Optimizer_Comparison.ipynb

optimizer_comparison.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Optimizer_Comparison.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "machine_shape": "hm",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "TPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/gist/utkucanaytac/018c47bcb69fb34902bedf1a55728cf2/optimizer_comparison.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FX3f0dwkvcPf",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "902c8505-d02d-4972-cf0e-50cbcdd8e854"
      },
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Mounted at /content/drive\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vPQQLX1jOshr",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "f1d1397b-35e2-4e56-92bb-cb1fac4cfc9a"
      },
      "source": [
        "!pip install pyyaml h5py\n",
        "import tensorflow as tf\n",
        "tf.test.gpu_device_name()\n",
        "import pickle\n",
        "import tensorflow.keras as keras\n",
        "import os\n",
        "from keras.callbacks import Callback\n",
        "import matplotlib.pyplot as plt\n",
        "from tensorflow.keras.models import Sequential\n",
        "from tensorflow.keras import backend as K\n",
        "from tensorflow.keras.utils import to_categorical\n",
        "import numpy as np\n",
        "os.environ['TF_KERAS'] = '1'\n",
        "from sklearn.utils.class_weight import compute_class_weight\n",
        "import logging, warnings\n",
        "from sklearn.metrics import classification_report\n",
        "from tensorflow.python.training import optimizer\n",
        "from tensorflow.python.ops import math_ops, state_ops, control_flow_ops\n",
        "from tensorflow.python.framework import ops\n",
        "import tensorflow as tf\n",
        "from tensorflow.keras import backend as K\n",
        "from tensorflow.keras.optimizers import SGD\n",
        "from tensorflow.keras.layers import Conv2D, Input, BatchNormalization, Activation, MaxPooling2D, Flatten, Dense,Dropout\n",
        "from tensorflow.keras import regularizers\n",
        "from tensorflow.keras.optimizers import Adadelta,Adam,Adagrad,RMSprop\n",
        "from keras.callbacks import EarlyStopping, ModelCheckpoint\n",
        "import os\n",
        "import sys\n",
        "np.set_printoptions(threshold=sys.maxsize)\n",
        "\n",
        "def loggings():\n",
        "\n",
        "    logging.disable(logging.WARNING)\n",
        "    os.environ[\"TF_CPP_MIN_LOG_LEVEL\"] = \"3\"\n",
        "    warnings.filterwarnings('ignore')\n",
        "\n",
        "loggings()\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Requirement already satisfied: pyyaml in /usr/local/lib/python3.7/dist-packages (3.13)\n",
            "Requirement already satisfied: h5py in /usr/local/lib/python3.7/dist-packages (2.10.0)\n",
            "Requirement already satisfied: numpy>=1.7 in /usr/local/lib/python3.7/dist-packages (from h5py) (1.19.5)\n",
            "Requirement already satisfied: six in /usr/local/lib/python3.7/dist-packages (from h5py) (1.15.0)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "J5DoolrX4uuk",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "9003f6c7-7c84-4b5c-96f7-c681ccc21296"
      },
      "source": [
        "from google.colab import files\n",
        "from google.colab import drive\n",
        "\n",
        "\n",
        "file = open(\"/content/drive/My Drive/Colab Notebooks/data_uca.pkl\" , \"rb\")\n",
        "\n",
        "\n",
        "print(\"pickle file open\")\n",
        "allX, allY = pickle.load(file) # allX = training data; allY = training label\n",
        "file.close()\n",
        "\n",
        "print ('file is ok')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "pickle file open\n",
            "file is ok\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "eTRC1OTexO5S",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "7021ee5f-288c-4472-957a-4598371c282f"
      },
      "source": [
        "from sklearn.model_selection import train_test_split\n",
        "from tensorflow.keras.callbacks import LearningRateScheduler\n",
        "x_train, x_test, y_train, y_test = train_test_split(allX, allY, test_size=0.25, random_state=42)\n",
        "\n",
        "y_train = to_categorical(y_train,4)\n",
        "y_test_cat = to_categorical(y_test,4)\n",
        "batch_size = 128\n",
        "epochs = 20\n",
        "\n",
        "print (\"number of training examples = \" + str(x_train.shape[0]))\n",
        "print (\"number of test examples = \" + str(x_test.shape[0]))\n",
        "\n",
        "\n",
        "y_integers = np.argmax(y_train, axis=1)\n",
        "#print(y_integers)\n",
        "class_weights = compute_class_weight('balanced', np.unique(y_integers), y_integers)\n",
        "#print(class_weights)\n",
        "d_class_weights = dict(enumerate(class_weights))\n",
        "#print(d_class_weights)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "number of training examples = 79905\n",
            "number of test examples = 26636\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ifT4TUZaZX2N",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "cec89731-26f5-4ea5-89fe-d853f22780e8"
      },
      "source": [
        "def cnn_model():\n",
        "  weight_decay = 0.0005 \n",
        "  model = Sequential()\n",
        "  model.add(Conv2D(32, (3, 3), padding='same', input_shape=(64,64,3), kernel_regularizer=regularizers.l2(weight_decay),activation='relu'))\n",
        "  model.add(BatchNormalization())\n",
        "  model.add(Dropout(0.3))\n",
        "  model.add(Conv2D(64, kernel_size=(3, 3), strides=(2,2), kernel_regularizer=regularizers.l2(weight_decay),activation='relu'))\n",
        "  model.add(BatchNormalization())\n",
        "  model.add(MaxPooling2D(pool_size=(2, 2)))\n",
        "  model.add(Conv2D(128, (3, 3), padding='same',kernel_regularizer=regularizers.l2(weight_decay),activation='relu'))\n",
        "  model.add(BatchNormalization())\n",
        "  model.add(Flatten())\n",
        "  model.add(Dense(512,kernel_regularizer=regularizers.l2(weight_decay),activation='relu'))\n",
        "  model.add(BatchNormalization())\n",
        "  model.add(Dropout(0.5))\n",
        "  model.add(Dense(4,activation = 'softmax'))\n",
        "  return (model)\n",
        "model = cnn_model() \n",
        "model.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "conv2d (Conv2D)              (None, 64, 64, 32)        896       \n",
            "_________________________________________________________________\n",
            "batch_normalization (BatchNo (None, 64, 64, 32)        128       \n",
            "_________________________________________________________________\n",
            "dropout (Dropout)            (None, 64, 64, 32)        0         \n",
            "_________________________________________________________________\n",
            "conv2d_1 (Conv2D)            (None, 31, 31, 64)        18496     \n",
            "_________________________________________________________________\n",
            "batch_normalization_1 (Batch (None, 31, 31, 64)        256       \n",
            "_________________________________________________________________\n",
            "max_pooling2d (MaxPooling2D) (None, 15, 15, 64)        0         \n",
            "_________________________________________________________________\n",
            "conv2d_2 (Conv2D)            (None, 15, 15, 128)       73856     \n",
            "_________________________________________________________________\n",
