi1RfO8hnC9tgL75dn+UwgB2wil21Z2BNSCywIGEm4wXg+8h1BQyJfXfpxwg/9HwvX62ti01xMC/wlF8sJd9nlsXy4v9Fv30oLVBYQMri/hTrlndGL1jk6KLcBRsd/9EHg++v854LzYtP+KftuTUB2zBXhdbPoZwLBCGxhNeyfhrm03oH+0zkyGcAfwy9hBSxKsDox99yJwepF9U0qwyrlsdg5WFwIDs06Crew4yecCfWPTz41t7weBhVnr/j/gn7H1PBObdhRZGXexTCM27Rrgb9H/ryWULA6LPv8VuDH6/xvAyKzf3gz8IXZR31lk/82Kb3P03ScImfy+hDvlKey48ZmTtY9ew47z9ONEpazY9B8Qnad5Lq7Zsc97Rsv6f4TMvB3YOzb9L8DtFThWIwiZwv550vUf4Oxi126OY1XsXMyb5gLnyHpCqdiB+9gRjG4iyoBj888APpnrGix03lEgWOX4/ZeAl6P/P0y4geiZ59gWClaHEkore0af7wEuzLPOfaN075O9nNg8HdsbnaMnx6Z9jigfIeRVr8bTTAj6H8q17hxpeRT4ebFlEfLsbYQqu8y0P2fvk6xlnwQ8QchzNxJupnsAB0bbH1/WFdF8m4hqRnLt8+j7vsC2YttWSmvAuwill+8QSlFx+xMyjwWx7xYQquYgVHksypqW8XZChrI0eii3lpChvblYgtx9DmGnHUvIiJ4ElpjZEYS7jOHFN2sny2L/txLuaislybJ32k/uvolQSsk5nV3349sy+zDaj78l3AzkS8MeCVtkfjBqUNJiZusIJeX9ozRuIdydfcvMdiPcaNwVS9MHs9L0TUJmnxHfnlzWEO5gO7j7CHff6u5rgZ8DhxBK/RDOhzfEZs/8vyHHtMz0DQXW37HP3L01+ncvwrFY7e7x35Zyzhc6VucAhwPTzWycmX0+K017EzKCXRQ6Vrm2iZ3PxUJpzue46PffIATh18e28VdZ23hQtI5C2gn5QdxrCBnrLszszWY2MGrgsR64mx3bexCwwN3bEmzHTtx9NqF67QtmtifwRUIVM2bWw8wuM7M50TrnRz/L3s/5vI1d88r4flmVlea8eZGZnWRmY81sdbSPT85KR75l9SIUFhIfb3d/yt2/QCh5n0qIBd8nXKPbgbfG5v2Nh+dWj0TrKeQAwg1vQYmDlbsvINStn0yoeolbSTiZ3h777mBCsRJC9cBBWdMyFhFKVvu7+77R3xvcPelD7+HAacDuHh7CDwfOBvYj1Jfn3JyEy661nfZTdJG8Kd90dt2P82L7cF9339vdT65Auu4l1G0f5O77EEqxFpt+ByEIfQZodfcxsTQNz0rTXu7+o9hvix2LSYSMuxCPpWcqOz/QPYZQxbCKUDXd08wOy5o+tcjyc1kCvDHeGorSzvm8x8rdZ7n7GYQbtsuBB83s9QDRzcWhhOqtXIodq0IKpTkvDx4gVF9eGNvGP2Vt457ufl+RxS0klADjDiF/RvoXwvE/2t3fAHyLHdu7CDg4zw1ZkjzgPsLN16nAtCiAQbhpP5VQs7JPLL2Z9RZb9hJ2zSuXJEjPTqJWrg8RajPeEgWHwSQ73i2EKsLOHO/t7v4socbsPdFN9QvAV5KnfidfJlQ/F1Tqe1bnAJ+OEtfB3dsJd9d/MrO9zeztwC8JdzlE0/7bzA6MmjH2i/12KfBv4Coze4OZ7WZm74xanyQxHPgpoeoEQouUnxGKm+15frOc8JygU6I07kG44zMz2yNqYl2uB4HPm9nHouVdzM7H6AHg/8xsPzM7kLCdGS8C683sfDN7XXT39x4z+0AF0rU3oRSx2cyOJ1ysHaLgtB24ih2lKggl3cPN7Cwze0309wEzexfJDSaUkgEws3eb2bHR9u0VrXMx4S4YQqn/HDM7KjrXLiBUy2RKqg8DF5vZ683so4RMJ57mRNx9EeFZw1+i43804fq4J5ql08fKzL5lZr3cfTs7SlCZc/l4QpVRvsy74LEqolCak7gMONfM/h/hudh5UUnPov19SlZwf0207zJ/PYH7gQuivGI3C83bv0C4NvJt70ZgrZkdwM4tdF8kBODLovXvER1zCHnAgUWu24GERxY/IipVxda5hVDrsSeh+iyuWP5yX7SNvSy8EnQhO/LKUuxOqIZvAdrM7KQovUVFeePDwEVmtqeZHUV4rpiTmZ1qZqdH54ZF59YnCc/eIDRU+56Z9TOzN0e/OZBwo5FreT3M7BAzu45QXfnHXPPFlRSs3H2OuzflmfwzQv3kXELLkHsJLbkgnLhDCXeDL7Fryexswo6fRihSPkisSFnEcMLJkwlWowgn0Ii8vwh3YxdE1RO/TrieuE8Q6oIHE+5GXiUE3LK4+1RCS8l7CRfZGkJrnow/Eu4w50Xruyv223bCRX1sNH0lcBvhzq9cPyZk8BsIF9YDOea5k/Bgv+Oii6rI/gs4nR0NIi4nXGBJPQEcaWaZapK3EDK09YRzrTeh1da2aJ1DCPXlwwj7agHwh6xteR2h7v4+4EfRfu+MM6L1LyFUd/zB3Z+OppVzrPoCU81sI/B3wjOlzdG0bxJKS/kkOVb55E1zEu4+mXA9/m+UT/yA8ExxDaG15HeyfjKYcO1k/i4i3KCNJlzHawjH8pvuPqVAmo8jNNQZRCxvie3nQwkltmZCdSWEUsFUYJmZrcyzPUsJpcWPEM65jDsJ+2kxIc8am/XTfwBHRfnLozkWfSmhteckYDIhT7w0x3wFRdfXfxOO8RrCjcnjJSzip4QqwWWEG7p/Fph3DeF4ziJce3cDV7r7PVFaRgGfJuSNM6MqySGEwsN1seV8ODqv10fT3gB8IDp3Csq0ahIpi5mdDZzr7h+rwrLPJTTe+Z9KL7uRRHesw4H3xYKXSLegYCVli56tPUdoBZjd+EZEpGzqyFbKYqHHiBZCPf29RWYXEekUlaxERCT1VLISEZHUq/iQGLWy//77e+/eveudDBGRhjJ+/PiV7t6r3ukoVcMGq969e9PUlK8VvYiI5GJmSXomSR1VA4qISOpVPVhZGPxuhZnle6kvM98HLAzMdVq10yQiIo2lFiWr2wlv5OdlYdC9ywm9XIiIiOyk6sHK3UdQvEfdnxE6ZFxR7fSIiEjjqfszq6jzyS9TuL+zzLznmlmTmTW1tLRUP3EiIpIKdQ9WhMHhzi/QQ3oHd7/F3fu4e59evRqu5aWIiHRSGpqu9wEGmhmEQcNONrM2d3+0rqkSEZHUqHuwcveO8U7M7HbgyUYLVLNXbGDVxq188B1vKj6ziIiUrOrByszuIwyutb+ZNRPGFnoNgLsXfU7VCE68OgydNf+yU+qcEhGRrqnqwSoanjvpvN+pYlJERKRBpaGBhYiISEEKViIiknoKViIiknoKViIiknoKViIiknoKViIiknoKVnk80LSI3v0GsXlb0V6gRESkyureg0XarNq4hcGTl3Lj83MAWL1pK2/b93V1TpWISPemYJXl5wMnMGr2ynonQ0REYlQNmGVN69Z6J0FERLIoWImISOopWImISOopWCUwb+UmXpi7qt7JEBHpttTAIoFP/fV5QEOAiIjUi0pWIiKSegpWIiKSegpWIiKSegpWWcx2/ty85tX6JERERDooWBUxt2VjvZMgItLtKViJiEjqVT1YmdkAM1thZlPyTP+mmU2K/kab2THVTpOIiDSWWpSsbgf6Fpg+D/ikux8NXALcUoM0JeZVWOa29u08+8ryKixZRKRrqnqwcvcRwOoC00e7+5ro41jgwGqnqd6ufXYW59zRxIiZLfVOiohIQ0jbM6tzgKfyTTSzc82sycyaWloaN6NftLoVgFWbttQ5JSIijSE1wcrMPkUIVufnm8fdb3H3Pu7ep1evXrVLnIiI1FUq+gY0s6OB24CT3L2uPcYaVnwmERGpqbqXrMzsYOBh4Cx3n1n/9Oz82avRwiK27LUa7FFEpKhaNF2/DxgDHGFmzWZ2jpmdZ2bnRbNcCLwJuNHMJphZU7XTVMiWbdsTzXfXmPn86oGJZa3rtpHzOPbipzueYYmISG5VrwZ09zOKTP8+8P1qpyOpGcs3JJrv949NBeCqr3f+tbBpS9cDsGhNKwe9cc9OL0dEpKurezWgiIhIMQpWIiKSegpWIiKSegpWFXbT83Po3W8QG7e01TspIiJdhoJVhd374gIAVm9Uk3SRRtfWvp2z/vECTfPz9hgnNaJglQJ6EVkknZau28zIWSv5+cAJ9U5Kt6dgVYRXod91y3rzeMay9RVfh4hIV6JgVUQ1e7DIuOiJaQybsaL6KxIRaVAKVikxr2VTvZNQc+MXrGbx2lfrnQwRaQCp6Mi2K6pG9WFX89WbxgAw/7JT6pwSEUk7lawqTI0lREQqT8GqSmrxrEtEqit7FAapHwWrCtPJLSJSed02WM1cvoHe/QYxe8XGeielrl5Zup4Hxi2qdzJERArqtsHq8QlLAHhq8tKarnfo1GU88vLiXb6vV63hSX8fyW8emlSntYuIJNNtg1VSna3Wm7cyd1P0H941vlPLc3f1NygNzd25+umZLFjV/V7TkPIpWBVRrKFE+/bcM3z39nFlr3vIlGVs2LwNgLtfWMh7/jCUhas0qrA0puY1r3Lts7Mqcm3UmqvFVN0pWJXphmGzd/pcqfYVc1s2ct7d4/n1vyYC8PS05eH7ld37GVsjumvMfL5x85h6JyM1trZtr3cSEsvuGk3qRy8Fl6laDTRat7YDsGi1enjIWLVxC+te3cY7eu1V76SU5PePTa13EkQaXrcvWXWmcL+lrb3i6aiVa56ZmbP13+Zt6d+mT1wxjE9fNbzeyRCROqh6sDKzAWa2wsym5JluZnatmc02s0lmdly10xTW2/nftrXnD3G5qg2mLF5XNMDVqk78mmdm5Wz9N3buqpqsvxybtqY/oIpktLVvZ9CkpXreVSG1KFndDvQtMP0k4LDo71zgphqkKbFyT7Nl6zbz+etG8fWbx3L4BU9VJE2yq+nL1itTkFS5bdQ8fnLvSzw+cUm9k9IlVD1YufsIoNAwm6cCd3owFtjXzN5a7XRlVDt/Wx+15pu4aG1DPVhOs/btzqqNWzo+/2f2SvpeM5K7X1hYx1Sly2evHs41z8ysdzK6jM5kE8vXbwZgpUYNr4g0PLM6AIg/RGmOvtuFmZ1rZk1m1tTS0lLWSpPWArZEJ1yll5vP8vWbeWnhmjKX0rVdOXQG77/0mY6AlXmn7ZWlGsQyY9aKjVzzzKx6J2MnjVjwVVvA9EhDsMp1PuQ8rd39Fnfv4+59evXqVZGVFxvK49rnZhecXmmf/uvzfOXG0Xmn1/t6Hzx5KTc9P6euafj3tGUArGndVtd0SOd059bgqzZu4fIh0/O+nyn5pSFYNQMHxT4fCFS1ktfdWbKutBJTreRrRJCW6/vH97zE5UOmF5ynrX17t+1z0d05/ZYxPPvK8nonRVLod49M4abn5zBiVnk1Q91RGoLV48DZUavADwHr3L2qHfYNHLeIB8c3V3MVNbN5Wzs/vmc8i1aX17NFJV9+vGLoDE68eni37G1jS9t2xs5dzY/ueakqy2/ZsIXfPDhxp1cNbhg2W1WgkfkrN3HmrWPZlNKuyba2h+fW21WyKlktmq7fB4wBjjCzZjM7x8zOM7PzolkGA3OB2cCtwI+rnaZx8wq198hvwqK1LF2Xjpd0t293Wre28fyMFgZPXsYlT04ra3mVLLm9GO3flZu2FJkzt0Z+j62Smte0cs8LC3b67i+DX+GBpmYGTdpxP3fl0Bl88fpRFV33utZtfPWm0WXfBNXa5UOmM3rOKobPbOHVre1c/fTMVDRsun/cQn5SpRuY7qIWrQHPcPe3uvtr3P1Ad/+Hu/d39/7RdHf3n7j7O939ve7eVO00NS3oXAOGL93wHz78l+fYtLXAXVuFcv1iweeSQdM46sKhFc/Y17VuY8ridRVdZikmNa/liAuGpKYabeSsFqYuqcz+aNmwhenLkpeAvnnbC/zukSkdLUoL2Vbg3b9Sbd7Wzlf7j2b8gjXcNLy+zyd/cf8EevcbxOjZK0v+7XXPzeLaZ2dx/7jyW4l2pnFIZtRwd+f8hyYzqEIjPMRbwnYnaagGrLmFZd4tDp2yLO+0SpVQXihS+stUY26p0F3jyugCOP3WsXz+ulEsXNXKX4fOqPm7Sy9FNxIjZiar09+8rZ2HXqpele5Z/3iRU66tTKnlk1cOo+81IxPPv3pTaPJc6iHI9/B+bsvGRNVPVz89s+gzR3fnpufnsGRtdWsaMsPpDMzqdeWeFxbQu98gtrXnP/83bwvTyrlGCtWOb9rSxsVPTMvb+0uu305YtLbTaYHw8v77L32GIQXyoK6qWwaruM7kxV2xtvmXD0xk3spNHc8+zr2rieuHzWbuyk187m8jUvnOzhMTl3Dk74fw8sK1FV/2xEVrGTBqXkWX2VqBHjgmNK8tOs8Xrts1uM5YtoFPXzWcvzz1StHfr8vRynJr23ZOvX4UY+aEnk4WrGrl8iHT+cGdyStCirW8LcUVQ2YAVO3Z1MYtbXz6r88zcVH+UvXNw+cw4D/zuH30/MTLzdyAdNak6PiPX9C5RxmNTMGq3gko08oKVgksXrPjLjlezz9j+YZ0vbMTHbS/P1u9NJ16w3+4uMzngNUwt2XnsaBylXyn5WhskSkB3TqycwF44epWJjav43ePTgZge7TezgRgq0Hb1vbt3vGKQ2dMWLiWuSs3ccXQ/C1f26JSqpqh10a3D1ZLO1GN8cwrK6qQkoSyrovMHWbLxi0Vvduam2fwSIFt7dv53SOTO3ooyOfMW8fyvov/XZU0lJI9PjFxCf+IlRLntKTjtQJ358qh0yv2TDDujtHzaV5TuSrKZSV2DtAZa1u3ctHjU1PRICSNun2w+tf45o5GCvMTZtBJn6dAac+wCj0fKtay/OWFa/nqTTuPmbStfTuvNmjnr53JaGr1LtqImS3c88JCfvvw5ILzjZ6zKhUvLv/svpcZFWug8JmrhvNEFfura93aRu9+g7hr7IKC821rd24YNocvF3gJvrOq/SytGv48+BVuHz2fJyeFYzN7xcacIyR0V90+WEHIGNvat/NUHR9auu9o8l2upvmraWvfzlduHM27LhxSkWXWSiZcPzu9jqXXhBq58iffe1kTF63tqOLrrJUbwnOZW0Yka0moDoiDzGgOmd3R95oRHSMkrN+8rdt3w6bBFwl3mgDn9z2y7GWV83LtxlIeFue5vl9euIbT+o/hp586lMl1bIIu5duwOf/5kC+DL6eKb/yCNXz1ptJKOfNWbsLdqz6ibtLFZ5rwV6qHmnr1HLOlrb3jmRjA9/45jqYFa/jfzx1RpxTVn0pWFZbd5Hfc/OrcDW3IE9hWbAgNLmYs39Dx3fCZLYyaVfw9lTT02ba0Dt1gzV6xkfPuGs+Nz9e2H8hKyI5Z5bSOW9bJfd+6tZ0nJi7hgabaV1llb//iClf/zWnp/LPbzOXUmYLjERfsXCMyqXldtKzuWwpVsKqiyc3r+O0jhZ9r1MK3B7zIt/7xQsWXu3jtq7z3D0M7nuGNmbOKk/4+sqwHxLeMmFup5CV24tXDGTJ1WUdjlWIyN7zPpaiqstY3Gs9kvbT9s/te5jcP7jqoZ1wmn124upWbC7xsfPyfnqF3v0E7vfCevXn5trdSmXkl9mdmGZVsst+NY5WCVSVl94VXyWblafTRy55jw5Y2zh7wIis2bOaCRyfzytL1HS9dp6CgltORv3+KvteM6PTvLx1UvEl7JVt05co4y82zysmMt7S18+fBuZt0J705u7bAaweZ2oHNW3fsw2pXM2ar9tpydVhdyjGt9f5IAwWrCur7985ngPmUek5OSvDSKOxaXZlEofdJVqzf0lFlknn29lIVXtaFzjWr37ilreNl183btjN92YYiv9hZvOeHBSnsoLeWd9yF1nXvCwsrWpIoxeevG1nycS3FmjJf6I1bUaApfDeMQ4koWMWUe5GV00OB4zkzgalL1u/UqqpYGm8YFuYtlnnl6uUgZ7piy5m/Kn+Q+HyO5V3y5DTaCnSHkzFl8bqOnhGq5QOXPsMxZbzzdNETUxPNV89qmqQ3KqUoJ980jOnL1tO73yAWFDh3kuyzJOmYsjh3C8dyuzjKeN8lT1e1SXx3fh6VhIJVA/jz4OkVr5Z4NU9/ZvXw+etGccatY6u6jlK295lpu3aie3eRd4aqqdCxj2dvNw6rZqezyTPSvz29o2uuh6I+LIdO3fW1kB3PdEKpZW1rdYZ/f3LS0ooFmVwNgIoFmX9PLa1TZpWsclOwahDDZlR/sLZi18jiCvYIkMuGAr2LF+qwNKNSF/n3S+jvrjP6XPpMVZY7ZOoyNhZo7h7X2S6Pkvzq0QnJXjiOL+t9lzzNsRc/XXJ6kobQpK0kN29rD9XdeTZ0S1s7KzaEgJX0fOvsKA+yMwWrBvPf902o27q/f0d1M/GLHs/fcGHW8nR0EVQJKzduoXe/QWV3j9W8pnWXqtn2hFVJlbx7z7fGpOvoTN961Sp8HPn7IfzygQl5p//gjiaO/9OzVVq7FKJgVU0lXFFt25M9MSvpxeEK25qgdFPMtCXr8w5TkWTcpq5k8OQyekxx+NjlwzpeaO+sJWtfrdl+jx/1TVvbmVdKQ5kaVo09NmFJ3pJnrlZ8L+Z5l7LUFntJrv/u/FRLwaqaSjizrhgyg1tHVvIdo+Ir36XaLce1deLV5WWGcRMWreXka0fWfUC/RpPp3zHeJ2WlWtx95LLn+NzfKt+KNYlShnZpqtLL9b95cGLZI/iOmNnCyFm7VtN3Nr7+4v6JeadlCs7d8bGWglUV5RqqoZBK9Q2YT3bHou+9qDo9gueTeeZVjV62q60ad7T/GDWP52fs+mLxqo1beO8fhnZ8Pv7Pz3LFkOmcPeDFKqSicK8haXnYv3B1a8Va9cU90NRckRF8a97zSkqOSy0pWMU8X+FGDFcOTdYjQq38/tEpBadXc5whd+e2UTtKjof+dnBJv0/SUqwW4yRV2h+f2PU53Zi5q3bpTuuxhI0Wkkq6p+a2bKJ3v0HMjD0zTNrCOr4Oz99mIbGVG7r2S/ZSWE2ClZn1NbMZZjbbzPrlmL6PmT1hZhPNbKqZfbcW6cpW7ZJNd5Spt5+/qnWnap+2HM+t8mWC/5m9kjNvq3x3UZUwYmZLqkZt3a2MolCun46O3n/LDClS6nOYsnpaSPjTR19enPN1g53TUeKq63Dfo9esCqt6r+tm1gO4Afgs0AyMM7PH3T1+S/kTYJq7f8HMegEzzOwed6/4ixdL1zXeODfZkrTKK+nhdQ10psXXSwvXcNzB+1Wl+iejpcy79UzV3PzLTilrOYMmLeX9b9+P/7fPHmUtJ2keu2LDFo6NvSQ9t4ze2rfkeYdt/qrWwn09ViAgtG5t43/un1D+gqjtiL/ZgenEq4d3qleZ7qQWJavjgdnuPjcKPgOBU7PmcWBvC7dhewGrgao0e6v0CVnu2D+dkd2JaC7l9BadFl+5cTSTmzv3fOupyUv5ZYJMLMn7W9XW1r6dn9z7El+7OfnwHOWedv+etpy1sYEhP11Gq8Kz/pHsWVp2f4BFY1XWNv5r/CKue3YWzWtaO0psJ1z5fLJEJvDoy4uTp63CFKiKq8V4VgcA8bEDmoEPZs1zPfA4sATYG/iGu9c/F0lgfcKXMBtBNfp0W9u6lc1ldOramc6A12/exo+iFl5Xf+PYTq+7Ftx37PWla8t/SJ+0UU8lM+Ok63w4Fgyg9CrCoVOXM3Tqcq56eiZvfP3uQO7q5M5qrXSvLlWIePXqdzENalGyynXIsvf454AJwNuAY4HrzewNuyzI7FwzazKzppaW6vfokEQpQ9yn3bh5lW8e/PErhvHRy56r+HJzyeR9R1ehlWMtC9DlNBS5fljjjckVN6ULDRha0QY/jdd2qOJqEayagYNinw8klKDivgs87MFsYB6wy7C97n6Lu/dx9z69evWqWoK7q/btlS/M5urcN3/GX/u7xtatbQU/x71cx2HFswshIxMMplmVdHTyd0mP7CVPFh9+BWB1BXtAzyVJqS8ejMbNW83/DHy5hOV3KlnduhFGLYLVOOAwMzvEzHYHTidU+cUtBD4DYGZvAY4Aaj8Kn1RNZy/OqUvWJX4F4M4xxTubzc7k/vD4zr2pH3XhUPL58o2juejx/L2vv7RwDRc+Vvj1gEIyVVpJ9lW+d4PSegPe1TPZf41v5tEJSwr2b1mWrP3XiK9plKvqz6zcvc3MfgoMBXoAA9x9qpmdF03vD1wC3G5mkwnX2/nuXp9bR0mVv/57ZvGZYoqNOXTcJTt3lrpqY2l36LePnp932ldvGl33TLk7DsqXTG32S/wR2rpX8weuTVvamb1iA4e+ee+Slt+dD28tGljg7oOBwVnf9Y/9vwT4r1qkReojyTVWiYy+lAYZ1RqSohTZmzx7xUaa1+w6uGPSTKoWeVmlHvLvVoeM99f/msjCVa08cN6HCw7t0dmgcNvIuSxfv5krTjuG+15cmHe+vz0zk789M7PsVx66k5oEK2kQVbxtKzRwYyU1lzBu0df6j6FHhXLMfg9NqshyKtkXY6meLvBibebUmLViI6+U2I1YPvPrMOLyg9H4Wi0btjB9WWW2I+6650IDlytOO6biy+7u1N2SdHi1QOOCco2evWMk4Gq+m/bdf45LPO+sCr7bMnDcouIz5VDvasO4R7KalsdtiL2i8dWbxlRkfdnvXe2iko3pspZ1wpXD2FrGKxXVHDFYclOwkg63jpxXtWXHM4uheUZOTVG+3SnVDDz1fqBe6PlLEml7Pyh7qI9ns160L7a3P1Kj1zFkh5KClZl9Lcl3Io0iTSWbQhI/s0o4Y1of1NfrcJxT5YFF85m3chP3vFC4FeuAUfM6xpJrlPO1GkotWf1fwu9EJIFKlziSBqFSM72UxrbEbk7pGGqnXj+K3z1S+HWHi2PvnmUOW1pvNqopUQMLMzsJOBk4wMyujU16A1Xqw69a1LRXupNyO+rtUOZ1s2J9wnTEgujAAq3pSvVAU3PRRg/xLRwzZ1Xe+Sqps921dcdcLGlrwCVAE/BFYHzs+w3ALyqdKOl6yn3m0d0tSzi4X3Ym9o1bcjeGWFXlHiCyFXo/LZ/7O9loJZ/sTouXFwigVz1d2vt9Un2JgpW7TwQmmtm97r4NwMz2Aw5y9/r1QSN1c/+40u56H2hqLjpPofdeurstnWy5NrdCve9nRnluZCf9feROn3/7yOQ6pUQ6o9RnVk+b2RvM7I3AROCfZnZ1FdIlKXf+Q41zoQ+ZsqzeSaiZatVyd6b3+7Sp9DAceqJQW6UGq33cfT3wFeCf7v5+4MTKJ0u6o+FV6MHe3bli6PT80+vcpFqFycaVtgFOu7pSg1VPM3sr8HXgySqkR7qxagzUOnrOqoJVYTOXa9C7NInfPMxNeTB4clLuzoSr6ZkCvYx0daUGq4sJHdLOcfdxZvYOoMhr6CL189z0FfVOQk3V++XhRjasAc6VpANddkUl9Q3o7v8C/hX7PBf4aqUTJVIpo+o07lMpKtXXHsCy9eWPNtxdPdsAwao7K7UHiwPN7BEzW2Fmy83sITM7sFqJE+nq3OGL1/+n3slIjQ2dfO+ou+mOjTtKrQb8J2HgxLcBBwBPRN+JiJRt+rIN9U6CpFSpwaqXu//T3duiv9sBjS8vqTVjebozv8XqvVskkVKD1Uoz+5aZ9Yj+vgXUpl8SERHptkoNVt8jNFtfBiwFTgO+W+lEiYiIxJU6UvAlwLczXSxFPVn8lRDEGkI3fC4pIjWW3Q+hlK/UktXR8b4A3X018L7KJklEpLF9+C/PVnX53fF9ulKD1W5RB7ZAR8mqaOnMzPqa2Qwzm21m/fLMc4KZTTCzqWY2vMR0iYikxsqNte3VvjsotRrwKmC0mT1IGHnm68CfCv3AzHoANwCfBZqBcWb2uLtPi82zL3Aj0NfdF5rZm0tMl4hIt9Ed37MqtQeLO82sCfg04fHPV+JBJ4/jgdlRbxeY2UDgVCD+uzOBh919YbQevUouIiIdSi1ZEQWnYgEq7gAgPopaM/DBrHkOB15jZs8DewN/d/c7sxdkZucC5wIcfPDBJSRBREQaWanPrDojV4E1u3/tnsD7gVOAzwG/N7PDd/mR+y3u3sfd+/TqpXeRRUS6i5JLVp3QDBwU+3wgsCTHPCvdfROwycxGAMcAGltaRERqUrIaBxxmZoeY2e7A6YT+BeMeAz5uZj3NbE9CNeErNUibiIg0gKqXrNy9zcx+ShgHqwcwwN2nmtl50fT+7v6KmQ0BJgHbgdvcfUq10yYiIo2hFtWAuPtgYHDWd/2zPl8JXFmL9IiINLIRM1v41X8dUe9k1FQtqgFFRKSCJjavq3cSak7BSkREUq/bBavsNvMiIpJ+3S5YDRg1r95JEBGREnW7YLVwdWu9kyAiIiXqdsHKXRWBIiKNptsFq2deUR+5IiKNptsFKxERaTwKViIiknoKViIiknoKViIiknoKViIiknoKViIiknoKViIiknoKViIiknoKViIiknoKViIiknoKViIiknoKViIiknoKViIikno1CVZm1tfMZpjZbDPrV2C+D5hZu5mdVot0iYhIY6h6sDKzHsANwEnAUcAZZnZUnvkuB4ZWO00iItJYalGyOh6Y7e5z3X0rMBA4Ncd8PwMeAjTglIiI7KQWweoAYFHsc3P0XQczOwD4MtC/0ILM7FwzazKzppaWloonVERE0qkWwcpyfJc9tvw1wPnu3l5oQe5+i7v3cfc+vXr1qlT6REQk5XrWYB3NwEGxzwcCS7Lm6QMMNDOA/YGTzazN3R+tQfpERCTlahGsxgGHmdkhwGLgdODM+AzufkjmfzO7HXhSgUpERDKqHqzcvc3Mfkpo5dcDGODuU83svGh6wedUIiIitShZ4e6DgcFZ3+UMUu7+nVqkSUREGod6sBARkdRTsBIRkdRTsBIRkdRTsBIRkdRTsBIRkdRTsBIRkdRTsBIRkdRTsBIRkdRTsBIRkdRTsBIRkdRTsBIRkdRTsBIRkdRTsBIRkdRTsBIRkdRTsBIRkdRTsBIRkdRTsBIRkdRTsBIRkdRTsBIRkdRTsBIRkdSrSbAys75mNsPMZptZvxzTv2lmk6K/0WZ2TC3SJSIijaHqwcrMegA3ACcBRwFnmNlRWbPNAz7p7kcDlwC3VDtdIiLSOGpRsjoemO3uc919KzAQODU+g7uPdvc10cexwIE1SJeIiDSIWgSrA4BFsc/N0Xf5nAM8lWuCmZ1rZk1m1tTS0lLBJIqISJrVIlhZju8854xmnyIEq/NzTXf3W9y9j7v36dWrVwWTKCIiadazButoBg6KfT4QWJI9k5kdDdwGnOTuq2qQLhERaRC1KFmNAw4zs0PMbHfgdODx+AxmdjDwMHCWu8+sQZpERKSBVL1k5e5tZvZTYCjQAxjg7lPN7Lxoen/gQuBNwI1mBtDm7n2qnTYREWkMtagGxN0HA4Ozvusf+//7wPdrkRYREWk86sFCRERST8FKRERST8FKRERST8FKRERST8FKRERST8FKRERST8FKRERST8FKRERST8FKRERST8FKRERST8FKRERST8FKRERST8FKRERST8FKRERST8FKRERST8FKRERST8FKRERST8FKRERST8FKRERST8FKRERSrybBysz6mtkMM5ttZv1yTDczuzaaPsnMjqtFukREpDFUPViZWQ/gBuAk4CjgDDM7Kmu2k4DDor9zgZuqnS4REWkctShZHQ/Mdve57r4VGAicmjXPqcCdHowF9jWzt1YjMR899E3VWKyIiFRRLYLVAcCi2Ofm6LtS58HMzjWzJjNramlp6VRi7vzeBzv1OxGRtPjlZw+vdxJqrmcN1mE5vvNOzIO73wLcAtCnT59dpifRYzdj/mWndOanIiJSJ7UoWTUDB8U+Hwgs6cQ8IiLSTdUiWI0DDjOzQ8xsd+B04PGseR4Hzo5aBX4IWOfuS2uQNhERaQBVrwZ09zYz+ykwFOgBDHD3qWZ2XjS9PzAYOBmYDbQC3612ukREpHHU4pkV7j6YEJDi3/WP/e/AT2qRFhERaTzqwUJERFJPwUpERFJPwUpERFJPwUpERFLPQtuGxmNmLcCCTv58f2BlBZNTT9qWdOoq29JVtgO0LRlvd/delUxMLTRssCqHmTW5e596p6MStC3p1FW2patsB2hbGp2qAUVEJPUUrEREJPW6a7C6pd4JqCBtSzp1lW3pKtsB2paG1i2fWYmISGPpriUrERFpIApWIiKSel06WJlZXzObYWazzaxfjulmZtdG0yeZ2XH1SGcSCbblm9E2TDKz0WZ2TD3SWUyx7YjN9wEzazez02qZvlIk2RYzO8HMJpjZVDMbXus0JpXg/NrHzJ4ws4nRtqRyZAQzG2BmK8xsSp7pjXTNF9uWhrjmK8bdu+QfYTiSOcA7gN2BicBRWfOcDDxFGKn4Q8AL9U53GdvyEWC/6P+T0rgtSbYjNt9zhJ76T6t3uss4JvsC04CDo89vrne6y9iW3wKXR//3AlYDu9c77Tm25RPAccCUPNMb4ppPuC2pv+Yr+deVS1bHA7Pdfa67bwUGAqdmzXMqcKcHY4F9zeyttU5oAkW3xd1Hu/ua6ONYwmjLaZPkmAD8DHgIWFHLxJUoybacCTzs7gsB3D2t25NkWxzY28wM2IsQrNpqm8zi3H0EIW35NMo1X3RbGuSar5iuHKwOABbFPjdH35U6TxqUms5zCHePaVN0O8zsAODLQH/SLckxORzYz8yeN7PxZnZ2zVJXmiTbcj3wLmAJMBn4ubtvr03yKqpRrvlSpfWar5iaDL5YJ5bju+x2+knmSYPE6TSzTxFO3I9VNUWdk2Q7rgHOd/f2cBOfWkm2pSfwfuAzwOuAMWY21t1nVjtxJUqyLZ8DJgCfBt4JPG1mI919fZXTVmmNcs0nlvJrvmK6crBqBg6KfT6QcFdY6jxpkCidZnY0cBtwkruvqlHaSpFkO/oAA6NAtT9wspm1ufujNUlhcknPr5XuvgnYZGYjgGOAtAWrJNvyXeAyDw9IZpvZPOBI4MXaJLFiGuWaT6QBrvmK6crVgOOAw8zsEDPbHTgdeDxrnseBs6MWQh8C1rn70lonNIGi22JmBwMPA2el8M49o+h2uPsh7t7b3XsDDwI/TmGggmTn12PAx82sp5ntCXwQeKXG6UwiybYsJJQQMbO3AEcAc2uayspolGu+qAa55iumy5as3L3NzH4KDCW0dhrg7lPN7Lxoen9Ca7OTgdlAK+HuMXUSbsuFwJuAG6NSSZunrFfmhNvREJJsi7u/YmZDgEnAduA2d8/ZDLmeEh6XS4DbzWwyoSrtfHdP3XAbZnYfcAKwv5k1A38AXgONdc1Dom1J/TVfSepuSUREUq8rVwOKiEgXoWAlIiKpp2AlIiKpp2AlIiKpp2AlItIAinVsm2P+r5vZtKjj4Xurnb5qU7ASqZGoB/Yn650OaVi3A32TzGhmhwH/B3zU3d8N/E/1klUbClYiIg0gV8e2ZvZOMxsS9T050syOjCb9ALgh09FtijtRTkzBSiSLmX3LzF6MxqG62cx6mNlGM7vKzF4ys2fNrFc077FmNjYaU+gRM9sv+v5QM3smGv/pJTN7Z7T4vczsQTObbmb3RL2YY2aXRVU2k8zsr3XadGk8twA/c/f3A78Gboy+Pxw43Mz+E52fiUpkaaZgJRJjZu8CvkGoPjkWaAe+CbweeMndjwOGE3oTALiT0JvD0YTeyDPf30O4sz2GMO5Qpkuf9xGqZI4ijB/1UTN7I6Gn+XdHy7m0mtsoXYOZ7UU4t/5lZhOAm4HMcCc9gcMIPWCcAdxmZvvWPpWV02W7WxLppM8QekofFxV6XkcYV2s7cH80z93Aw2a2D7Cvu2dGAL6DkHHsDRzg7o8AuPtmgGh5L7p7c/R5AtCbMBbRZkKGMgjQcy1JYjdgbXRTla0ZGOvu24B5ZjaDELzG1TB9FaWSlcjODLjD3Y+N/o5w94tyzFeon7JCY5tsif3fDvR09zbCAIgPAV8ChpSWZOmOouFZ5pnZ1wCiznkzQ9s/Cnwq+n5/QrVgI3Y83EHBSmRnzwKnmdmbAczsjWb2dsK1clo0z5nAKHdfB6wxs49H358FDI8ykWYz+1K0jNdGva7nFFXn7OPugwlVhMdWfKuk4UUd244BjjCzZjM7h1BFfY6ZTQSmsmOE56HAKjObBgwD/rfRhxBRNaBIjLtPM7MLgH+b2W7ANuAnwCbg3WY2HlhHeK4F8G2gfxSM5rKjF++zgJvN7OJoGV8rsNq9gcfMbA9CqewXFd4s6QLc/Yw8k3ZpPBGNO/bL6K9LUK/rIgmY2UZ336ve6RDprlQNKCIiqaeSlYiIpJ5KViIiknoKViIiknoKViIiknoKViIiknoKViIiknr/H88fwYi5Bs2SAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "yHat = nn3.predict(X_test)\n",
    "print('Accuracy Score: ', nn3.check_accuracy(y_test, yHat))\n",
    "nn3.plot_cost(title='Model with 1 hidden layer (500 nodes) and ReLU activation and SGD')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "1c53b8cc",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x1080 with 25 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "nn3.plot_images_with_prediction(test_data, y_test, yHat)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "44d2d859",
   "metadata": {},
   "source": [
    "# Fit the Model with 2 hidden layer with 500 and 500 nodes, ReLU activation function with mini-batchs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "a03d53a1",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Input features = 784 \n",
      "Output labels = 10\n",
      "\n",
      "Hidden Layer 0 \n",
      "Number of nodes = 500 \n",
      "weight shape: (784, 500) \n",