            "batch_normalization_2 (Batch (None, 15, 15, 128)       512       \n",
            "_________________________________________________________________\n",
            "flatten (Flatten)            (None, 28800)             0         \n",
            "_________________________________________________________________\n",
            "dense (Dense)                (None, 512)               14746112  \n",
            "_________________________________________________________________\n",
            "batch_normalization_3 (Batch (None, 512)               2048      \n",
            "_________________________________________________________________\n",
            "dropout_1 (Dropout)          (None, 512)               0         \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 4)                 2052      \n",
            "=================================================================\n",
            "Total params: 14,844,356\n",
            "Trainable params: 14,842,884\n",
            "Non-trainable params: 1,472\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "e21zPxp-ZmwP",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "25ab2aaf-eb49-4b01-f979-4ead50bd2f81"
      },
      "source": [
        "model = cnn_model()\n",
        "sgd = SGD(lr=0.01, momentum=0.9, decay=0.001, nesterov=False)\n",
        "model.compile(loss=keras.losses.categorical_crossentropy, optimizer=sgd, metrics=['accuracy'])\n",
        "init_loss = model.evaluate(x_test, y_test_cat)[0]"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "833/833 [==============================] - 15s 17ms/step - loss: 1.9517 - accuracy: 0.2691\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4l3IgHyLZt6e",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "f242bfb6-9a0f-4be9-c66e-30ce2653d952"
      },
      "source": [
        "loss_list = [init_loss]\n",
        "class history_(keras.callbacks.Callback): \n",
        "    def on_epoch_end(self, epoch, logs={}):\n",
        "        loss = model.evaluate(x_test, y_test_cat)[0]\n",
        "        loss_list.append(loss)\n",
        "        for i in range(epoch+1):\n",
        "          loss_t_minus = np.array(loss_list[i-2], np.float32)\n",
        "          loss_t = np.array(loss_list[i-1], np.float32)\n",
        "          beta = 0.90\n",
        "          momentum = beta / (1. + np.exp(-1. * abs(1. / np.sqrt(np.exp(-1*(loss_t * loss_t_minus))))))\n",
        "          print(\"loss\", loss)\n",
        "          print(\"momentum\", momentum)\n",
        "          return momentum\n",
        "loss_history_ = history_()\n",
        "\n",
        "sgd = SGD(lr=0.01, momentum=history_.on_epoch_end(epochs, epochs), nesterov=False)\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "833/833 [==============================] - 19s 22ms/step - loss: 1.9516 - accuracy: 0.2729\n",
            "loss 1.9516260623931885\n",
            "momentum 0.898910566849575\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BEq93wKyk9QF",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "d908f94a-dfb9-400e-ae23-f63c07b5e56a"
      },
      "source": [
        "earlyStopping = EarlyStopping(monitor='val_loss', patience=10, verbose=0, mode='min')\n",
        "\n",
        "checkpoint_path = \"/Users/utkucanaytac/Desktop/model\"\n",
        "checkpoint_dir = os.path.dirname(checkpoint_path)\n",
        "\n",
        "cp_callback = tf.keras.callbacks.ModelCheckpoint(filepath=checkpoint_path,\n",
        "                                                 save_best_only=True, save_weights_only=True,monitor = 'val_loss',\n",
        "                                                 verbose=1)\n",
        "callbacks_list_ = [loss_history_,earlyStopping,cp_callback]\n",
        "\n",
        "model.compile(loss=keras.losses.categorical_crossentropy, optimizer=sgd, metrics=['accuracy'])\n",
        "\n",
        "history = model.fit(x_train, y_train, \n",
        "     validation_data=(x_test, y_test_cat), \n",
        "     epochs=epochs, \n",
        "     batch_size=batch_size, \n",
        "     callbacks=callbacks_list_,\n",
        "     verbose = 1)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/20\n",
            "625/625 [==============================] - 133s 211ms/step - loss: 1.7403 - accuracy: 0.5497 - val_loss: 1.2285 - val_accuracy: 0.7345\n",
            "833/833 [==============================] - 13s 16ms/step - loss: 1.2285 - accuracy: 0.7345\n",
            "loss 1.2285151481628418\n",
            "momentum 0.8684804967117162\n",
            "\n",
            "Epoch 00001: val_loss improved from inf to 1.22851, saving model to /Users/utkucanaytac/Desktop/model\n",
            "Epoch 2/20\n",
            "625/625 [==============================] - 133s 213ms/step - loss: 1.1384 - accuracy: 0.7723 - val_loss: 0.9699 - val_accuracy: 0.8344\n",
            "833/833 [==============================] - 13s 16ms/step - loss: 0.9699 - accuracy: 0.8344\n",
            "loss 0.9698764681816101\n",
            "momentum 0.7739033617870735\n",
            "\n",
            "Epoch 00002: val_loss improved from 1.22851 to 0.96988, saving model to /Users/utkucanaytac/Desktop/model\n",
            "Epoch 3/20\n",
            "625/625 [==============================] - 134s 214ms/step - loss: 0.9429 - accuracy: 0.8425 - val_loss: 0.8307 - val_accuracy: 0.8807\n",
            "833/833 [==============================] - 13s 16ms/step - loss: 0.8307 - accuracy: 0.8807\n",
            "loss 0.8306760191917419\n",
            "momentum 0.7352857945964214\n",
            "\n",
            "Epoch 00003: val_loss improved from 0.96988 to 0.83068, saving model to /Users/utkucanaytac/Desktop/model\n",
            "Epoch 4/20\n",
            "625/625 [==============================] - 135s 216ms/step - loss: 0.8214 - accuracy: 0.8839 - val_loss: 0.7508 - val_accuracy: 0.9070\n",
            "833/833 [==============================] - 13s 16ms/step - loss: 0.7508 - accuracy: 0.9070\n",
            "loss 0.7508325576782227\n",
            "momentum 0.7170523528494643\n",
            "\n",
            "Epoch 00004: val_loss improved from 0.83068 to 0.75083, saving model to /Users/utkucanaytac/Desktop/model\n",
            "Epoch 5/20\n",
            "625/625 [==============================] - 139s 222ms/step - loss: 0.7355 - accuracy: 0.9075 - val_loss: 0.7175 - val_accuracy: 0.9118\n",
            "833/833 [==============================] - 15s 18ms/step - loss: 0.7175 - accuracy: 0.9118\n",
            "loss 0.7175431251525879\n",
            "momentum 0.7086336605098984\n",
            "\n",
            "Epoch 00005: val_loss improved from 0.75083 to 0.71754, saving model to /Users/utkucanaytac/Desktop/model\n",
            "Epoch 6/20\n",
            "625/625 [==============================] - 141s 226ms/step - loss: 0.6782 - accuracy: 0.9266 - val_loss: 0.6866 - val_accuracy: 0.9212\n",
            "833/833 [==============================] - 15s 18ms/step - loss: 0.6866 - accuracy: 0.9212\n",
            "loss 0.6865735650062561\n",
            "momentum 0.7040995003881895\n",
            "\n",
            "Epoch 00006: val_loss improved from 0.71754 to 0.68657, saving model to /Users/utkucanaytac/Desktop/model\n",
            "Epoch 7/20\n",
            "625/625 [==============================] - 138s 221ms/step - loss: 0.6367 - accuracy: 0.9374 - val_loss: 0.7253 - val_accuracy: 0.9037\n",
            "833/833 [==============================] - 14s 17ms/step - loss: 0.7253 - accuracy: 0.9037\n",