      "Bias Shape (1, 500) \n",
      "Activation function: ReLU\n",
      "\n",
      "Hidden Layer 1 \n",
      "Number of nodes = 500 \n",
      "weight shape: (500, 500) \n",
      "Bias Shape (1, 500) \n",
      "Activation function: ReLU\n",
      "\n",
      "Hidden Layer 2 \n",
      "Number of nodes = 10 \n",
      "weight shape: (500, 10) \n",
      "Bias Shape (1, 10) \n",
      "Activation function: softmax\n",
      "\n"
     ]
    }
   ],
   "source": [
    "network_size = [784, 500, 500,  10]\n",
    "activation_fct = ['ReLU', 'ReLU', 'softmax']\n",
    "nn4 = NeuralNetwork(network_size, activation_fct, learning_rate=0.1)\n",
    "nn4.model_summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "0ce26b17",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 0 - Loss:0.0648\n",
      "=======================================================================\n",
      "Epoch: 10 - Loss:0.0309\n",
      "=======================================================================\n",
      "Epoch: 20 - Loss:0.0253\n",
      "=======================================================================\n"
     ]
    }
   ],
   "source": [
    "nn4.fit(X_train, y_train, nb_epochs = 500, loss_every=10, batch_size = 100, stop=0.025)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "bed17f39",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy Score:  88.88000000000001\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "yHat = nn4.predict(X_test)\n",
    "print('Accuracy Score: ', nn4.check_accuracy(y_test, yHat))\n",
    "nn4.plot_cost(title='Model with 2 hidden layer (500 nodes) and ReLU activation and mini batch')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8a9521cd",
   "metadata": {},
   "source": [
    "Try to use 0.02 as stopping "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "0aa5cef6",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Input features = 784 \n",
      "Output labels = 10\n",
      "\n",
      "Hidden Layer 0 \n",
      "Number of nodes = 500 \n",
      "weight shape: (784, 500) \n",
      "Bias Shape (1, 500) \n",
      "Activation function: ReLU\n",
      "\n",
      "Hidden Layer 1 \n",
      "Number of nodes = 500 \n",
      "weight shape: (500, 500) \n",
      "Bias Shape (1, 500) \n",
      "Activation function: ReLU\n",
      "\n",
      "Hidden Layer 2 \n",
      "Number of nodes = 10 \n",
      "weight shape: (500, 10) \n",
      "Bias Shape (1, 10) \n",
      "Activation function: softmax\n",
      "\n"
     ]
    }
   ],
   "source": [
    "network_size = [784, 500, 500,  10]\n",
    "activation_fct = ['ReLU', 'ReLU', 'softmax']\n",
    "nn5 = NeuralNetwork(network_size, activation_fct, learning_rate=0.1)\n",
    "nn5.model_summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "adebc036",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 0 - Loss:0.0644\n",
      "=======================================================================\n",
      "Epoch: 10 - Loss:0.0312\n",
      "=======================================================================\n",
      "Epoch: 20 - Loss:0.0257\n",
      "=======================================================================\n",
      "Epoch: 30 - Loss:0.0218\n",
      "=======================================================================\n"
     ]
    }
   ],
   "source": [
    "nn5.fit(X_train, y_train, nb_epochs = 500, loss_every=10, batch_size = 100, stop=0.02)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "7c2be30e",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy Score:  89.74\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "yHat = nn5.predict(X_test)\n",
    "print('Accuracy Score: ', nn5.check_accuracy(y_test, yHat))\n",
    "nn5.plot_cost(title='Model with 2 hidden layer (500 nodes) and ReLU activation and mini batch')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "15ec4991",
   "metadata": {},
   "outputs": [],
   "source": [
    "# model 0\n",
    "yHat0 = nn0.predict(X_test)\n",
    "accuracy_0 = nn0.check_accuracy(y_test, yHat0)\n",
    "\n",
    "# model 1\n",
    "yHat1 = nn1.predict(X_test)\n",
    "accuracy_1 = nn1.check_accuracy(y_test, yHat1)\n",
    "\n",
    "# model 2\n",
    "yHat2 = nn2.predict(X_test)\n",
    "accuracy_2 = nn2.check_accuracy(y_test, yHat2)\n",
    "\n",
    "# model 3\n",
    "yHat3 = nn3.predict(X_test)\n",
    "accuracy_3 = nn3.check_accuracy(y_test, yHat3)\n",
    "\n",
    "# model 4\n",
    "yHat4 = nn4.predict(X_test)\n",
    "accuracy_4 = nn4.check_accuracy(y_test, yHat4)\n",
    "\n",
    "# model 5\n",
    "yHat5 = nn2.predict(X_test)\n",
    "accuracy_5 = nn5.check_accuracy(y_test, yHat5)\n",
    "\n",
    "accuracy = [accuracy_0, accuracy_1, accuracy_2, accuracy_3, accuracy_4, accuracy_5]\n",
    "models = ['sigmoid act', 'ReLU act', 'ReLU mini-batch', 'ReLU SGD', '2 hidden layers minibatch', 'model 4 change stopping']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "5ba31448",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAl4AAAGmCAYAAACtCIpvAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjQuMywgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/MnkTPAAAACXBIWXMAAAsTAAALEwEAmpwYAABER0lEQVR4nO3dd5hkVdXF4d+CIQcBGZBBgiASBEWSoIgEkSxZEMkIgigqBoKKCioGzBgARRAUJSlJPgkSRSVIkKQgGUYYcoYB1vfHOQ1FM6GmZ+pWV/V6n2ee6bp1q3t3dVfXvufss49sExERERGdN123A4iIiIgYKZJ4RURERDQkiVdEREREQ5J4RURERDQkiVdEREREQ5J4RURERDQkiVdEh0m6UNJHevXzjzQqfiXpEUmXt3H+opIsaVS9fbaknVru/5qkByX9r97eXNLdkp6U9I7OfSfDj6Q7JL2vjfNe9ZxG9JMkXhHTQH1Deaa+md5f37hnn8LPMdk3G0lfkXT81Ecck7A6sC7wRturTOmDbW9g+1gASQsBnwGWsf2GesphwMdtz2776mkVdDuSpEd0XxKviGlnE9uzAysAKwNf7HI8PaWONA2Hv0mLAHfYfmoafa6HbD8w6NgNQ/lkGQGK6H3D4Y9cRF+xfS9wNrDs4PskTSfpi5LulPSApF9Lel29++L6/6N15Gy1QY9dHzgQ2Kbef23L3YtI+qukJySdI2nelsetKukySY9KulbSmhOLXdL+kv5bP8+NkjYfdP/ukm5quX+FenwhSadKGifpIUmH1+OvGqGbwLTchZK+LumvwNPAYpJ2afkat0n66KAYNpV0jaTHa6zrS9pa0lWDzvuMpD9O5PscI+l0SQ9LulXS7vX4bsAvgNXqc/zVCTx2ekmH1enD24CNBt1/oaSP1Cm1c4Ex9XOdIOlJYHrgWkn/bYnllPrc3S5pn5bP9RVJJ0s6XtLjwM6SXifpl5LGSrq3TmVOX8/fWdKlNb5H6ufboN73deA9wOE1nsMn8L0N/Hx2UZkOfUTSnpJWlnRd/R06vOX8Sf0+I2mHet9Dkr4w6GtN1/L79pCkEyXNM5Gf1871d+GJ+j19eELnRfQE2/mXf/k3lf+AO4D31Y8XooxoHFJvXwh8pH68K3ArsBgwO3AqcFy9b1HAwKhJfJ2vAMcPOnYh8F/gLcAs9fY3630LAg8BG1IutNatt0dP5PNvDYyp524DPAUs0HLfvZTRPAFvpozeTA9cC3wfmA2YGVh9QvEO/h5rrHcBbwVGATNQEpnF69d4LyUhW6GevwrwWP0+pqvf31LATMDDwNItX+tqYMuJfJ8XAT+tsS4PjAPWqfftDFw6iZ/BnsDN9ec8D3DBBL6ngZ/3msA9gx5v4M314+mAq4CDgBnr78VtwHotz994YLN67izAH4Ej6nM9H3A58NGW2McDu9efy17AfYAGxzaR723g5/Pz+ty8H3i2fs356vP9APDeNn6flwGeBNaoP5/vAS/wyuvkU8DfgTfW+48AThj8e1K/z8eBJet9CwBv7fZrPv/yb6j/MuIVMe38UdKjwKWUN/ZvTOCcDwPfs32b7SeBA4BtNfVTSL+y/R/bzwAnUpIJgO2BP9n+k+2XbJ8LXElJxF7D9km276vn/h64hZLsAHwE+LbtK1zcavvOev8Y4HO2n7L9rO1LpyD2Y2zfYPsF2+Ntn2X7v/VrXAScQxmpAdgNONr2uTXGe23fbPs54Pf1+0XSWylv3mcO/mIqdVerA/vVWK+hjHLt0Ga8HwR+YPtu2w8Dh07B9zrYypQk+GDbz9u+DTgK2LblnL/Z/qPtl4A5gQ2AT9Xn+gFKwtt6/p22j7L9InAsJVGZfwrjOqQ+N+dQku8TbD/gMpp7CTCwKGBSv89bAWfavrj+fL4EvNTyNT4KfMH2PfX+rwBbTeS18BKwrKRZbI+1PaSp2ojhIIlXxLSzme25bC9i+2M1CRpsDHBny+07KVf1U/rGONj/Wj5+mjL6AGVEaus6RfRoTQxXp7wZv4akHes03sC5ywID05YLUUbWBluI8mb/whBjv3tQDBtI+nudBnyUkiROLgYoScZ2kkRJok6sb+iDjQEetv1Ey7E7KaM57RgzKOY7J3ZiGxahTEW2/nwO5NW/D3cPOn8GYGzL+UdQRqMGvPy7YPvp+uEULfQA7m/5+JkJ3B74fJP6fX7V8+RSM/fQoO/lDy3fx03Aiwx6LdTHbUMZaRwr6SxJS03h9xMxbKRQM6JZ91HecAYsTJl+uZ/23vg9hV/vbsrUz+6TO1HSIpTRlnUooywvSrqGMuU38LkWn8jXWFjSqAkkX08Bs7bcfgOv9fL3JGkm4BRgR+A02+NrndbkYsD23yU9Txkd267+m5D7gHkkzdGSfC1MmUZtx1hKAjhg4TYfNyF3A7fbXmIS57T+zO8GngPmHWKiO6W/P5Mzqd/nscDSA3dImhV4fcu5dwO72v7r4E8qadHW27b/DPxZ0izA1yi/p+8Z/LiIXpARr4hmnQB8WtKbVNpNfAP4fX0THUeZUllsEo+/H1hU7a/+Ox7YRNJ6tSh8ZklrSnrjBM6djfLGPA5A0i68eoHAL4DPSlpRxZtrsnY55U32m5Jmq1/j3fUx1wBrSFq4Fl0fMJl4Z6TU+4wDXqiF4e9vuf+XwC6S1qnF2QsOGv34NXA48MLEpjtt3w1cBhxaY30bZQrzN5OJbcCJwD6S3ihpbmD/Nh83IZcDj0vaT9Is9We0rKSVJxL7WMrU63clzVmfg8UlvbfNr3c/k/79mlKT+n0+GdhY0uqSZgQO5tXvOT8Hvl5/h5A0WtKmg7+ApPklfUDSbJSk80nKyFhET0riFdGso4HjKCsYb6cULn8CXp4W+jrw1zr9suoEHn9S/f8hSf+c3BerScamlOmrcZRRhs8xgde+7RuB7wJ/o7xBLwf8teX+k2p8vwWeoBRcz1NriTahFNvfBdxDmRqi1pT9HriOUkT+mpqrQTE8AexDSW4eoYxand5y/+XALpS6pscotXStIy7HUZLF4yb5xMCHKDVg9wF/AL5cY23HUcCfKQsK/kkpKB+Sluduecrvw4OUBPd1k3jYjpQE9UbKc3QyE5k6noAfUuqoHpH0oyGG3WpSv883AHtTfl/G1ljvGRTL6cA5kp6gFNq/cwJfYzpKL7T7KAso3gt8bBrEHtEVAytdIiJ6Xp2KeoCyCvKWbscTETFYRrwiop/sBVyRpCsihqsU10dEX5B0B6UIf7PuRhIRMXGZaoyIiIhoSKYaIyIiIhqSxCsiIiKiIT1R4zXvvPN60UUX7XYYEREREZN11VVXPWh79ITu64nEa9FFF+XKK6/sdhgRERERkyVpoluJZaoxIiIioiFJvCIiIiIaksQrIiIioiFJvCIiIiIaksQrIiIioiFJvCIiIiIaksQrIiIioiFJvCIiIiIaksQrIiIioiFJvCIiIiIaksQrIiIioiE9sVdjRESvWXT/s7odwpDc8c2Nuh3CkOT5jl6REa+IiIiIhmTEK2KEyIhAREwrvfr3BLr/NyUjXhERERENSeIVERER0ZAkXhERERENSeIVERER0ZAkXhERERENSeIVERER0ZAkXhERERENSeIVERER0ZAkXhERERENSeIVERER0ZAkXhERERENSeIVERER0ZAkXhERERENSeIVERER0ZBR3Q4gRq5F9z+r2yEMyR3f3KjbIURERI/KiFdEREREQ5J4RURERDQkiVdEREREQ5J4RURERDSko4mXpE9LukHS9ZJOkDSzpHkknSvplvr/3J2MISIiImK46FjiJWlBYB9gJdvLAtMD2wL7A+fbXgI4v96OiIiI6HudnmocBcwiaRQwK3AfsClwbL3/WGCzDscQERERMSx0LPGyfS9wGHAXMBZ4zPY5wPy2x9ZzxgLzdSqGiIiIiOGkk1ONc1NGt94EjAFmk7T9FDx+D0lXSrpy3LhxnQozIiIiojGdnGp8H3C77XG2xwOnAu8C7pe0AED9/4EJPdj2kbZXsr3S6NGjOxhmRERERDM6mXjdBawqaVZJAtYBbgJOB3aq5+wEnNbBGCIiIiKGjY7t1Wj7H5JOBv4JvABcDRwJzA6cKGk3SnK2dadiiIiIiBhOOrpJtu0vA18edPg5yuhXRERExIiSzvURERERDUniFREREdGQJF4RERERDUniFREREdGQJF4RERERDUniFREREdGQJF4RERERDUniFREREdGQJF4RERERDUniFREREdGQJF4RERERDUniFREREdGQJF4RERERDUniFREREdGQJF4RERERDUniFREREdGQJF4RERERDUniFREREdGQJF4RERERDUniFREREdGQJF4RERERDUniFREREdGQJF4RERERDUniFREREdGQJF4RERERDUniFREREdGQJF4RERERDUniFREREdGQJF4RERERDUniFREREdGQJF4RERERDUniFREREdGQJF4RERERDUniFREREdGQJF4RERERDUniFREREdGQJF4RERERDUniFREREdGQJF4RERERDUniFREREdGQJF4RERERDUniFREREdGQySZekr7VzrGIiIiImLR2RrzWncCxDaZ1IBERERH9btTE7pC0F/AxYDFJ17XcNQfw104HFhEREdFvJpp4Ab8FzgYOBfZvOf6E7Yc7GlVEREREH5roVKPtx2zfYftDwELA2rbvBKaT9KbGIoyIiIjoE+0U138Z2A84oB6aETi+k0FFRERE9KN2ius3Bz4APAVg+z5KnVdERERETIF2Eq/nbRswgKTZ2v3kkuaSdLKkmyXdJGk1SfNIOlfSLfX/uYcafEREREQvaSfxOlHSEcBcknYHzgOOavPz/xD4P9tLAW8HbqIU6p9vewngfF5duB8RERHRtya1qhEA24dJWhd4HFgSOMj2uZN7nKQ5gTWAnevneR54XtKmwJr1tGOBCyk1ZBERERF9bbKJV51a/IvtcyUtCSwpaQbb4yfz0MWAccCvJL0duAr4JDC/7bEAtsdKmm/qvoWIiIiI3tDOVOPFwEySFqRMM+4CHNPG40YBKwA/s/0OSnF+29OKkvaQdKWkK8eNG9fuwyIiIiKGrXYSL9l+GtgC+LHtzYFl2njcPcA9tv9Rb59MScTul7QAQP3/gQk92PaRtleyvdLo0aPb+HIRERERw1tbiZek1YAPA2fVY+3Uhv0PuLtOTwKsA9wInA7sVI/tBJw2RRFHRERE9KjJJlCUuqwDgD/YvkHSYsAFbX7+TwC/kTQjcBtlmnI6ykrJ3YC7gK2nPOyIiIiI3tPOyNXFlDqvgdu3Afu088ltXwOsNIG71mkzvoiIiIi+0c5UY0RERERMA0m8IiIiIhrSzibZ727nWERERERMWjsjXj9u81hERERETMJEi+trC4l3AaMl7dty15zA9J0OLCIiIqLfTGpV44zA7PWcOVqOPw5s1cmgIiIiIvrRRBMv2xcBF0k6xvadAJKmA2a3/XhTAUZERET0i3ZqvA6VNGfdLPtG4N+SPtfhuCIiIiL6TjuJ1zJ1hGsz4E/AwsAOnQwqIiIioh+1k3jNIGkGSuJ1mu3xgDsaVUREREQfaifxOgK4A5gNuFjSIpQC+4iIiIiYAu3s1fgj4Ecth+6UtFbnQoqIiIjoT+10rp9f0i8lnV1vLwPs1PHIIiIiIvpMO1ONxwB/BsbU2/8BPtWheCIiIiL6VjuJ17y2TwReArD9AvBiR6OKiIiI6EPtJF5PSXo9dSWjpFWBxzoaVUREREQfmmxxPbAvcDqwuKS/AqOBrTsaVUREREQfaifxugF4L7AkIODftDdSFhEREREt2kmg/mb7Bds32L6+NlD9W6cDi4iIiOg3Ex3xkvQGYEFgFknvoIx2AcwJzNpAbBERERF9ZVJTjesBOwNvBL7LK4nX48CBnQ0rIiIiov9MNPGyfSxwrKQtbZ/SYEwRERERfWmyNV5JuiIiIiKmjaxOjIiIiGhIEq+IiIiIhrTTxwtJ7wIWbT3f9q87FFNEREREX5ps4iXpOGBx4Bpe2aPRQBKviIiIiCnQzojXSsAytt3pYCIiIiL6WTs1XtcDb+h0IBERERH9rp0Rr3mBGyVdDjw3cND2BzoWVUREREQfaifx+kqng4iIiIgYCSabeNm+SNL8wMr10OW2H+hsWBERERH9Z7I1XpI+CFwObA18EPiHpK06HVhEREREv2lnqvELwMoDo1ySRgPnASd3MrCIiIiIftPOqsbpBk0tPtTm4yIiIiKiRTsjXv8n6c/ACfX2NsCfOhdSRERERH9qp7j+c5K2AFYHBBxp+w8djywiIiKiz7S1V6PtU4FTOxxLRERERF9LrVZEREREQ5J4RURERDSknT5eG0tKghYRERExldpJqLYFbpH0bUlLdzqgiIiIiH412cTL9vbAO4D/Ar+S9DdJe0iao+PRRURERPSRtqYQbT8OnAL8DlgA2Bz4p6RPdDC2iIiIiL7STo3XJpL+APwFmAFYxfYGwNuBz3Y4voiIiIi+0U4fr62B79u+uPWg7acl7dqZsCIiIiL6TzuJ15eBsQM3JM0