            "loss 0.725324809551239\n",
            "momentum 0.7046234587569375\n",
            "\n",
            "Epoch 00007: val_loss did not improve from 0.68657\n",
            "Epoch 8/20\n",
            "625/625 [==============================] - 135s 216ms/step - loss: 0.6043 - accuracy: 0.9477 - val_loss: 0.6371 - val_accuracy: 0.9343\n",
            "833/833 [==============================] - 14s 17ms/step - loss: 0.6371 - accuracy: 0.9343\n",
            "loss 0.6370843648910522\n",
            "momentum 0.7011107735256915\n",
            "\n",
            "Epoch 00008: val_loss improved from 0.68657 to 0.63708, saving model to /Users/utkucanaytac/Desktop/model\n",
            "Epoch 9/20\n",
            "625/625 [==============================] - 133s 213ms/step - loss: 0.5850 - accuracy: 0.9505 - val_loss: 0.8074 - val_accuracy: 0.8968\n",
            "833/833 [==============================] - 14s 17ms/step - loss: 0.8074 - accuracy: 0.8968\n",
            "loss 0.8073847889900208\n",
            "momentum 0.7062325436866438\n",
            "\n",
            "Epoch 00009: val_loss did not improve from 0.63708\n",
            "Epoch 10/20\n",
            "625/625 [==============================] - 135s 215ms/step - loss: 0.5638 - accuracy: 0.9545 - val_loss: 0.7296 - val_accuracy: 0.9050\n",
            "833/833 [==============================] - 13s 16ms/step - loss: 0.7296 - accuracy: 0.9050\n",
            "loss 0.7296409010887146\n",
            "momentum 0.7136140476410293\n",
            "\n",
            "Epoch 00010: val_loss did not improve from 0.63708\n",
            "Epoch 11/20\n",
            "625/625 [==============================] - 136s 218ms/step - loss: 0.5434 - accuracy: 0.9594 - val_loss: 0.6146 - val_accuracy: 0.9345\n",
            "833/833 [==============================] - 13s 16ms/step - loss: 0.6146 - accuracy: 0.9345\n",
            "loss 0.6146386861801147\n",
            "momentum 0.6997820266816814\n",
            "\n",
            "Epoch 00011: val_loss improved from 0.63708 to 0.61464, saving model to /Users/utkucanaytac/Desktop/model\n",
            "Epoch 12/20\n",
            "625/625 [==============================] - 135s 216ms/step - loss: 0.5145 - accuracy: 0.9659 - val_loss: 0.6109 - val_accuracy: 0.9359\n",
            "833/833 [==============================] - 14s 17ms/step - loss: 0.6109 - accuracy: 0.9359\n",
            "loss 0.6108519434928894\n",
            "momentum 0.6927117502727237\n",
            "\n",
            "Epoch 00012: val_loss improved from 0.61464 to 0.61085, saving model to /Users/utkucanaytac/Desktop/model\n",
            "Epoch 13/20\n",
            "625/625 [==============================] - 135s 216ms/step - loss: 0.5127 - accuracy: 0.9655 - val_loss: 0.5865 - val_accuracy: 0.9400\n",
            "833/833 [==============================] - 13s 16ms/step - loss: 0.5865 - accuracy: 0.9400\n",
            "loss 0.5864992737770081\n",
            "momentum 0.6910598768455486\n",
            "\n",
            "Epoch 00013: val_loss improved from 0.61085 to 0.58650, saving model to /Users/utkucanaytac/Desktop/model\n",
            "Epoch 14/20\n",
            "625/625 [==============================] - 135s 217ms/step - loss: 0.4948 - accuracy: 0.9678 - val_loss: 0.5648 - val_accuracy: 0.9482\n",
            "833/833 [==============================] - 14s 17ms/step - loss: 0.5648 - accuracy: 0.9482\n",
            "loss 0.564750075340271\n",
            "momentum 0.6884717977362946\n",
            "\n",
            "Epoch 00014: val_loss improved from 0.58650 to 0.56475, saving model to /Users/utkucanaytac/Desktop/model\n",
            "Epoch 15/20\n",
            "625/625 [==============================] - 137s 219ms/step - loss: 0.4842 - accuracy: 0.9691 - val_loss: 0.5679 - val_accuracy: 0.9410\n",
            "833/833 [==============================] - 14s 17ms/step - loss: 0.5679 - accuracy: 0.9410\n",
            "loss 0.5679117441177368\n",
            "momentum 0.6874705185422612\n",
            "\n",
            "Epoch 00015: val_loss did not improve from 0.56475\n",
            "Epoch 16/20\n",
            "625/625 [==============================] - 136s 217ms/step - loss: 0.4912 - accuracy: 0.9667 - val_loss: 0.6003 - val_accuracy: 0.9312\n",
            "833/833 [==============================] - 14s 17ms/step - loss: 0.6003 - accuracy: 0.9312\n",
            "loss 0.6002951264381409\n",
            "momentum 0.6893977415691743\n",
            "\n",
            "Epoch 00016: val_loss did not improve from 0.56475\n",
            "Epoch 17/20\n",
            "625/625 [==============================] - 138s 221ms/step - loss: 0.4830 - accuracy: 0.9699 - val_loss: 0.5792 - val_accuracy: 0.9368\n",
            "833/833 [==============================] - 14s 17ms/step - loss: 0.5792 - accuracy: 0.9368\n",
            "loss 0.5791895985603333\n",
            "momentum 0.6900457028970646\n",
            "\n",
            "Epoch 00017: val_loss did not improve from 0.56475\n",
            "Epoch 18/20\n",
            "625/625 [==============================] - 136s 217ms/step - loss: 0.4670 - accuracy: 0.9729 - val_loss: 0.5871 - val_accuracy: 0.9432\n",
            "833/833 [==============================] - 14s 17ms/step - loss: 0.5871 - accuracy: 0.9432\n",
            "loss 0.587140679359436\n",
            "momentum 0.689316550903217\n",
            "\n",
            "Epoch 00018: val_loss did not improve from 0.56475\n",
            "Epoch 19/20\n",
            "625/625 [==============================] - 136s 217ms/step - loss: 0.4657 - accuracy: 0.9721 - val_loss: 0.5761 - val_accuracy: 0.9337\n",
            "833/833 [==============================] - 13s 16ms/step - loss: 0.5761 - accuracy: 0.9337\n",
            "loss 0.5761176347732544\n",
            "momentum 0.6891440844140218\n",
            "\n",
            "Epoch 00019: val_loss did not improve from 0.56475\n",
            "Epoch 20/20\n",
            "625/625 [==============================] - 136s 218ms/step - loss: 0.4623 - accuracy: 0.9722 - val_loss: 0.5411 - val_accuracy: 0.9464\n",
            "833/833 [==============================] - 13s 16ms/step - loss: 0.5411 - accuracy: 0.9464\n",
            "loss 0.5411344766616821\n",
            "momentum 0.6866163977822795\n",
            "\n",
            "Epoch 00020: val_loss improved from 0.56475 to 0.54113, saving model to /Users/utkucanaytac/Desktop/model\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "JUhGVPz4lC0M",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "06f7ad3e-ede7-4a3b-b8a7-f668f2dc7366"
      },
      "source": [
        "K.clear_session()\n",
        "model2 = cnn_model()\n",
        "\n",
        "sgd_old = SGD(learning_rate=0.01, nesterov=False)\n",
        "\n",
        "model2.compile(loss=keras.losses.categorical_crossentropy, optimizer=sgd_old, metrics=['accuracy'])\n",
        "history2 = model2.fit(x_train, y_train, \n",
        "     validation_data=(x_test, y_test_cat), \n",
        "     epochs=epochs, \n",
        "     batch_size=batch_size, \n",
        "     verbose = 1)\n",
        "\n",
        "# fit CNN model using Adadelta optimizer\n",
        "\n",
        "K.clear_session()\n",
        "model3 = cnn_model()\n",
        "model3.compile(loss=keras.losses.categorical_crossentropy,\n",
        "              optimizer=Adadelta(lr=1.0, rho=0.95, epsilon=1e-08, decay=0.0),\n",