CzG/7DtvndyyyiIiIiD7TTo3XScBLLbdfrMciIiIiYgq0k3iNsv38wI368YydCykiIiKiP7WTeI2T9IGBG5I2BR7sXEgRERER/amdGq89gd9IOhwQcDewY7tfQNL0wJXAvbY3ljQP8HtgUeAO4IO2H5nCuCMiIiJ6TjsNVP9re1VgGWAZ2++yfesUfI1PAje13N4fON/2EsD59XZERERE32tnxAtJGwFvBWaWBIDtg9t43BuBjYCvA/vWw5sCa9aPjwUuBPabgpgjIiIielI7DVR/DmwDfIIy1bg1sEibn/8HwOd59arI+W2PBaj/zzeRr7uHpCslXTlu3Lg2v1xERETE8NVOcf27bO8IPGL7q8BqwEKTe5CkjYEHbF81lMBsH2l7JdsrjR49eiifIiIiImJYaWeq8dn6/9OSxgAPAW9q43HvBj4gaUNgZmBOSccD90tawPZYSQsADwwl8IiIiIhe086I1xmS5gK+A/yTshLxhMk9yPYBtt9oe1FgW+AvtrcHTgd2qqftBJw25WFHRERE9J5JjnhJmo6yAvFR4BRJZwIz235sKr7mN4ETJe0G3EWpGYuIiIjoe5NMvGy/JOm7lLoubD8HPDelX8T2hZTVi9h+CFhnSj9HRERERK9rZ6rxHElbaqCPREREREQMSTvF9fsCswEvSHqW0lLCtufsaGQRERERfWayiZftOZoIJCIiIqLfTTbxkrTGhI7bvnjahxMRERHRv9qZavxcy8czA6sAVwFrdySiiIiIiD7VzlTjJq23JS0EfLtjEUVERET0qXZWNQ52D7DstA4kIiIiot+1U+P1Y8D15nTA8sC1HYwpIiIioi+1U+N1ZcvHLwAn2P5rh+KJiIiI6FvtJF4nA8/afhFA0vSSZrX9dGdDi4iIiOgv7dR4nQ/M0nJ7FuC8zoQTERER0b/aSbxmtv3kwI368aydCykiIiKiP7WTeD0laYWBG5JWBJ7pXEgRERER/amdGq9PASdJuq/eXgDYpmMRRURERPSpdhqoXiFpKWBJygbZN9se3/HIIiIiIvrMZKcaJe0NzGb7etv/AmaX9LHOhxYRERHRX9qp8drd9qMDN2w/AuzesYgiIiIi+lQ7idd0kjRwQ9L0wIydCykiIiKiP7VTXP9n4ERJP6dsHbQn8H8djSoiIiKiD7WTeO0H7AHsRSmuPwc4qpNBRURERPSjyU412n7J9s9tb2V7S+AG4MedDy0iIiKiv7Qz4oWk5YEPUfp33Q6c2sGYIiIiIvrSRBMvSW8BtqUkXA8Bvwdke62GYouIiIjoK5Ma8boZuATYxPatAJI+3UhUEREREX1oUjVeWwL/Ay6QdJSkdSjF9RERERExBBNNvGz/wfY2wFLAhcCngfkl/UzS+xuKLyIiIqJvtLOq8Snbv7G9MfBG4Bpg/04HFhEREdFv2ulc/zLbD9s+wvbanQooIiIiol9NUeIVEREREUOXxCsiIiKiIUm8IiIiIhqSxCsiIiKiIUm8IiIiIhqSxCsiIiKiIUm8IiIiIhqSxCsiIiKiIUm8IiIiIhqSxCsiIiKiIUm8IiIiIhqSxCsiIiKiIUm8IiIiIhqSxCsiIiKiIUm8IiIiIhqSxCsiIiKiIUm8IiIiIhqSxCsiIiKiIUm8IiIiIhqSxCsiIiKiIUm8IiIiIhrSscRL0kKSLpB0k6QbJH2yHp9H0rmSbqn/z92pGCIiIiKGk06OeL0AfMb20sCqwN6SlgH2B863vQRwfr0dERER0fc6lnjZHmv7n/XjJ4CbgAWBTYFj62nHApt1KoaIiIiI4aSRGi9JiwLvAP4BzG97LJTkDJhvIo/ZQ9KVkq4cN25cE2FGREREdFTHEy9JswOnAJ+y/Xi7j7N9pO2VbK80evTozgUYERER0ZCOJl6SZqAkXb+xfWo9fL+kBer9CwAPdDKGiIiIiOGik6saBfwSuMn291ruOh3YqX68E3Bap2KIiIiIGE5GdfBzvxvYAfiXpGvqsQOBbwInStoNuAvYuoMxRERERAwbHUu8bF8KaCJ3r9OprxsRERExXKVzfURERERDknhFRERENCSJV0RERERDknhFRERENCSJV0RERERDknhFRERENCSJV0RERERDknhFRERENCSJV0RERERDknhFRERENCSJV0RERERDknhFRERENCSJV0RERERDknhFRERENCSJV0RERERDknhFRERENCSJV0RERERDknhFRERENCSJV0RERERDRnU7gOFi0f3P6nYIQ3LHNzfqdggRERHRpox4RURERDQkiVdEREREQ5J4RURERDQkiVdEREREQ5J4RURERDQkiVdEREREQ5J4RURERDQkiVdEREREQ5J4RURERDQkiVdEREREQ5J4RURERDQkiVdEREREQ5J4RURERDQkiVdEREREQ5J4RURERDQkiVdEREREQ5J4RURERDQkiVdEREREQ5J4RURERDQkiVdEREREQ5J4RURERDQkiVdEREREQ5J4RURERDQkiVdEREREQ5J4RURERDQkiVdEREREQ5J4RURERDQkiVdEREREQ5J4RURERDSkK4mXpPUl/VvSrZL270YMEREREU1rPPGSND3wE2ADYBngQ5KWaTqOiIiIiKZ1Y8RrFeBW27fZfh74HbBpF+KIiIiIaJRsN/sFpa2A9W1/pN7eAXin7Y8POm8PYI96c0ng3x0ObV7gwQ5/jXi1POfNyvPdrDzfzcrz3aw835O2iO3RE7pjVNORAJrAsddkf7aPBI7sfDiFpCttr9TU14s8503L892sPN/NyvPdrDzfQ9eNqcZ7gIVabr8RuK8LcUREREQ0qhuJ1xXAEpLeJGlGYFvg9C7EEREREdGoxqcabb8g6ePAn4HpgaNt39B0HBPQ2LRmvCzPebPyfDcrz3ez8nw3K8/3EDVeXB8RERExUqVzfURERERDknhFRERENCSJ1zQiaUJtMmIqScrvaIwI+RsSMTLkTW0qSZoVwLYlLdztePqJpLko20ohaWVJb+huRDEhSY6nDQ8quM3z2l+SWMeAFNdPJUlbAmsDvwcOAHaznb5k04CklYG1gOWApYA1bD/T3ahiYiStCNwJPGv7yW7H00skrUn5OzIX8Bywn+2XuhhSdIiktwNPA0/bvrfb8fQzScvbvqbbcQyWK6qpZPsU4D3AOcCXbN8naYYuh9XTWq70rwaWALYAjhtIunLlODxIWkvSAfXjvYA/AocBB2V0sn2S1qMszb8TuI1ykfFPSWPq/fk73cMkLSTppPrxKsDZwNcpr5M1uhpcHxp4f5D0ZuBPks7sckivkRGvIZKkOr04B/AZ4P3Ag7Y/0OXQetrA81o/Xh54kpJ4vQ64Gfid7fGSZs+oSndJWhK4jpI0PAH8jLITxUaUkZtDbN/ftQB7gKQPAF8C9rB9dcvxY4Hlbb+9a8HFNCPpOuAh4BLgD8A4YFNgVeBI25d0Mby+I2ljYDfgb8DOwJ22N+hqUC2SeA1BS9K1CfBu4Ae2/yfpAmC87fdLWhZ4k+0zuhttb5L0McoL5/3AY8BHKaNfFwFzA3MCh9t+oWtBjlD1inI62y9KWgy4ELjF9jp1dGYFYDNKEvZ52w90LdhhquVvyKnACrYXrcdnaRnZ/RPwd9sHdzHUmAqSZrA9vn78B+BdwDK2H6qvnfUpU8yH276we5H2jzrjdApwvO0T67GLgEdsb9bN2AZkCHsI6h/MdwPfAs6w/b96fC1g1vpDPhF4toth9ixJawF7AJvZfqgmV7+kjHitCxwInJekq3kDCUNNut5D2Xt1TWBFSZ+3/ZLtK4EzKdNm+RszYbMB2N4CuLletGH7mYEFO5Tt1abvUnwxleprZbykVSS9xfbmwL+B3wHYvo1SonIx8EgXQ+0rNdG9j1JHN2BfYLU6ktx1+aM4dO8CTrL9V0nTD9R12V4d+CGwo+1zuxph75oduMT23ZJGSRpl+1nbPwc+Caxq+/ouxzgitUwDfxr4JrBIfQN5J/BZSfvV8/4OfGvgoiReIel9wKGSNgewvT7wfEvyNfCG8RzwVH1M6hp7TL1A3xA4DpivHlsDmKGOZmL7VuAo29d2L9Le1lLTtbSk+SSNAs4DfiHpLfW0UcDhwEJ1pqqrkngN3cOUBAHKlO14SatJWs72qfWqPyaj9Q2l5ePbKVcn69h+oe7vuZekfWyPt/1gd6INgDrauw2wke1bamL8b8q0+9clfQbA9nPdjHM4qm/EhwF/AR4fOG57PWD8QPJVk7LtKQsWXtNqIoa/ujjiAMpF+KWSZgSwvSYwz8DPmsyMDJmk6WqC+z7gfODHwHcpe0EfCJwh6cfAScBplJqvrmt8k+xe1FKP8Q7K8OWzlIz6E5K2Ay6TNC/wK8ofy2jDoEL6DwEL1CLUC4CfAzvXpdePUuq9duxWrCOZpNlsD4y8zERJGP4NPCZpZtvP1j+At9SVRDN2M97hqv4ufx/YqY4IDhzfyPZZtTb0bEmPUJ7frWz/p1vxxpSprw1sPyfpdZTXyZ3AA5KmB16q581qe9W6wjFJ9RC0lDy8JOldlM4Cm1LenzejzDp9hrKYYT7gB8CCwIbAMV0I+VUy4tWGmnRtBBxLudL/KaW4+8PABylTLt+nFBJnpKtNLUnX3sAnKPPyJ1FWoVxGeYGsDCwP7Gz7xm7EOZLVeqO1Ja0naRfgI5Tf/TGUxSMDV+sfkvRZ23ckWZioBShT6H+vixCQ9H3gdEk/Aqgrr06jrHLM73uPqInVOymvgy2Bgyg/79mBVWy/WEfu3wkcVldlX97FkHuWSqPyHSTNUGdJDqe8Z1xp+wbKqtFbKe/TM9r+K6VW8kuU95FbuhP5KzLi1QZJywBfpmTLmwELUX6o+9neTKWlxNy272odxYnJq8/txsAGwHaUacbtgdcDh9k+v4vhBbwIPAN8h9IiYmXbD0q6DDhGpUfOzJSLkM27FuUwVqdiX6Bc6M4OUK/UlwIeoFyJ/07SAbYPtb1z96KNoaiLTR4Cvkhp+LxbHQH+CSXRWoHyOtoaONBphTM1pgf+RXnPfaCWPlwF/ILyvN9YE+HpeGVxyj2UEeTHuhLxIGknMRGDpsGWoCSpA0OWWwG7AFtSmqae3K04e0nLlO10bunKLen1wIrA/rbXrvUtJwC7A7+1/WKXQh6xBv3+v4VSa/QvyoKSk+vxbYDRlAuRo2udV7SQtC5l5OMkSi+6fwJftn1UvX+mOjW1J2UU8ZC6Kit6TK3hOhkwcCnlNfFQnWJemXLhcoXti3KBPjQt7yEzA2dR2q18oY7MX0UZUd6jnvtyicRwk6nGQSTNAi9PL6p+fIvtm4BlKT27/kuZu7+IkklHexaHl6/2p2t5fh+iLK8fKDZ+HjgduCBJV/MGJV1z16nDt1NWZ20oabd66tWUxHi/JF2vVQvpf0TpQzddXeG5C2Wa5GPwcj3QhylTJccn6eotLSvqFrD9vEsD7QOARSgrsAHuAv5h+zDbF0HquoZi0IX7s8BewNslHeSyEngFYH1JxwAM16QLMuL1KnXk5ROULPr/6rHWN6H9KFNiRwOfoxTJ/rNb8faK+sdpRkqyeqzt/erxl0e+JC0OfJUyqjga+OBwmIsfySTtA7yP0nH7T7ZPqotJ1qXUec1HGb5Pd/pB6ijhicDutq9oedOYi3IB92vgekox8HLANk6LlJ5UE+yvUOpSn7f9+VrLtQNlFHMZYAfbV3Qvyv5QR5A/RBl9vwi4l1J7fbHtb9SRr5VsX9zFMCcrNV6vNgvlOVlD0vO2/9I6NWb7W3U4eRngC0m62lMT1+ckrQpcJOk52wcNjHzV5Os2ypXiSsB1dVQxuqROfW1Bqbf7DvBDSfPYPkLSLZRalYOSdE3UnMB9NemahTLKtRGlUeafKH9DVqTMOtxuOyPnPUhlr8VvURZdbQrsK2l+2ztJuofyGvpZkq6pV+vk9gfOoOxe8lXgJ8CuwImSprd9CHDxcJ/KzYhXNZAAqDRX246y9Pcnti+r97du/TBw9Tqsf7jDgaQ5bD/RcnthSi+VX9o+qB6bPlOK3TVoZHdmylXlaZQWHmtSlmcfDXzH9k+7FWevqCNbh1MSq2WAa4AHKTVeOwCfyvRsb6srU9cC7qcskDiYMmV8DPAf2zu0nJv3iqmgsi/sGcC3bf9CpV3HapQLw12AJYFZ3SMrRVPjxcsvipckrUfJoi+jFMRuIWltKNsQ1Bfay/PzeSFNmqSlgS9IWnHgmO27KC+YnSV9rh5L0tVFg98UXHYJ+BWl7m59YG/bF1BquvaUNPdAbUsULbU+A39TH6M0Sr2OMhVygO3P2v5tvS+9znqYSnd0u6y6vpVysX5IrQW+EFhJ0vID5+e9YqrdQ5kV2RfAZXXiFZSR5TfZvr5Xki4Y4YmXpIVVmtlZZcufDYHv2/4xpV/R05QpgvdAKQrvYri96BFKvdbmKs1ngZeTr/UphdpLdiu4AEmLUlscSNpH0hGSfidpIcoIzWPAmyTtCtwNrGP7kbyRvEbrwpFRLq6x/U3b37c9FkClx9NiZG++niJp/jpaj6SNKZsw/0HSlrXQ+2le2Y5mcWAT29d0LeAe13Ihs0ydzn2J0jPz75L+LGmgFvgt9OBFzIhOvCjJ1ZL1in88MI6SDMzlsv/cCcDqwKYqneljCtRVXAdSllF/sDX5Am4B7iDbZXSFitkp04d71gLhLSlF3//jlTqKSyivk08Dv7A9rkshD0v1eZwJuFTStwBcGmVON+i80ZI+S+kHuGtqunrO/sA3ahLwCeBnwO8pPbo24pVmz/tTVvre2q1A+0EdDFmXkuB+jbIrzOaUOuBnKAtTPgXs0ouLUkZ8jVe9sv81pTDyDcCelO06jqasSPkBZZrg5m7F2EvqCNZits9uOfYGSmPBR4HzbF9Yrwz3p6yKG9uVYIM6HXIoJQE+d6B+S9IhwHrAqpQGqaNsPz6xzzPS1ZHDiyirdgdqFwev2v0gcJrTkb5nSJofmB+4kdI0ezHgRtv71PvXAn5L2af0bmBm20+kpmvqSHor8G3g07b/I2kHSkubk4FrKfsxLmx743p+Tz3fI33EC9t3U5bLn0DpIn0BpR/IJcCZlCZ4SbraUKdrtwQ2rvVywMsjX9+gdEHfW9JplFq6jybp6p6aGFxDqZuYBVhlYGTX9pco7T9eb/vpJF2vpbJjBQC276DsF7ebpIPrsZda6kL/SykMTtLVI+rPbn1gvMvOAx+lvOkvL+ktKs1vL6AkAwvYHj+wkKiXkoDhRtJswCbAKsDS9fDvKFOKH7b9DPB5yqbyv4Hee75H3IhXy4rEt1JGuK62/bCknwMLA9vafryOBDyZIeMpU0e3dqRcJZ7XOvJV738zJeF/yva9XQgxWgysKJW0LGW/0UsoxcFvAL4OvCvTi69VF47sROnkf1XL8YUpXct/bPs73Yovpo1aRD83ZYr4D8BfgJ9T2g6dS+kj9RvKyH3PFHcPNy3vywPdBeag9MpcEPiV7UslbUDZmmx328/UUok5bd/XzdiHYsQlXgCSNqU0vBtLKR6+3Pb3a/L1NmA9t7RAiEkbPMxbR012pUzVnmv7rHp8H2Bj2+/vTqQjW72Y+LDtz9Xbo4CXWqbD3kh5U5kPOA84rq7SikHqBcbXKX9DTrF9dct9y1D6C+3ptIzoSa1/0ySNppSgzEvZ+umvwPcoe8yeAvy5jnzFVKjlJ1tT/v4cCjxMGXHcgVJz+h5KO5szuhbkNDLiphprlrwbsKPtDSkbay4saWPbe1K2d1immzH2kkF/oLaVtAWwiu1vU96U1pa0gqRdgL0pxfbRHbcA35G0CpQi8Hp1+W5J/6AsLjmAsnT7h0m6Ji4LR/pbHX1ZRdJqlNGtb1HeG7al1HPtS7k4OTFJ19Srf5O+BhxF6R/4RWBRSq31WZSVokfZPmNgxWMvG3GJF+VFND9lCBPK1cuDwPsBbG9r+x9diq3ntCRdHwc+DrwAnFpbcBxDaS54KKX7+da2r+xSqCOey95lDwFfknQGgEojwm9TehA9Z/tflOn2dKQfRNKSdboDgPocfQ2YA9hS0pr1rvUpy9yfbzrGmDoDNXmS3k2ZWtyX8vdrTcq+m/+lNEldw/Zezu4lQyJp+kGHxgBX2b7E9s+AH1Oe7zkogyP/ANaTtFKv1XNNSN8nXi39QN4k6fW2H6VMp2whaeXag+VKYB5Jsw1eBh6TJ2kM5Q/TOsCbgfMp+13eT3muz6DUCl3XtSBHqNarQ72yQ8AHgcclnejSiHBL22fW1ggCslHzIFk40t/qtPvAgoi1KNOIG1C6ov+FUs+3FmU3gn9TLtZjCFS23dtQ0ryS3qaySfwTwPSSFlHpg3c6ZWutBerClP8DLqeMxve8EVHjpdKj6JvAU5QXzl2Upal7UFYufgj42OBC8JgwSatTNva9nbL9z5PAEZR2EYtT6oielrQXpcA+m113waBp4L2ANwGPumwmOxNwPGXF1naDz4/XysKR/iRpAWB3yqrTZyUdDnwMWM72DfXCch1KH6mj8j4xdVQ2sn4/ZcHC6ygX7WMpvbruoIxuPVZvbzVQP6mWbft6Xd8mXi2rJGam9F/5HrAQ5Wr/EspVzBso88h32v5bt2LtJfVq/zDKsmoBN9s+RNLPKAX1s7tsr7QdZZh+M6dZZFdJ2oyymOQrlFqVi23vXpOvU4EHbe/UtQCHsSwcGRkkLUW5MH/J9r2SjqIUcy9X/56NAdYFrnU60k81SStRRrT+DWxv+856YbM3ZRHDEpRdZM7qxwvCvk28AFQ6Ci9LeQFtVa9mNqKsnLiGsmrroS6G2FNU9q38I7Ck7bF1Fcqmtj9S7z+asir075TGmzu7B7sK97r6JvJAbZOyLmVV0Em1MHUWyp6LF9resyZf82aE5rUGLxyh1Gw9a/tPkvajrL76DWX0fH/KSG9qGHtIy/Q7ko6h7P23d/379ktKN/pV6nvHjLZTtzdELYMhqwE3U9o3rUAZSTzU9t8kvYky+jVzLQvqS31bz1SXzn+VUuw9L6W7MPUK9VTgnZRNgKN9D1Kes7UA6rLepSR9vk7n7gYcRNkUeOskXc2TNBdlGH+gXmt6ykjvapLGuDQfXIFS4/ijWlCfpGsCsnCkv9VE4MW6wATbO1P6cn2nvlZ2o2xNc11qH6deTbo2pDRDXc72tZRSnwsoC34+TtkpZv5+TrqgT0e8JC0HfAa4zvb3JM1JKfKe3vY29ZzRTmPIKVaHiM+lTCOOAbYD/kXpfD7QhPMn/TY03EvqKNYSlLqVAygXGbtRClTPs/2/OvI1phauxkTUKaYfURo37k2p9dmsTj/NSan5Osf2f7oYZgxRXaX6JcqCoFNtXy3pJ5SN4w+s047L1dW+MRVUtuc7C9ipPs9LATNQCutXp/y9+o7tM7sYZiP6LvGqNV0LUa5CnwC+Yvu/9Y/krwFsb6aWfdRiykhaGTgHeNj24i3HNwauSU1X8wb/PtdRmW2ARyiNPtekJMkXA6fbfqAbcQ53WTgyctRE4KeUEfqVgZkoe2leUKcZZ6W0jng+F5JTT9KClIL6f1Iu0pcB5qG8R/9R0uy2n+zHmq7B+mKqcWDJvMoGzT8GnqaMyABsKmlRl73mdqQUGJOka+hsXwG8l9KCY4eW42cm6eoOv9J9fgVJi9i+hPKmMgdl+vdCyp5yq5DGnhNUF478jNIgcwdgn1r/Mx74BPDBmnRtRxlBfKZrwcZUkbQisCLwL9snU0bq7wE+IOl9dZrxa3Uqvq+TgKbUkoY7gLdSuv2vDRxHeS+B8r7dc/suDkXPj3i1FOytR0msVgAuotRezEwper0FOMH27d2LtP/UacfLgd1s/6rb8YxEgwrAPwHsR0myoCQPS1OShFH1vuldGqlGiywcGTnqaPBvKVPvuwGr275MZWugPSglFF+y/XAXw+wrE2oFodKt/ijgM7bP605k3dGziVfr1IrKhrVnUFZHLAq8g/LiOZgylHkg8EXbt3Un2v6lslXK086edI0blHTNQ2nyeCJlWuynlIUQW1OmzralLM9OXeMESHobZbXnDrZ/W49dCpxOKbA+m9JQcxyl/UYu4npQnRX5EvAL2xdK2oOyc8PGLhsxzwfMlp/v0LUMhswNPOPSpByVxqgv1I9XoJRA/NR9sPfilOrJxKvOzW9MaWb3gsr2Dgfa3qjevzyv7K31ReCRLAOOfjIo6doD2IiSaH2uFq7OSkm+xgDrATPkNTBpWTjS31R2JdkL2J7yc/667eck7UrZlmZN2xd3M8Z+Uds27QdcB8xoe48JnLOw7btGQk3XYL1a4/UscBkwb10KfA0ws6TdAVwa3P2dso3HNsBLylZA0Udakq4NKCMxv6W0O1hT0uK2n6aswrsNWDBJ1+TVdhDvpzRb3tX2W132bt0U+CTwx5H2BtFP6gzJLylJ1pyUliqjbB8N7EkpTYmpJGlVSiun3YE7gVUlzd5y/8D2THfV/0fca6rnRrwGGt7VROp0Sv3W14E1KH80n6JMO36L8mb0TspKpN76RiMmo47QnA3sZfvk2pjwo5SpsdPT4mBo6rTjRZTi+uO6HU9MGwPlKbWVyq7AYpQRmd8O1B+NxNGXaa3OQM0NmDKtu63tOyS9w3X7n5Gup0aBWhreLU2ZVtmDsv/cRylb2BwBLEB5Ue0JXEWp8ZqjOxFHTBstK3dbX7PXA38GDpb0Opdtr34CrAasr7Kxc0whl83c1wWOlbRLt+OJKdfyelmp1hMNbIAtlybCR1NWMa4EjB54XJKuoZO0sMqWWtNRRhUPoUzf3iFpTeDT9f4Rr6cSr1qwtwlwArC07fsoHaXfSVnReIvLhr+7UF5MPwH2q60kInrZ4vDym8d09eNnKQX15wJ/lDR3bfXxDeDkwauIon112nFFSklD9Jj6XrEZpc5x9kHHR9Xk6+fAYfV9JIagJcFdnrKYbYfayuablFGvN0naitLm6STbD3Yr1uGkp6Ya6w/3GMr2HLfUIvsXKY1ST6asSvpiLbjfHviH0+Awelj9wzYjpVbiWNv71eOte8zNQPlDtzawlvt8u42IyVHZ8+8EynvF3ZIWARa1fVG9Pw20p5HaOPtTlDKfeSnP+y8oCxlWq6f9yvbZmcotei3xWpqyUuJvlB/wupR55EMom26OcfZLiz4kaVFK3dGxtg+qx1qTrxkpzYGPsH1nt+KM6DaVrWhGU3YvOZoyWrwUZbunfWoxfUwDtY3N7yirqa9V2Ux+DeAG2z+p52Rz8UF6aqoRuBu4kjKteBMloz4FWNj2fUm6op9Ierk20fYdwHuA3SQdXI8NLDLB9vO2D0zSFSNRy5TX0pRZkRsouxBsA1wKbEXZ/mfu7kTYt56i5BHLANj+HfAAsIuknVX2jU3JwyA9lXjZftL24ZTplFMpBfYfpSRkEX2jvoF8QWVrE+Dl5derATtL+lw9lumSGPFq7da7KUnX92w/bPtYYL3aoPPdlBqk67oYZs9rSXDnkzTG9nPA8cCyKvucAvyJsjXQB4A5MrX4Wj2VeLV4sb4h/YRS03V+twOKmMYeoUyXbF53BwBeTr7WBzasXbgjorgZWJjSIHXADPV18l3KQqtzuxJZn6gJ7gcoydX5kj