        "              metrics=['accuracy'])\n",
        "history3 = model3.fit(x_train, y_train, \n",
        "                     validation_data=(x_test, y_test_cat), \n",
        "                     epochs=epochs, \n",
        "                     verbose=1)\n",
        "\n",
        "# fit CNN model using RMSprop optimizer\n",
        "K.clear_session()\n",
        "model4 = cnn_model()\n",
        "model4.compile(loss=keras.losses.categorical_crossentropy,\n",
        "              optimizer=RMSprop(lr=0.001, rho=0.9, epsilon=1e-08, decay=0.0),\n",
        "              metrics=['accuracy'])\n",
        "history4 = model4.fit(x_train, y_train, \n",
        "                     validation_data=(x_test, y_test_cat), \n",
        "                     epochs=epochs, \n",
        "                     verbose=1)\n",
        "\n",
        "# fit CNN model using Adam optimizer\n",
        "K.clear_session()\n",
        "model5 = cnn_model()\n",
        "model5.compile(loss=keras.losses.categorical_crossentropy,\n",
        "              optimizer=Adam(lr=0.001, beta_1=0.9, beta_2=0.999, epsilon=1e-08, decay=0.0),\n",
        "              metrics=['accuracy'])\n",
        "history5 = model5.fit(x_train, y_train, \n",
        "                     validation_data=(x_test, y_test_cat), \n",
        "                     epochs=epochs, \n",
        "                     verbose=1)\n",
        "\n",
        "# fit CNN model using Adagrad optimizer\n",
        "K.clear_session()\n",
        "model6 = cnn_model()\n",
        "model6.compile(loss=keras.losses.categorical_crossentropy,\n",
        "              optimizer=Adagrad(lr=0.01, epsilon=1e-08, decay=0.0),\n",
        "              metrics=['accuracy'])\n",
        "history6 = model6.fit(x_train, y_train, \n",
        "                     validation_data=(x_test, y_test_cat), \n",
        "                     epochs=epochs, \n",
        "                     verbose=2)\n",
        "\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/20\n",
            "625/625 [==============================] - 135s 214ms/step - loss: 1.8698 - accuracy: 0.5217 - val_loss: 1.3630 - val_accuracy: 0.7000\n",
            "Epoch 2/20\n",
            "625/625 [==============================] - 131s 210ms/step - loss: 1.2496 - accuracy: 0.7323 - val_loss: 1.1344 - val_accuracy: 0.7741\n",
            "Epoch 3/20\n",
            "625/625 [==============================] - 131s 210ms/step - loss: 1.0895 - accuracy: 0.7957 - val_loss: 1.0221 - val_accuracy: 0.8210\n",
            "Epoch 4/20\n",
            "625/625 [==============================] - 130s 208ms/step - loss: 0.9894 - accuracy: 0.8350 - val_loss: 0.9153 - val_accuracy: 0.8576\n",
            "Epoch 5/20\n",
            "625/625 [==============================] - 131s 209ms/step - loss: 0.9156 - accuracy: 0.8617 - val_loss: 0.8550 - val_accuracy: 0.8831\n",
            "Epoch 6/20\n",
            "625/625 [==============================] - 131s 210ms/step - loss: 0.8511 - accuracy: 0.8838 - val_loss: 0.8206 - val_accuracy: 0.8945\n",
            "Epoch 7/20\n",
            "625/625 [==============================] - 135s 216ms/step - loss: 0.8018 - accuracy: 0.9016 - val_loss: 0.8011 - val_accuracy: 0.8986\n",
            "Epoch 8/20\n",
            "625/625 [==============================] - 138s 220ms/step - loss: 0.7601 - accuracy: 0.9175 - val_loss: 0.7638 - val_accuracy: 0.9146\n",
            "Epoch 9/20\n",
            "625/625 [==============================] - 140s 223ms/step - loss: 0.7281 - accuracy: 0.9265 - val_loss: 0.7513 - val_accuracy: 0.9135\n",
            "Epoch 10/20\n",
            "625/625 [==============================] - 137s 220ms/step - loss: 0.6946 - accuracy: 0.9360 - val_loss: 0.7232 - val_accuracy: 0.9261\n",
            "Epoch 11/20\n",
            "625/625 [==============================] - 136s 218ms/step - loss: 0.6690 - accuracy: 0.9455 - val_loss: 0.7164 - val_accuracy: 0.9244\n",
            "Epoch 12/20\n",
            "625/625 [==============================] - 140s 224ms/step - loss: 0.6497 - accuracy: 0.9508 - val_loss: 0.6956 - val_accuracy: 0.9306\n",
            "Epoch 13/20\n",
            "625/625 [==============================] - 138s 221ms/step - loss: 0.6314 - accuracy: 0.9546 - val_loss: 0.6777 - val_accuracy: 0.9385\n",
            "Epoch 14/20\n",
            "625/625 [==============================] - 139s 223ms/step - loss: 0.6093 - accuracy: 0.9623 - val_loss: 0.6638 - val_accuracy: 0.9424\n",
            "Epoch 15/20\n",
            "625/625 [==============================] - 139s 222ms/step - loss: 0.5950 - accuracy: 0.9650 - val_loss: 0.6488 - val_accuracy: 0.9464\n",
            "Epoch 16/20\n",
            "625/625 [==============================] - 135s 217ms/step - loss: 0.5810 - accuracy: 0.9682 - val_loss: 0.6434 - val_accuracy: 0.9454\n",
            "Epoch 17/20\n",
            "625/625 [==============================] - 135s 217ms/step - loss: 0.5707 - accuracy: 0.9703 - val_loss: 0.6609 - val_accuracy: 0.9382\n",
            "Epoch 18/20\n",
            "625/625 [==============================] - 140s 224ms/step - loss: 0.5600 - accuracy: 0.9712 - val_loss: 0.6295 - val_accuracy: 0.9488\n",
            "Epoch 19/20\n",
            "625/625 [==============================] - 140s 224ms/step - loss: 0.5473 - accuracy: 0.9760 - val_loss: 0.6221 - val_accuracy: 0.9491\n",
            "Epoch 20/20\n",
            "625/625 [==============================] - 135s 216ms/step - loss: 0.5346 - accuracy: 0.9783 - val_loss: 0.6122 - val_accuracy: 0.9515\n",
            "Epoch 1/20\n",
            "2498/2498 [==============================] - 260s 104ms/step - loss: 1.7424 - accuracy: 0.5746 - val_loss: 2.1045 - val_accuracy: 0.6346\n",
            "Epoch 2/20\n",
            "2498/2498 [==============================] - 258s 103ms/step - loss: 1.0100 - accuracy: 0.7849 - val_loss: 0.8036 - val_accuracy: 0.8420\n",
            "Epoch 3/20\n",
            "2498/2498 [==============================] - 251s 101ms/step - loss: 0.8149 - accuracy: 0.8436 - val_loss: 0.7149 - val_accuracy: 0.8729\n",
            "Epoch 4/20\n",
            "2498/2498 [==============================] - 253s 101ms/step - loss: 0.7323 - accuracy: 0.8685 - val_loss: 0.6665 - val_accuracy: 0.8911\n",
            "Epoch 5/20\n",
            "2498/2498 [==============================] - 256s 103ms/step - loss: 0.6740 - accuracy: 0.8887 - val_loss: 0.6456 - val_accuracy: 0.8956\n",
            "Epoch 6/20\n",
            "2498/2498 [==============================] - 253s 101ms/step - loss: 0.6379 - accuracy: 0.9007 - val_loss: 0.6465 - val_accuracy: 0.8943\n",
            "Epoch 7/20\n",
            "2498/2498 [==============================] - 259s 104ms/step - loss: 0.6108 - accuracy: 0.9082 - val_loss: 0.6245 - val_accuracy: 0.8993\n",
            "Epoch 8/20\n",
            "2498/2498 [==============================] - 258s 103ms/step - loss: 0.5942 - accuracy: 0.9139 - val_loss: 0.6221 - val_accuracy: 0.9061\n",
            "Epoch 9/20\n",
            "2498/2498 [==============================] - 264s 106ms/step - loss: 0.5723 - accuracy: 0.9201 - val_loss: 0.5974 - val_accuracy: 0.9139\n",
            "Epoch 10/20\n",
            "2498/2498 [==============================] - 262s 105ms/step - loss: 0.5625 - accuracy: 0.9234 - val_loss: 0.5741 - val_accuracy: 0.9173\n",
            "Epoch 11/20\n",
            "2498/2498 [==============================] - 262s 105ms/step - loss: 0.5481 - accuracy: 0.9263 - val_loss: 0.5723 - val_accuracy: 0.9172\n",
            "Epoch 12/20\n",
            "2498/2498 [==============================] - 270s 108ms/step - loss: 0.5351 - accuracy: 0.9303 - val_loss: 0.5647 - val_accuracy: 0.9203\n",
            "Epoch 13/20\n",
            "2498/2498 [==============================] - 259s 104ms/step - loss: 0.5343 - accuracy: 0.9311 - val_loss: 0.5737 - val_accuracy: 0.9205\n",
            "Epoch 14/20\n",
            "2498/2498 [==============================] - 257s 103ms/step - loss: 0.5268 - accuracy: 0.9320 - val_loss: 0.5717 - val_accuracy: 0.9155\n",
            "Epoch 15/20\n",
            "2498/2498 [==============================] - 248s 99ms/step - loss: 0.5274 - accuracy: 0.9324 - val_loss: 0.6606 - val_accuracy: 0.9088\n",
            "Epoch 16/20\n",
            "2498/2498 [==============================] - 251s 100ms/step - loss: 0.5046 - accuracy: 0.9381 - val_loss: 0.5429 - val_accuracy: 0.9251\n",
            "Epoch 17/20\n",
            "2498/2498 [==============================] - 250s 100ms/step - loss: 0.5055 - accuracy: 0.9372 - val_loss: 0.5814 - val_accuracy: 0.9214\n",
            "Epoch 18/20\n",
            "2498/2498 [==============================] - 253s 101ms/step - loss: 0.5032 - accuracy: 0.9378 - val_loss: 0.5716 - val_accuracy: 0.9210\n",
            "Epoch 19/20\n",
            "2498/2498 [==============================] - 254s 102ms/step - loss: 0.5014 - accuracy: 0.9377 - val_loss: 0.5466 - val_accuracy: 0.9252\n",
            "Epoch 20/20\n",
            "2498/2498 [==============================] - 261s 104ms/step - loss: 0.4849 - accuracy: 0.9405 - val_loss: 0.5768 - val_accuracy: 0.9220\n",
            "Epoch 1/20\n",
            "2498/2498 [==============================] - 295s 118ms/step - loss: 1.7176 - accuracy: 0.5481 - val_loss: 1.1136 - val_accuracy: 0.7077\n",
            "Epoch 2/20\n",
            "2498/2498 [==============================] - 280s 112ms/step - loss: 1.0492 - accuracy: 0.7324 - val_loss: 0.9277 - val_accuracy: 0.7690\n",
            "Epoch 3/20\n",
            "2498/2498 [==============================] - 277s 111ms/step - loss: 0.9436 - accuracy: 0.7644 - val_loss: 0.8567 - val_accuracy: 0.7907\n",
            "Epoch 4/20\n",
            "2498/2498 [==============================] - 276s 110ms/step - loss: 0.8872 - accuracy: 0.7763 - val_loss: 0.7989 - val_accuracy: 0.8007\n",
            "Epoch 5/20\n",
            "2498/2498 [==============================] - 275s 110ms/step - loss: 0.8420 - accuracy: 0.7872 - val_loss: 0.7836 - val_accuracy: 0.8007\n",
            "Epoch 6/20\n",
            "2498/2498 [==============================] - 266s 106ms/step - loss: 0.8241 - accuracy: 0.7912 - val_loss: 0.7608 - val_accuracy: 0.8080\n",
            "Epoch 7/20\n",
            "2498/2498 [==============================] - 266s 107ms/step - loss: 0.7979 - accuracy: 0.8005 - val_loss: 0.8084 - val_accuracy: 0.7960\n",
            "Epoch 8/20\n",
            "2498/2498 [==============================] - 255s 102ms/step - loss: 0.7817 - accuracy: 0.8046 - val_loss: 0.7838 - val_accuracy: 0.8093\n",
            "Epoch 9/20\n",
            "2498/2498 [==============================] - 253s 101ms/step - loss: 0.7648 - accuracy: 0.8070 - val_loss: 0.7173 - val_accuracy: 0.8189\n",
            "Epoch 10/20\n",
            "2498/2498 [==============================] - 255s 102ms/step - loss: 0.7537 - accuracy: 0.8114 - val_loss: 0.7864 - val_accuracy: 0.7968\n",
            "Epoch 11/20\n",
            "2498/2498 [==============================] - 257s 103ms/step - loss: 0.7434 - accuracy: 0.8129 - val_loss: 0.7712 - val_accuracy: 0.8133\n",
            "Epoch 12/20\n",
            "2498/2498 [==============================] - 254s 102ms/step - loss: 0.7426 - accuracy: 0.8122 - val_loss: 0.7211 - val_accuracy: 0.8205\n",
            "Epoch 13/20\n",
            "2498/2498 [==============================] - 253s 101ms/step - loss: 0.7404 - accuracy: 0.8093 - val_loss: 0.6797 - val_accuracy: 0.8287\n",
            "Epoch 14/20\n",
            "2498/2498 [==============================] - 257s 103ms/step - loss: 0.7171 - accuracy: 0.8198 - val_loss: 0.6842 - val_accuracy: 0.8294\n",
            "Epoch 15/20\n",
            "2498/2498 [==============================] - 259s 104ms/step - loss: 0.7149 - accuracy: 0.8204 - val_loss: 0.6817 - val_accuracy: 0.8231\n",
            "Epoch 16/20\n",
            "2498/2498 [==============================] - 259s 104ms/step - loss: 0.7134 - accuracy: 0.8196 - val_loss: 0.6693 - val_accuracy: 0.8255\n",
            "Epoch 17/20\n",
            "2498/2498 [==============================] - 259s 104ms/step - loss: 0.7059 - accuracy: 0.8215 - val_loss: 0.7052 - val_accuracy: 0.8239\n",
            "Epoch 18/20\n",
            "2498/2498 [==============================] - 268s 107ms/step - loss: 0.7013 - accuracy: 0.8244 - val_loss: 0.6824 - val_accuracy: 0.8276\n",
            "Epoch 19/20\n",
            "2498/2498 [==============================] - 260s 104ms/step - loss: 0.7071 - accuracy: 0.8231 - val_loss: 0.6648 - val_accuracy: 0.8359\n",
            "Epoch 20/20\n",
            "2498/2498 [==============================] - 259s 104ms/step - loss: 0.7030 - accuracy: 0.8219 - val_loss: 0.6781 - val_accuracy: 0.8315\n",
            "Epoch 1/20\n",
            "2498/2498 [==============================] - 237s 94ms/step - loss: 2.0177 - accuracy: 0.5445 - val_loss: 1.6860 - val_accuracy: 0.7230\n",
            "Epoch 2/20\n",
            "2498/2498 [==============================] - 239s 96ms/step - loss: 1.6679 - accuracy: 0.7391 - val_loss: 1.5275 - val_accuracy: 0.7978\n",
            "Epoch 3/20\n",
            "2498/2498 [==============================] - 250s 100ms/step - loss: 1.5212 - accuracy: 0.7852 - val_loss: 1.4213 - val_accuracy: 0.8104\n",
            "Epoch 4/20\n",
            "2498/2498 [==============================] - 241s 97ms/step - loss: 1.4571 - accuracy: 0.8009 - val_loss: 1.3840 - val_accuracy: 0.8189\n",
            "Epoch 5/20\n",
            "2498/2498 [==============================] - 239s 95ms/step - loss: 1.3673 - accuracy: 0.8136 - val_loss: 1.2362 - val_accuracy: 0.8387\n",
            "Epoch 6/20\n",
            "2498/2498 [==============================] - 236s 94ms/step - loss: 1.2719 - accuracy: 0.8280 - val_loss: 1.3117 - val_accuracy: 0.8287\n",
            "Epoch 7/20\n",
            "2498/2498 [==============================] - 247s 99ms/step - loss: 1.2649 - accuracy: 0.8267 - val_loss: 1.1475 - val_accuracy: 0.8513\n",
            "Epoch 8/20\n",
            "2498/2498 [==============================] - 269s 108ms/step - loss: 1.1716 - accuracy: 0.8371 - val_loss: 1.0641 - val_accuracy: 0.8615\n",
            "Epoch 9/20\n",
            "2498/2498 [==============================] - 266s 106ms/step - loss: 1.1141 - accuracy: 0.8434 - val_loss: 1.0492 - val_accuracy: 0.8586\n",
            "Epoch 10/20\n",
            "2498/2498 [==============================] - 253s 101ms/step - loss: 1.0989 - accuracy: 0.8447 - val_loss: 1.0791 - val_accuracy: 0.8526\n",
            "Epoch 11/20\n",
            "2498/2498 [==============================] - 249s 100ms/step - loss: 1.0750 - accuracy: 0.8478 - val_loss: 0.9805 - val_accuracy: 0.8730\n",
            "Epoch 12/20\n",
            "2498/2498 [==============================] - 245s 98ms/step - loss: 1.0383 - accuracy: 0.8507 - val_loss: 0.9726 - val_accuracy: 0.8685\n",