4MXELprfkFSUcCv6Y0Ln+OOhIWr9ZTNV6tJM0GzGf79hTsRT+SND+lD84TwIkDPXBqMf2RwFcytRgjVcvWNMtS9iK9gdIE9QbKqt59W85d0Pa9ea+YOpKWoSSx+1LqrA+n7It8NqUv2sqURGxuSrPa97lsjh0tenXEC9tPue6nlRdS9ANJS9ZO9ADYvh/4GqUP3Za1Fw6UEa+3AClYjRGrJl2bUlbQ7Qr8ntLH8R3ARpJ+1nLuvQOP6Uas/UDSgpQdHGYCbq1tIz4OfJ7SpPxq20dSkq7vANsk6Zqwnh3xiugndRTrc8CClK7zf265bwyllnEZSmuJhYAdbV/fjVgjhoOaCBwHbALsRtkIfnPb90t6PaW90PrATUm4hqZ1hLDWd21N2cP0LOD3th+vF4Q/ATZw2XtxFmCeJF0Tl8QrYpiQ9AbKit35gfNsnz3o/jdTRqmfyh+1GMkkzUy5CPk2ZdeS7YGdbN8qaS3bF9RGqS90NdAe1jKV+17KSOIDtv8iaWvgvZTE9mTbj9XmzY+0triJievZqcaIfjCwSgjA9v8ofYfuB9aTtFHLefsAP7X9nyRdMZLV5qiH1JvTA58FPlqTrnWA70p6c5KuqdOyU8z3KQt9DpS0v+2TgPMpK0W3kTQ98Fh9TJKuNozqdgARI9WgYfxtKTVbz9r+tqT9gLUljQXeDuwNfLh70UYMG2MohdzPU1bQPUdJCi4APg183vatXYyvL0h6I2U/5E0pPQTnAlaVdLDtgySNAm5MsjXlMtUY0WWSPk6pT/k2cCJlR4b/UPZhXIeyZ+DaLps3R4xIkuYa2A5L0rHAY7b3qe0iNgJeAK6v02FZvTiFBj9nkuamrFycCziKkoCtAPwAOMH2/l0Isy9kxCuii2rh/JqUBGtvyhD+322Pl/Rz4GngE7b/070oI7pL0mLA/pLG2v4yZW/S7STNafvfwL9bz0/SNeVaRt+Xo0wdPuWyJ/J6lAU/d9btmE6k7BgTQ5QRr4gG1e7OywG3U/YcfRI4AniUsqfch20/LWkvSoF9NnmPEWnQVPw8lOnFrwI3UhagLA8cavuErgXZByQtTOkJuGvt/n8SpRfXQ5SRrucpDVN/TRmF3972xd2Ktx8k8YpoSL1yPIyyCkvAzbYPqf2GdgVmryNd21EaFG5m+57uRRzRHS0r6taiXqjYPqO2XXkLsDawJ3ArsHdeJ1NH0m2UhPYq4HTKheCmlOR2P0px/duBe23/pTtR9o8kXhENkLQ28EdgSdtj62qhTW1/pN5/NPA24O/AqsDO6dMVI1FL0rUB8D3gIMqo8PeBn9l+sJ63AvApyt6M13Qp3J4maQbb4+vH51GS2qXqqPtbKD3SVgW+MbBzRky9tJOIaMaDlE3d1wKoG/cuJenzkjakNIA8CDgW2DpJV4w0khaq/aBce9rtC2xBmfL6H+W185m6lRa2/0np5bVxt2LuZTXBHS9pJUmL2H4f5e/UcQC1rvQs4HLgpS6G2ndSXB/RANvXSXoncK6kmShL4uemrBJ6N/BjyhX92SkMjhHq08A6tQHq/yTtBMwDfNP2MrXo+yrgQUlHAC8CswJ/6F7IvasmuBtRyh92A+60vYKkKyX93vY2tm+WdJftp7scbl/JVGNEgyStDJwDPGx78ZbjGwPXpFYlRrJa77gsZRr+YUnvAr5qe926ou7bwMG2r6znpzv9ENWi+t8A+9q+YtC04w2U/Rg37WqQfSpTjRENsn0FZbuNeSTt0HL8zCRdMVLV7ufY3gu4CfhjXcl4LXCnpLOAU4EfDiRd9fwkXUP3BHAPcF9thvoSvFz39VbgW90Mrp8l8YpoWG2Eui5wrKRduh1PRLfZfrEl+doDuJnSK2omyqq6I4DdbZ/fvSh7m6TB7/czAbMD77D9Qv0ZrAp8R9Isti9rPsqRIVONEV0i6R3A07UBZMSIVvctnW5gC5pax7U8sKHth7oZWy+T9PqB52/wJtaSPkDZ9/I04BlgR2A/26d3JdgRIolXREQ0qqVlxALA9K3T7K3JgaRfAr+w/bduxdrL6vN7IXCW7X3rselsv9RyzuqUVjbzARfavjBbLnVWEq+IiGhc7WV3CDAzpT/XD1vuS9H8NCBpEUp7iFmAq2zv2XJfnuMuSY1XREQ0StLSlM7z2wIfBHaW9MmB+5MQTL06dXsvZfuf7YH5JX1H0myS3pDnuHuSeEVERGPq9NengDkpW9BcR9kya3tJn+9mbP3ExQuUJrMrUJLcd1KSsdXhldWk0awkXhER0VF19AUA22MpTU8fAHaoxd9XU0bAdpC0aHei7C8tqxivq/+/QKnjugdYA8pq0i6ENuIl8YqIiI5pKaTfSNJXJB1KKfg+DlgS2ErSvLavAt5l+44uhtuzai+ul5PclgL6y4CPUEa6jqKMfi0tadluxBnZMigiIjqoJl3rAl+lTCn+AZjD9sfr9lnvA0bV9hFPdTHUniXp9cApkvas2/y0rkocC/wVON72r+r5m2YboO5J4hUREdNU3cj6jXUUC0py9VFgQcqG14cB2P69pOeB/6TYe+hsPyTpCuB4SR+yfctA8mX7aUnfs/0ovNxOIklXF6WdRERETDOSZqQkWe8Efmz7H5IOAt5M2Rx+r5oYfBiY1fZRXQy35w3qe/Y1YBNgq/ocv6pnVwwPqfGKiIhpxvbzlH0VbwB2q8XyZwIfAI6pCcEqwBeA27oWaB+oo1ov1pWi2P4iZePrUyQtYful1oUNMTxkqjEiIqaJltGXJYHlgGWArwBfBLYCvidpPWAJytY02XtxiFoXLQAfl3Q9cJ/tb0uaFThB0oezJdnwk6nGiIiYKq1TWrU56hnA5sAbgRWBhSkjXC9SNmeezfat2Zpm6khaATiGMpr4NeB1lGnG5+q04+bAyqnpGl4y1RgREUMmaQzwBUlvqIfmAG63/S/bZ1M2YF4QOBxYwvZY27dCWfHYlaB7lKTF6sbWA+YHfkZJcJcA9qlJ1+J12nGzJF3DTxKviIiYGuOB9wJ7SpoXuBZA0scAbP8LuBJ4mLSLmFqLUVYublZv3wV8DPgFsInt2+vU4xclzQ7c2p0wY1KSeEVExJDUjZbHAd+irKb7NjArZXRrWUk/kbQ2sBFwpO3ruxdtb6v1c+cBnwW+L+mDtm+gjCieC6woaXXg68AfbD+ZEcXhKTVeERExxQYK6SVtAOwLnEXZg/G3lBGYWYDPUUbEzrR9Wrdi7ReSNqRsKj4zsAWwDfB3YD1gR0p3+pNsn576ueEriVdERLRN0iLAo7Yfq53njwPOsn2spPnq7TuBT9XmnaNsv5BEYOhqS4h5gbOB/W2fVxPeY4GP2T659k9TrfHKcz2MZaoxIiKmxGbA4nXE6zngJuD1kmaz/QCwD7ADcICkGQc60icRGLragX4cpX7u4frcnw18AzhR0ha2n68/jzzXw1wSr4iIaJvtHwL3ARdLmoeyD+BqwFvrRs0vUBqonlmbqcYQDDQ+lTRa0lz18GPAbrzy3n0ZZe/LhxoPMIYsU40RETFZg3p1Cfg5sAilV9R2wLqUHl1vAT5ZC8FjKtSarkOAe4DHgU8AvwKeAe6n7IG5q+0rMr3YO5J4RUTERA3UaNWP3wzMaPvGevtHwFKU5GsGSsf652xf06Vw+4akpSgLFT4O/LN+/CSwKyXJHQPcZvuCrgUZQ5LEKyIiJqg2Rd0IOAlYFTgKuBmYzva69ZzvA+8GNq41XjEN1D0uDwH2sP1MPXYhcKrtH3UxtJhKqfGKiIiJeTNlO5rdKAXzW9hejzLbeD6A7U9TWhos3rUo+0BLTddMtVbuScqo1iotp/0OeK4L4cU0lBGviIiYKEnvBfagbE/zOdtX1+N/Bua0vVo34+sntSP97sB/KTV0CwKHAb+k1HV9itKm49wuhRjTQEa8IiLiNQZGYGxfROlE/xDwLklvqsfXA56X9M7uRdn7Wka65gJ2pvRBu4FS03UvpabrdcDKwKeTdPW+Ud0OICIihofWlYu2PXDb9t8kzUxJAl6SdI7t/9p+b3cj7n31eV4FWAm4yvbvACQ9CxxPWSF6SFYt9o8kXhERgaQlge0kvUh5w79jIAkDsH2BJFNW2Y2S9Avg2SQDQzOQSElalbLF0p3AfJIuBS6tOwGMAo6UtAZlk/EXuxhyTCOp8YqIGOFq64JTKXVF7wXGUUZaXlPIXTe9fsj2tc1G2X/qNO3BwGdsXy/pEGAu4GTgMtvjJY2xfV8344xpKzVeEREjWJ1C/Brwi9qmYBtKb66dB503UPP1lyRd08zrgHWA99fbB1NGtnYCVgdI0tV/knhFRIxgtp+l7Pl3TN0D8AXKZswzDTov0yPTmO1zgC2B3SRtZ3s8pXfX/4D0ROtTqfGKiIhrbbfWDz1I6eGFpLdTutVf0ZXI+pzt0ySNBw6pm4ofAxzY5bCigzLiFRExwg0kXQPTicCMwHOS3kbpWp/3ig6y/Sfgq8DnJY2RNH23Y4rOSXF9RES8iqR1KI07nwEOtX1Gl0MaESSNtj2u23FEZ2WqMSIiBnscWBrY3PbZ3Q5mpEjSNTJkxCsiIl6l9o9ayPbtadwZMW0l8YqIiIlK4hUxbSXxioiIiGhIVqpERERENCSJV0RERERDknhFRERENCSJV0T0JEmWdFzL7VGSxkk6cwo/zx2S5p3acyIi2pHEKyJ61VPAspJmqbfXBe7tYjwREZOVxCsietnZwEb14w8BJwzcIWkeSX+UdJ2kv9ftb5D0eknnSLpa0hGAWh6zvaTLJV0j6YjBW7dImk3SWZKulXS9pG06/y1GRD9J4hURvex3wLaSZgbeBvyj5b6vAlfbfhtl0+Ff1+NfBi61/Q7gdGBhAElLA9sA77a9PPAi8OFBX2994D7bb7e9LPB/HfmuIqJvZcugiOhZtq+TtChltOtPg+5eHdiynveXOtL1OmANYIt6/CxJj9Tz1wFWBK6oe0XPAjww6HP+CzhM0reAM21fMu2/q4joZ0m8IqLXnU7Z0HlN4PUtxzWBcz3o/1YCjrV9wMS+kO3/SFoR2BA4VNI5tg8eUtQRMSJlqjEiet3RwMG2/zXo+MXUqUJJawIP2n580PENgLnr+ecDW0mar943j6RFWj+hpDHA07aPpyR7K3TiG4qI/pURr4joabbvAX44gbu+AvxK0nXA08BO9fhXgRMk/RO4CLirfp4bJX0ROEfSdMB4YG/gzpbPuRzwHUkv1fv3mvbfUUT0s+zVGBEREdGQTDVGRERENCSJV0RERERDknhFRERENCSJV0RERERDknhFRERENCSJV0RERERDknhFRERENCSJV0RERERD/h8KYUJsQkUqWQAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = plt.figure(figsize = (10, 5))\n",
    " \n",
    "plt.bar(models, accuracy, align='edge', width = 0.5)\n",
    "plt.xticks(rotation=45)\n",
    "plt.xlabel(\"Models\")\n",
    "plt.ylabel(\"Accuracy on test set\")\n",
    "plt.title(\"Plot the accuracy of different models\")\n",
    "plt.show()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.8.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}