            "Epoch 13/20\n",
            "2498/2498 [==============================] - 241s 96ms/step - loss: 1.0305 - accuracy: 0.8514 - val_loss: 0.9911 - val_accuracy: 0.8702\n",
            "Epoch 14/20\n",
            "2498/2498 [==============================] - 245s 98ms/step - loss: 1.0213 - accuracy: 0.8532 - val_loss: 0.9717 - val_accuracy: 0.8754\n",
            "Epoch 15/20\n",
            "2498/2498 [==============================] - 245s 98ms/step - loss: 1.0203 - accuracy: 0.8532 - val_loss: 0.9349 - val_accuracy: 0.8811\n",
            "Epoch 16/20\n",
            "2498/2498 [==============================] - 245s 98ms/step - loss: 0.9906 - accuracy: 0.8572 - val_loss: 0.9403 - val_accuracy: 0.8808\n",
            "Epoch 17/20\n",
            "2498/2498 [==============================] - 234s 94ms/step - loss: 0.9976 - accuracy: 0.8535 - val_loss: 0.9233 - val_accuracy: 0.8750\n",
            "Epoch 18/20\n",
            "2498/2498 [==============================] - 238s 95ms/step - loss: 0.9797 - accuracy: 0.8573 - val_loss: 0.9407 - val_accuracy: 0.8765\n",
            "Epoch 19/20\n",
            "2498/2498 [==============================] - 234s 94ms/step - loss: 0.9674 - accuracy: 0.8581 - val_loss: 0.9082 - val_accuracy: 0.8810\n",
            "Epoch 20/20\n",
            "2498/2498 [==============================] - 231s 92ms/step - loss: 0.9601 - accuracy: 0.8602 - val_loss: 0.9270 - val_accuracy: 0.8782\n",
            "Epoch 1/20\n",
            "2498/2498 - 229s - loss: 1.4176 - accuracy: 0.6747 - val_loss: 1.0773 - val_accuracy: 0.7982\n",
            "Epoch 2/20\n",
            "2498/2498 - 224s - loss: 1.0228 - accuracy: 0.8174 - val_loss: 0.8988 - val_accuracy: 0.8576\n",
            "Epoch 3/20\n",
            "2498/2498 - 222s - loss: 0.8553 - accuracy: 0.8729 - val_loss: 0.7903 - val_accuracy: 0.8924\n",
            "Epoch 4/20\n",
            "2498/2498 - 222s - loss: 0.7459 - accuracy: 0.9052 - val_loss: 0.7200 - val_accuracy: 0.9109\n",
            "Epoch 5/20\n",
            "2498/2498 - 221s - loss: 0.6710 - accuracy: 0.9257 - val_loss: 0.6653 - val_accuracy: 0.9218\n",
            "Epoch 6/20\n",
            "2498/2498 - 224s - loss: 0.6112 - accuracy: 0.9394 - val_loss: 0.6178 - val_accuracy: 0.9335\n",
            "Epoch 7/20\n",
            "2498/2498 - 240s - loss: 0.5586 - accuracy: 0.9522 - val_loss: 0.5880 - val_accuracy: 0.9382\n",
            "Epoch 8/20\n",
            "2498/2498 - 225s - loss: 0.5145 - accuracy: 0.9610 - val_loss: 0.5631 - val_accuracy: 0.9380\n",
            "Epoch 9/20\n",
            "2498/2498 - 229s - loss: 0.4838 - accuracy: 0.9650 - val_loss: 0.5392 - val_accuracy: 0.9438\n",
            "Epoch 10/20\n",
            "2498/2498 - 233s - loss: 0.4553 - accuracy: 0.9696 - val_loss: 0.5255 - val_accuracy: 0.9433\n",
            "Epoch 11/20\n",
            "2498/2498 - 229s - loss: 0.4278 - accuracy: 0.9747 - val_loss: 0.5061 - val_accuracy: 0.9448\n",
            "Epoch 12/20\n",
            "2498/2498 - 232s - loss: 0.4102 - accuracy: 0.9756 - val_loss: 0.4954 - val_accuracy: 0.9468\n",
            "Epoch 13/20\n",
            "2498/2498 - 225s - loss: 0.3944 - accuracy: 0.9778 - val_loss: 0.4788 - val_accuracy: 0.9473\n",
            "Epoch 14/20\n",
            "2498/2498 - 225s - loss: 0.3753 - accuracy: 0.9799 - val_loss: 0.4647 - val_accuracy: 0.9490\n",
            "Epoch 15/20\n",
            "2498/2498 - 224s - loss: 0.3641 - accuracy: 0.9809 - val_loss: 0.4556 - val_accuracy: 0.9506\n",
            "Epoch 16/20\n",
            "2498/2498 - 225s - loss: 0.3523 - accuracy: 0.9817 - val_loss: 0.4492 - val_accuracy: 0.9503\n",
            "Epoch 17/20\n",
            "2498/2498 - 227s - loss: 0.3407 - accuracy: 0.9828 - val_loss: 0.4370 - val_accuracy: 0.9519\n",
            "Epoch 18/20\n",
            "2498/2498 - 226s - loss: 0.3288 - accuracy: 0.9840 - val_loss: 0.4399 - val_accuracy: 0.9494\n",
            "Epoch 19/20\n",
            "2498/2498 - 235s - loss: 0.3223 - accuracy: 0.9838 - val_loss: 0.4291 - val_accuracy: 0.9506\n",
            "Epoch 20/20\n",
            "2498/2498 - 234s - loss: 0.3140 - accuracy: 0.9844 - val_loss: 0.4215 - val_accuracy: 0.9498\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "EiYqpv1xZ8PQ",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 717
        },
        "outputId": "4bf0f7e3-0d36-45c9-ab14-f05cbf501b9f"
      },
      "source": [
        "fig = plt.figure(figsize=(12,8))\n",
        "plt.plot(range(epochs),history.history['val_loss'],label='Adaptive_Momentum')\n",
        "plt.plot(range(epochs),history2.history['val_loss'],label='SGD')\n",
        "#plt.plot(range(epochs),history3.history['val_loss'],label='Adadelta')\n",
        "plt.plot(range(epochs),history4.history['val_loss'],label='RMSprop')\n",
        "plt.plot(range(epochs),history5.history['val_loss'],label='Adam')\n",
        "#plt.plot(range(epochs),history6.history['val_accuracy'],label='Adagrad')\n",
        "\n",
        "plt.legend(loc=0)\n",
        "plt.xlabel('epochs')\n",
        "plt.xlim([0,epochs])\n",
        "plt.ylabel('accuracy on test set')\n",
        "plt.grid(True)\n",
        "plt.title(\"Comparing Model Accuracy\")\n",
        "plt.show()\n",
        "fig.savefig(r'C:\\Users\\ucan.aytac\\Desktop\\compare-accuracy.jpg')\n",
        "plt.close(fig)\n",
        "\n",
        "target_names = ['Oligodendroglioma', 'Glioblastoma', 'Astrocytoma','Unidentified']\n",
        "\n",
        "predictions = model.predict_classes(x_test)\n",
        "print('Classification Report of Proposed Model')\n",
        "print(classification_report(y_test,predictions,target_names=target_names))\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAtwAAAHwCAYAAACVL7i5AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nOzdd3RVVf7+8ffOTUhCSCWNTkRBpEsRexwFQWwUKSpKU9FBFAZUEKUIjoogVhRQwZmhiQwqDlZEvmKhiXQrHRJqIIGEtP37417yC0hCIPfmJOF5rXWXuafs89x941ofdvbZx1hrERERERER3/BzOoCIiIiISHmmgltERERExIdUcIuIiIiI+JAKbhERERERH1LBLSIiIiLiQyq4RURERER8SAW3iIgPGWPuMsZ87nSOojLG1DbGWGOMfxGO7WWM+bYkcomIlGUquEWkTDDG3GmMWWmMSTPG7DHGLDLGXOV0rjOx1v7HWtvWF20bY7YaYzKNMdGnbP/JUzTX9sV1z4YxppLnO1vkdBYREaeo4BaRUs8YMxiYBDwLxAE1gTeA25zMdSZFGSX2gi1Aj3zXbARULIHrFlVn4DjQxhgTX5IXLqH+FxE5IxXcIlKqGWPCgTHA36218621R621Wdbaj621Qz3HBBpjJhljdntek4wxgZ59icaYncaYx4wxez2j47cbY24yxvxqjDlojBme73qjjDHzjDFzjDGpxpjVxpgm+fY/YYz5w7NvozGmY759vYwxy4wxLxljDgCjTp124Rl57m+M+c0Yk2KMed0YYzz7XMaYCcaY/caYLcaYAUWY3vEv4J587+8F3ju1D40x7xlj9hljthljRhhj/PJd80XPNf8EOpzm3Lc9/bbLGDPWGOMq2reXl+dNYC1w9yltX2WM+c7TDzuMMb0824M9/bDNGHPYGPOtZ1uiMWbnKW1sNcbc4Pn5xHf3b2PMEaCXMaaVMeZ7zzX2GGNeM8ZUyHd+A2PMF57fg2RjzHBjTLwx5pgxpnK+4y719F/AWXx2ERFABbeIlH6XA0HAfws55kmgNdAUaAK0Akbk2x/vaaMa8DQwFXfx1xy4GnjKGJOQ7/jbgPeBKGAmsCBfofWH55xwYDTwb2NMlXznXgb8iXskflwBeW8GWgKNga7AjZ7t9wHtPZ/jUuD2Qj7zCT8AYcaY+p5CuDvw71OOedWT9wLgWtwFeu9817wZaAa0ALqccu50IBu40HNMW6BfEXJhjKkFJAL/8bzuOWXfIk+2GNyfeY1n94u4v5srcH8HjwG5Rbkm7u9uHhDhuWYOMAiIxv27dD3wkCdDKPAl8ClQ1fMZv7LWJgFLcH83J/QEZltrs4qYQ0QkjwpuESntKgP7rbXZhRxzFzDGWrvXWrsPdyHcM9/+LGCcp1iajbv4etlam2qt3QBsxF2on7DKWjvPc/xE3MV6awBr7fvW2t3W2lxr7RzgN9wF/gm7rbWvWmuzrbXpBeR9zlqbYq3dDnyNu9gEd4H3srV2p7X2EPDcGfrmhBOj3G2ATcCuEzvyFeHDPJ93KzAhX/90BSZZa3dYaw8C/8x3bhxwE/Co5y8Le4GXPO0VRU9grbV2I+5+b2CMaebZdyfwpbV2lucvFgestWs8I+99gEestbustTnW2u+stceLeM3vrbULPN9PurV2lbX2B8/3sRV4C/c/OsD9D40ka+0Ea22Gp39+9OybgWdE3tOHPXD3s4jIWdP8NhEp7Q4A0cYY/0KK7qrAtnzvt3m25bVhrc3x/HyiCE7Otz8dqJTv/Y4TP1hrcz3TGKoCGGPuAQYDtT2HVMJdwP/l3EIk5fv5WL5rVz3l/KK0Be5CcCmQwCnTSTzZAvhr/1Qr4Jr5j6vlOXePZ9YLuAdqiprrHtx/TcBau8sY8w3uKSY/ATVw/7XgVNG4/4Fzun1FcVI2Y0xd3P9oaoF7brs/sMqzu6AMAB8Cb3r+8lEPOGytXX6OmUTkPKcRbhEp7b7HfdNdYdMrduMuDk+o6dl2rmqc+MEz4lod2O2ZBjEVGABUttZGAOsBk+9cW4zr7vFc6y85CmOt3Yb75smbgPmn7N6Pe4T/1P45MQq+55Tr1Mz38w7cfR9trY3wvMKstQ3OlMkYcwVwETDMGJNkjEnCPd3mTs+c9B1AndOcuh/IKGDfUfLdEOoZeY455ZhT+38ysBm4yFobBgzn/39fO3BPs/kLa20GMBf3KHdPNLotIsWggltESjVr7WHc865f99zsWNEYE2CMaW+MecFz2CxghDEmxriXyHuav85jPhvNjTGdPIXho7iLzh+AENwF3T4AY0xvoGExrnOqucAjxphqxpgI4PGzOLcv8Ddr7dH8Gz0j+3OBccaYUM8/Ggbz//tnLjDQGFPdGBMJPJHv3D3A58AEY0yYMcbPGFPHGHMtZ3Yv8AVwCe4pM01x91Uw7nnq/wFuMMZ0Ncb4G2MqG2OaWmtzgXeAicaYqp6bOi837ptgfwWCjDEdPHPqRwCBZ8gRChwB0owxFwMP5tu3EKhijHnUuG+8DTXGXJZv/3tAL+BWVHCLSDGo4BaRUs9aOwF3kTgCd7G7A/co8wLPIWOBlbhXwlgHrPZsO1cfAt2AQ7hHNzt55hlvxD3/+XvcU1IaAcuKcZ1TTcVd4K7FPe3if7hvWMwp7CQAa+0f1tqVBex+GPfo8J/At7hvBH0n3zU/A37G3W+njpDfA1TAPc/9EO4bEqtQCGNMEO654a9aa5PyvbbgLlzv9cxfvwn4B3AQ9w2TJ+bRD8H9Pa7w7Hse8PP84+shYBruEfqjwEmrlpzGENzzxVM9n3XOiR3W2lTc895vwT3N5zfgunz7l+G+WXO1568IIiLnxFhbnL9+ioiUL8aYUcCF1tq7z3RsCWRpD7xpra11xoPFJ4wxi4GZ1tppTmcRkbJLI9wiIqWEZ63pmzxTLKoBIyl8OUTxIWNMS9zLM84507EiIoVRwS0iUnoY3EsaHsI9pWQT7vnoUsKMMTNwr9H9qGfqiYjIOdOUEhERERERH9IIt4iIiIiID6ngFhERERHxoTL3pMmIiAh74YUXOh2j3Dh69CghISFOxyg31J/eo770LvWnd6k/vUd96V3qT+9atWrVfmvtqQ/YOmtlruCOi4tj5cqClpqVs7VkyRISExOdjlFuqD+9R33pXepP71J/eo/60rvUn95ljPHKGvyaUiIiIiIi4kMquEVEREREfEgFt4iIiIiID5W5OdwiIiIihcnKymLnzp1kZGQ4HaXEhYeHs2nTJqdjlDlBQUFUr16dgIAAn7SvgltERETKlZ07dxIaGkrt2rUxxjgdp0SlpqYSGhrqdIwyxVrLgQMH2LlzJwkJCT65hqaUiIiISLmSkZFB5cqVz7tiW86NMYbKlSv79C8iKrhFRESk3FGxLWfD178vKrhFRERERHxIBbeIiIiIDyxYsABjDJs3bz7t/sTExHN+mN+CBQvYuHFj3vunn36aL7/88pzaOp3ExERq1qyJtTZv2+23306lSpW8do2zkZKSwhtvvOHItb1BBbeIiIiID8yaNYurrrqKWbNmeb3tUwvuMWPGcMMNN3j1GhERESxbtgxwF7x79uzxavtno6wX3FqlRERERMqt0R9vYOPuI15t85KqYYy8pUGhx6SlpfHtt9/y9ddfc8sttzB69GjS09Pp3bs3P//8MxdffDHp6el5xz/44IOsWLGC9PR0unTpwujRowGoXbs2Xbt2ZdGiRQQHBzNz5kz27t3LRx99xDfffMPYsWP54IMPeOaZZ7j55pvx8/Nj1qxZvP/++4D7Ue8vvvgiCxcu5PPPP2fkyJEcP36cOnXq8O677xY6Yt29e3dmz57NVVddxfz58+nUqRMbNmwA3Ct7PPbYYyxatAhjDCNGjKBbt24sWbKEkSNHEhERwbp16+jatSuNGjXi5ZdfJj09nQULFlCnTh327dtH//792b59OwCTJk3iyiuvZNSoUWzfvp0///yT7du38+ijjzJw4ECeeOIJ/vjjD5o2bUqbNm3o0KFD3ucCGDBgAC1atKBXr17Url2bHj16sGjRIvz9/ZkyZQrDhg3j999/Z+jQofTv3//cv/xzpBFuERERES/78MMPadeuHXXr1qVy5cqsWrWKyZMnU7FiRTZt2sTo0aNZtWpV3vHjxo1j5cqVrF27lm+++Ya1a9fm7QsPD2fdunUMGDCARx99lCuuuIJbb72V8ePHs2bNGurUqZN37HXXXcePP/7I0aNHAZgzZw7du3dn//79jB07li+//JLVq1fTokULJk6cWOhnuP7661m6dCk5OTnMnj2bbt265e2bP38+a9as4eeff+bLL79k6NCheSPgP//8M2+++SabNm3iX//6F7/++ivLly+nX79+vPrqqwA88sgjDBo0iBUrVvDBBx/Qr1+/vLY3b97MZ599xvLlyxk9ejRZWVk899xz1KlThzVr1jB+/Pgz9n/NmjVZs2YNV199Nb169WLevHn88MMPjBw58ozn+oJGuEVERKTcOtNItK/MmjWLRx55BHCPFM+aNYvff/+dgQMHAtC4cWMaN26cd/zcuXOZMmUK2dnZ7Nmzh40bN+bt79GjR95/Bw0aVOh1/f39adeuHR9//DFdunThk08+4YUXXuCbb75h48aNXHnllQBkZmZy+eWXF9qWy+XiqquuYvbs2aSnp1O7du28fd9++y09evTA5XIRFxfHtddey4oVKwgLC6Nly5ZUqVIFgDp16tC2bVsAGjVqxNdffw3Al19+edKUmCNHjpCWlgZAhw4dCAwMJDAwkNjYWJKTkwvNeTq33npr3jXT0tIIDQ0lNDSUwMBAUlJSiIiIOOs2i0MFt4iIiIgXHTx4kMWLF7Nu3TqMMeTk5GCMoVmzZqc9fsuWLbz44ousWLGCyMhIevXqddKa0PmXrCvK8nXdu3fntddeIyoqihYtWhAaGoq1ljZt2pz1fPLu3bvTsWNHRo0aVeRzAgMD83728/PLe+/n50d2djYAubm5/PDDDwQFBRV6vsvlyjsnP39/f3Jzc/Pen7qGdv5rnprndO35mqaUiIiIiHjRvHnz6NmzJ9u2bWPr1q3s2LGDhIQEmjdvzsyZMwFYv3593rSRI0eOEBISQnh4OMnJySxatOik9ubMmZP33xOj0qGhoaSmpp72+tdeey2rV69m6tSpdO/eHYDWrVuzbNkyfv/9dwCOHj3Kr7/+esbPcvXVVzNs2LC8Ufb82+fMmUNOTg779u1j6dKltGrVqqhdRNu2bfOmlwCsWbOm0ONP/by1atVi48aNHD9+nJSUFL766qsiX9sJGuEWERER8aJZs2bx+OOPn7Stc+fO/PTTT6Snp1O/fn3q169P8+bNAWjSpAnNmjXj4osvpkaNGnnTPk44dOgQjRs3JjAwMG+Eunv37tx333288sorzJs376TjXS4XN998M9OnT2fGjBkAxMTEMH36dHr06MHx48cBGDt2LHXr1i30sxhjGDJkyF+2d+zYke+//54mTZpgjOGFF14gPj6+wCUQT/XKK6/w97//ncaNG5Odnc0111zDm2++WeDxlStX5sorr6Rhw4a0b9+e8ePH07VrVxo2bEhCQkKBfz0oLUz+9RXLgovr1LGb//jD6RjlxpIlS0hMTHQ6Rrmh/vQe9aV3qT+9S/3pPb7oy02bNlG/fn2vtumU2rVrs3LlSqKjo4t0fGpqKqGhoT5OVT6d7vfGGLPKWtuiuG2XuSklrqQkco54d3kfERERERFfKXMFt8nNZd/LrzgdQ0RERMTntm7dWuTR7XPRsWNHmjZtetLrs88+89n1zldlbg53bqVQDs2aRXjHjgQ3dGapHxEREZHy4L///a/TEc4LZW6EOyciHFdUFEmjR2NzcpyOIyIiIiJSqDJXcOPnR9zjj5Oxbh0pc+c6nUZEREREpFBlr+AGwm7uQMXWrdk78SWy9+93Oo6IiIiISIHKZMFtjCH+6afIzchg7/jxTscRERER+Ytx48bRoEEDGjduTNOmTfnxxx/Jzs5m+PDhXHTRRXk3KY4bNy7vHJfLRdOmTWnQoAFNmjRhwoQJJz1RUcqmMnfT5AmBF1xA5b59OPDmW4R37kzIWTzdSERERMSXvv/+exYuXMjq1asJDAxk//79ZGZmMmLECJKSkli3bh1BQUGkpqYyYcKEvPOCg4Pznrq4d+9e7rzzTo4cOcLo0aOd+ijiBWVyhPuE6AceIKBaNZJGj8FmZjodR0RERASAPXv2EB0dTWBgIADR0dFEREQwdepUXn31VYKCggD3I8tHjRp12jZiY2OZMmUKr732GmXtQYVysjI7wg3gFxxM3Ign2fngQxyYMYPo++5zOpKIiIiUJouegKR13m0zvhG0f67QQ9q2bcuYMWOoW7cuN9xwA926dSMyMpKaNWue1ZMgL7jgAnJycti7dy9xcXHFTS4OKdMj3ACh111HpRuuZ/8bk8natcvpOCIiIiJUqlSJVatWMWXKFGJiYujWrRtLliw56Zh3332Xpk2bUqNGDXbs2OFMUCkRZXqE+4T44cP5o8PNJI17lhpvvO50HBERESktzjAS7Usul4vExEQSExNp1KgRb731Ftu3byc1NZXQ0FB69+5N7969adiwITkFPFvkzz//xOVyERsbW8LpxZvK/Ag3QEDVqsT8/SHSFi8mdfFip+OIiIjIee6XX37ht99+y3u/Zs0a6tWrR9++fRkwYAAZGRkA5OTkkFnAfWj79u2jf//+DBgwAGNMieQW3ygXI9wAUffeS8qCBSSPHUdI69b4VazodCQRERE5T6WlpfHwww+TkpKCv78/F154IVOmTCE8PJynnnqKhg0bEhoaSnBwMPfeey9Vq1YFID09naZNm5KVlYW/vz89e/Zk8ODBDn8aKa5yU3CbgACqjBzJtp73sP/Nt4gdPMjpSCIiInKeat68Od99991p9z333HM899zpp7oUNLVEyrZyMaXkhIotWxJ+++0cePddjv/xh9NxRERERETKV8ENEPvYUPwqVnSvza01K0VERETEYeWu4PaPiiJ20CCOLV/OkY8/djqOiIiIiJznyl3BDRDR9Q6CGjcm+fkXyDlyxOk4IiIiInIeK5cFt/Hzo8qokeQcOsS+SZOcjiMiIiIi57FyWXADBF1yCZF33cWhWbNJX7fe6TgiIiIicp4qtwU3QMzAh3FFVyZp1CisltkRERGREuJyuWjatCkNGzbklltuISUlBYCtW7dijGHEiBF5x+7fv5+AgAAGDBgAuB+ak5iYSNOmTalfvz7333+/I59BvKdcF9yu0FDinniCjA0bODRnjtNxRERE5DwRHBzMmjVrWL9+PVFRUbz++ut5+xISEvjkk0/y3r///vs0aNAg7/3AgQMZNGgQa9asYdOmTTz88MNFvq61ltzcXO98CPGacl1wA4TddBMVL2/Nvpcmkb1vn9NxRERE5Dxz+eWXs2vXrrz3FStWpH79+qxcuRKAOXPm0LVr17z9e/bsoXr16nnvGzVqBMD06dO57bbbSExM5KKLLmL06NGAe9S8Xr163HPPPVx22WXs2LGDoUOH0rBhQxo1asQcz6DjkiVLuOaaa+jQoQP16tWjf//+Ks5LSLl50mRBjDHEP/00W269jeTx46n2wgtORxIREZES8vzy59l8cLNX27w46mIeb/V4kY7Nycnhq6++om/fvidt7969O7NnzyYuLg6Xy0XVqlXZvXs3AIMGDeJvf/sbV1xxBW3btqV3795EREQAsHz5ctavX0/FihVp2bIlHTp0IDo6mt9++40ZM2bQoEEDPv/8c9asWcPPP//M/v37admyJddcc03e+Rs3bqRWrVq0a9eO+fPn06VLFy/2jpxOuR/hBghMSCCqX1+OfPQxR3/40ek4IiIiUs6lp6fTtGlT4uPjSU5Opk2bNiftb9euHV988QWzZ8+mW7duJ+3r3bs3mzZt4o477mDJkiW0bt2a48ePA9CmTRsqV65McHAwnTp14ttvvwWgVq1atG7dGoBvv/2WHj164HK5iIuL49prr2XFihUAtGrVigsuuACXy0WPHj3yzhffKvcj3CdEP/AARz5eSNKYMVyw4L+YChWcjiQiIiI+VtSRaG87MYf72LFj3Hjjjbz++usMHDgwb3+FChVo3rw5EyZMYOPGjXz00UcnnV+1alX69OlDnz59aNiwIevXu1dcM8acdNyJ9yEhIUXKVdD54ls+G+E2xrxjjNlrjClwTT5jTKIxZo0xZoMx5htfZQHwCwoi/qkRZP75Jwfene7LS4mIiIgA7vnar7zyChMmTCA7O/ukff/4xz94/vnniYqKOmn7p59+SlZWFgBJSUkcOHCAatWqAfDFF19w8OBB0tPTWbBgAVdeeeVfrnn11VczZ84ccnJy2LdvH0uXLqVVq1aAe0rJli1byM3NZc6cOVx11VW++NhyCl9OKZkOtCtopzEmAngDuNVa2wC4w4dZAKh07bWEtmnD/smTydy568wniIiIiBRTs2bNaNy4MbNmzTppe4MGDbj33nv/cvznn39Ow4YNadKkCTfeeCPjx48nPj4ecE8J6dy5M40bN6Zz5860aNHiL+d37NiRxo0b06RJE/72t7/xwgsv5J3fsmVLBgwYQP369UlISKBjx44++MRyKp9NKbHWLjXG1C7kkDuB+dba7Z7j9/oqS35xw4eR1mEZyePGUWPyGyVxSRERETnPpKWlnfT+448/zvv5xPSQ/Hr16kWvXr0AmDhxIhMnTjxtu9WrV2fBggUnbatdu/ZJbRpjGD9+POPHj//L+WFhYSxcuLDIn0O8w8k53HWBAGPMEiAUeNla+97pDjTG3A/cDxATE8OSJUuKdeGK7dph58/n+5df5niTJsVqq6xLS0srdn/K/6f+9B71pXepP71L/ek9vujL8PBwUlNTvdpmaZCRkUFmZmahny0nJ6fA/ceOHSM7O7tc9o03ZGRk+Oz/a2Ot9UnDAJ4R7oXW2oan2fca0AK4HggGvgc6WGt/LazNevXq2V9++aVYuWxWFls6dSbnaBp1Fi7Er2LFYrVXli1ZsoTExESnY5Qb6k/vUV96l/rTu9Sf3uOLvty0aRP169f3aptlRWpqKqGhoU7HKJNO93tjjFllrf3rvJ2z5OSygDuBz6y1R621+4GlQIkMN5uAAOJHjSR79x72T36zJC4pIiIiIucpJwvuD4GrjDH+xpiKwGXAppK6eMXmzQnv2JED777L8d9/L6nLioiIiMh5xpfLAs7CPU2knjFmpzGmrzGmvzGmP4C1dhPwKbAWWA5Ms9YWuISgL8QOHYJfSAhJo8fgy6k1IiIiInL+8uUqJT2KcMx44K+30JYQ/6goYgcPJmnkSI589BHht93mVBQRERERKafOi0e7Fybiji4ENWlM8gvjyTl82Ok4IiIiUk4sWLAAYwybN28+7f7ExERWrlxZwqnECed9wW38/KgyciQ5hw6xd9Ikp+OIiIhIOTFr1iyuuuqqvzzwRs4/533BDRB0ySVE3n0XKbPnkL5undNxREREpIxLS0vj22+/5e2332b27NkApKen0717d+rXr0/Hjh1JT0/PO/7BBx+kRYsWNGjQgJEjR+Ztr127NsOGDaNp06a0aNGC1atXc+ONN1KnTh3efFMrrZUVTj74plSJGTiQ1EWfkjRqNLXnzsG4XE5HEhERkWJKevZZjm86/ZSOcxVY/2Lihw8v9JgPP/yQdu3aUbduXSpXrsyqVav45ptvqFixIps2bWLt2rVceumlecePGzeOqKgocnJyuP7661m7di2NGzcGoGbNmqxZs4ZBgwbRq1cvli1bRkZGBg0bNqR///5e/WziGxrh9nBVqkTcsCfI2LCBQ55/iYqIiIici1mzZtG9e3cAunfvzqxZs1i6dCl33303AI0bN84rqAHmzp3LpZdeSrNmzdiwYQMbN27M23frrbcC0KhRIy677DJCQ0OJiYkhMDCQlJSUEvxUcq40wp1PaPv2hMybx76XJhHWti3+MTFORxIREZFiONNItC8cPHiQxYsXs27dOowx5OTkYIyhWbNmpz1+y5YtvPjii6xYsYLIyEh69epFRkZG3v7AwEAA/Pz88n4+8T47O9u3H0a8QiPc+RhjiHvqKezx4yQ//4LTcURERKQMmjdvHj179mTbtm1s3bqVHTt2kJCQQPPmzZk5cyYA69evZ+3atQAcOXKEkJAQwsPDSU5OZtGiRU7GFx/QCPcpAhMSqHxfP/a/MZmILp0Jad3a6UgiIiJShsyaNYvHH3/8pG2dO3fmp59+Ij09nfr161O/fn2aN28OQJMmTWjWrBkXX3wxNWrU4Morr3QitviQCu7TqHz//Rz+eCFJo8eQ8OEC/CpUcDqSiIiIlBFff/31X7YNHDiw0HOmT59+2u1bt27N+7lXr1706tXrtPukdNOUktPwCwoi/qkRZG7ZwsF33nU6joiIiIiUYSq4C1DpmmsIbduW/ZMnk7lzp9NxRERERKSMUsFdiLjhw8DlIvmZsVhrnY4jIiIiImWQCu5CBMTHEzNgAGnffEPaV185HUdERESKSANlcjZ8/fuigvsMonreTWDduiSNe5bcY8ecjiMiIiJnEBQUxIEDB1R0S5FYazlw4ABBQUE+u4ZWKTkDExBA/KiRbLvzLva/8QaxQ4Y4HUlEREQKUb16dXbu3Mm+ffucjlLiMjIyfFo4lldBQUFUr17dZ+2r4C6CipdeSnjnThyYPoPw224j8KKLnI4kIiIiBQgICCAhIcHpGI5YsmRJgU+0FOdoSkkRxQ4ZgiskhKTRY/QnKhEREREpMhXcReQfGUnMkH9wbOVKDn/4odNxRERERKSMUMF9FiI6dya4aVP2vjCenJQUp+OIiIiISBmggvssGD8/4keNJOfwYXb9Ywg2M9PpSCIiIiJSyqngPktBF19MlTFjOLpsGbufGIbNzXU6koiIiIiUYlql5BxEdO5EzqGD7H1xAq7ISOJGPIkxxulYIiIiIlIKqeA+R1F9+5J94CAH330X/+jKRD/4oNORRERERKQUUsF9jowxxA4dQs7Bg+x7+RVckVFEdu/mdCwRERERKWVUcBeD8fOjythnyElJIWn0aKe0YbkAACAASURBVFwREYS1u9HpWCIiIiJSiuimyWIyAQFUm/QSwU2bsnvoUI7+8IPTkURERESkFFHB7QV+wcHUeHMyFWrXZudDfyd9wwanI4mIiIhIKaGC20tc4eHUmDYVV0QEO+67n8ytW52OJCIiIiKlgApuLwqIi6PG29PAWrb37UdW8l6nI4mIiIiIw1Rwe1lgQgI1pkwh59Ahdtx3HzlHjjgdSUREREQcpILbB4IbNaT6a69yfMsWdjz4ELkZGU5HEhERERGHqOD2kZArrqDaC8+Tvno1uwYNxmZnOx1JRERERByggtuHwtq3J/7pp0j7+mv2PPU01lqnI4mIiIhICdODb3wsskcPsg8cZP9rr+FfOYrYIUOcjiQiIiIiJUgFdwmI/vtD5Bw8wIFpb+OKjKJy3z5ORxIRERGREqKCuwQYY4h78kmyDx1i7/jxuKKiiOh4u9OxRERERKQEqOAuIcblourzz7Pz8GH2jBiBKyKc0OuuczqWiIiIiPiYbposQX4VKlDtlVcJql+fXYMGc2z1aqcjiYiIiIiPqeAuYa5KIdSY8hYB8fHs6P8gGb/+6nQkEREREfEhFdwO8I+Kosa0afgFBbGj331k7tzldCQRERER8REV3A6pUL0aNaZNJTcjgx19+5J94IDTkURERETEB1RwOyiobl1qvDmZrORkdtz/ADlpR52OJCIiIiJepoLbYRUvvZRqk14iY/Nmdj48gNzMTKcjiYiIiIgXqeAuBUITE6kybizHvv+B3UMfw+bkOB1JRERERLxEBXcpEXH77cQ+/jipn31G0jPPYK11OpKIiIiIeIEefFOKVO7dy/0I+KnT8K8cTczDA5yOJCIiIiLFpIK7lIkZPJjsgwfZ//rruCpHEXXnnU5HEhEREZFiUMFdyhhjqDJ6NDmHUkh+Ziz+kZGEtW/vdCwREREROUeaw10KGX9/qk2cQPCll7Lrscc5+t13TkcSERERkXOkgruU8gsKosbkNwhMSGDHgIdJX7fO6UgiIiIicg5UcJdirrAwakybin9kJDvuf4Djf25xOpKIiIiInCUV3KVcQGwsNd+eBn5+bO/Xl6zkZKcjiYiIiMhZUMFdBlSoXZsaU94i9/ARdvTrR05KitORRERERKSIVHCXEcENGlD99dfJ3LqNHf0fJDc93elIIiIiIlIEKrjLkJDWl1F1woukr13LzkcfxWZlOR1JRERERM5ABXcZE9a2LfEjR3L0m6XsfORR0tdv0GPgRUREREoxPfimDIrs1pXco0fZ9/LLpC1eTGC9ekR07kzYLTfjHxnpdDwRERERyUcj3GVU5T69uej/lhI/aiQmIIDkZ5/l92uuZeegQaR9uwybk+N0RBERERFBI9xlmissjMju3Yns3p2MX34h5YMPOPLhR6Qu+hT/KlWI6NiR8E4dqVC9utNRRURERM5bGuEuJ4Lq1SN++HAu/L+lVJv0EoF16rB/8mT+uKEN23r35vDHC8nNyHA6poiIiMh5x2cFtzHmHWPMXmPM+jMc19IYk22M6eKrLOcTvwoVCGvXjprTpnLh4q+IHvgwWdt3sHvoUH675lqSxowhfYNutBQREREpKb4c4Z4OtCvsAGOMC3ge+NyHOc5bAVWqEPPQQ9T54nNqTn+XStdcQ8q8D9jauQtbOnbi4L/+jUlLczqmiIiISLnmsznc1tqlxpjaZzjsYeADoKWvcggYPz9CWrcmpHVrcg6P4PAnn3D4g/kkjxtHjL8/uxYvJrxTZ0KuuBzjp1lGIiIiIt5kfDm1wFNwL7TWNjzNvmrATOA64B3PcfMKaOd+4H6AmJiY5nPnzvVV5POK/86duJZ8Q9hPP+F39Cg5UVGkX96a9MuvIDe6stPxyqS0tDQqVarkdIxyQX3pXepP71J/eo/60rvUn9513XXXrbLWtihuO06uUjIJeNxam2uMKfRAa+0UYApAvXr1bGJiou/TnSeWVK9O/ddfI23xYlLmfYDrf4uo9L9FhFzemvBOnQltcwN+gYFOxywzlixZgn4/vUN96V3qT+9Sf3qP+tK71J+lk5MFdwtgtqfYjgZuMsZkW2sXOJjpvOQXGEhY+/aEtW9P1u7dpPz3vxye/192DxmCX1gY4TffTESXzgRdconTUUVERETKHMcKbmttwomfjTHTcU8pUbHtsICqVYn5+9+JfvBBjv34IynzPiBl3jwOzZxJYP36RHTuTPjNHXBFRDgdVURERKRM8FnBbYyZBSQC0caYncBIIADAWvumr64r3mH8/Ai5/HJCLr+cnMOHObxwoftGy7Fj2fvCC4TecAPhnTsRcrlutBQREREpjC9XKelxFsf28lUOKT5XeDhRd91F1F13kbFpEykfzOfwxx9z5H//I6BqVaq9PIngRo2cjikiIiJSKmloUs5KUP36xI94kouWfkO1iROw1rJ76GN6iqWIiIhIAVRwyznxCwwk7KabqDpuLJlbt7L/9dedjiQiIiJSKqnglmIJueIKIu7owoG33yF93Tqn44iIiIiUOiq4pdhiH3sM/5gY9gx/EpuZ6XQcERERkVJFBbcUmys0lPhRIzn+22/sf2uK03FEREREShUV3OIVodddR9itt7D/rbfI+OUXp+OIiIiIlBplruCukHnI6QhSgLhhw3CFh7Nn2HBsdrbTcURERERKhTJXcAcePwgHtzgdQ07DPzKS+KefJmPjRg68867TcURERERKhTJXcAOwYprTCaQAYTe2JfTGG9n/2msc//NPp+OIiIiIOK7MFdzZ/iHw078g86jTUaQA8U+NwC842L1qSU6O03FEREREHFXmCu7MChGQcRjWznU6ihTAPzqauCeHk75mDYf+8x+n44iIiIg4qswV3DmuIIhvBMungrVOx5EChN1yC5WuvZa9E18ic/t2p+OIiIiIOKbMFdwAtLof9m6AbcucTiIFMMYQP3oUxt+fPU89jc3NdTqSiIiIiCPKZsHd6A4IjoTleshKaRYQH0/s449x7McfSZn7vtNxRERERBxRNgvugGBo1hM2LYTDO51OI4WI6NKFipe3Zu/48WTt3u10HBEREZESVzYLboCW/cDmwsp3nE4ihTDGUOWZZ7DWsmfkKKzm3YuIiMh5puwW3JG1oF57WDUdsjKcTiOFqFC9OrGDB3P0//6Pwws+dDqOiIiISIkquwU3uG+ePHYANvzX6SRyBpF39iC4eXOS//lPsvbudTqOiIiISIkp2wX3BYkQXVc3T5YBxs+PKmOfwR4/TtKYMZpaIiIiIueNsl1wG+Me5d69GnaudDqNnEFgQgIxAx8m7cuvSP30U6fjiIiIiJSIsl1wAzTpDhVCNcpdRkTdey9BjRqR9MxYsg8edDqOiIiIiM+V/YI7MBSa3gnr50Oa5gaXdsbfnyrjxpKTmkryuGedjiMiIiLic2W/4AZodR/kZsGqGU4nkSIIqluX6Af7c+STT0j96iun44iIiIj4VPkouKMvgjp/g5VvQ06W02mkCKLvu4/Aiy8madRocg4fdjqOiIiIiM+Uj4IboNUDkLoHNn3sdBIpAhMQQJVx7nncyc+/4HScYsvYvJkdD/0d/5168qmIiIicrPwU3Be1gYhasHyq00mkiIIbNKByv34cnj+ftP/71uk45yxlwQK2dutO2uLFhP3r39jcXKcjiYiISClSfgpuP5d7Lvf27yBpndNppIiiH3qQCnXqsGfk0+SkHXU6zlnJzcxkz+jR7HliGMFNmhD7xOMEbNvG4fnznY4mIiIipUj5KbgBmt4F/sFaIrAM8QsMpMrYZ8jek8S+iROcjlNkWXv2sO3unqTMmk3lfn2p+c7bRN17L5l16rB34kvkHDnidEQREREpJcpXwV0xChp3hbXvwzGt8VxWVGzWjKh77uHQzFkcXb7c6ThndPS779jSqTOZf/xBtVdfIXbIEIy/P8YYUrt1I+fQIfa9+prTMUVERKSUKF8FN7ifPJmdDj/92+kkchZiHn2EgJo12TPiKXLT052Oc1o2N5f9b01he7/78I+uTO157xPWps1Jx2TXrEFEt64cmjmTjF9/dSipiIiIlCblr+CObwi1roQVUyE3x+k0UkR+wcFUeeYZsrZvZ9/Lrzgd5y9yjhxh54CH2ffSS4S1b0/t2bMJTEg47bExjzyCq1IlkseOw1pbwklFRESktCl/BTe4R7lTtsOvnzmdRM5CyGWtiOjRnYMzZpC+Zo3TcfJk/PILW+64g7SlS4kbPpyqL47HLySkwOP9IyOJefQRji1fTuqnn5ZgUhERESmNymfBfXEHCK2qmyfLoNh/DMG/Sjy7nxxB7vHjTsfh8EcfsbVbd+yxdGq9N4Ooe3pijDnjeRFduxJYvz7Jz79A7rFjJZBURERESqvyWXC7AqBlH/jza9inebRliatSCFVGjyHzjz/Y/8Zkx3LYzEySxjzD7sceJ7hRIxLmf0DFSy8t8vnG5SJ+xJNkJyWx/y39w09EROR8Vj4LboBLe4Grgnsut5Qpla6+ivBOnTgwbRrpGzaU+PWzkpLY1vMeDs2cSVSfPtR89x38Y2LOup2KzZsTdustHHznHTK3bfNBUhERESkLym/BXSkGGnSCNTMhQ2silzVxjz+Gf1QUe54cgc3KKrHrHv3hB7Z06szx336j2qRJxD02FOPvf87txQ4ZggkIIPmfz3kxpYiIiJQl5bfgBvfNk5lp8PNsp5PIWXKFhxM/ehTHN2/mwLRpPr+etZYD06axvU9fXJGR7iX/2t1Y7HYDYmOJ/vtDpC1ZQto333ghqYiIiJQ15bvgrt4cqjV33zyZm+t0GjlLoX/7G2E33cS+NyZz/LfffHadnNRUdg0cyN4XJxB6Y1sS5s4h8IILvNZ+VM+eVEhIIOnZZ8nNzPRauyIiIlI2lO+CG6DVA3DgN9iyxOkkcg7iRjyJq1Ildg9/Epud7fX2M379la1d7iB18dfEPvE41SZOLHTJv3NhKlQgbvhwsrZt5+D0GV5tW0REREq/8l9wN7gdQmLgR60UURb5R0UR/9QIMtat4+CM97za9uGFn7C1W3dyjh2l1ozpVO7Vq0hL/p2LSldfRaXrr2f/m2+SlZzsk2uIiIhI6VT+C27/QGjeC379FA5ucTqNnIPQ9u2pdMP17HvlFY5vKf53aDMzSRo7jt1DhhB0ySUkfPABFVu08ELSwsU98ThkZ7P3hfE+v5aIiIiUHuW/4AZo3huMH6x82+kkcg6MMcQ//TQmMJA9I57CFmM+flZyMtvu7cWhf/+bqHvvpdb0dwmIjfVi2oJVqFGDyv36cuSTTzi6fHmJXFNEREScd34U3OHVoP4tsPpfkKmn/pVFAbGxxA0bRvqqVRyaOeuc2jj643K2dOpMxi+/UO2licQNewITEODlpIWrfN99+FetQvLYcT6Zky4iIiKlzxkLbmNMYFG2lXqt7oeMFFj3vtNJ5ByF334bIVdfzd6JE8ncubPI51lrOfD2O2zv0wdXWBgJc+cQ1r69D5MWzC84mLjHn+D4r79yaPYcRzKIiIhIySrKCPf3RdxWutW6AuIaupcItNbpNHIOjDFUGT0KYwx7nnoKW4TvMSctjV2PPMre8eMJvf56ar8/l8ALLyyBtAULbduGipe3Zt8rr5B98KCjWURERMT3Ciy4jTHxxpjmQLAxppkx5lLPKxGoWGIJvcUYaHUfJK+H7WXv3wviFlC1KrFDh3Ls+x9ImTev0GOP//47W+/oSupXXxH72GNUe3kSrkqVSihpwYwxxD/5JLnHjrHvpUlOxxEREREfK2yE+0bgRaA6MBGY4HkNAob7PpoPNOoKQRHw41tOJ5FiiOh6BxVbtWLv8y8UuMTekf/9jy1du5Fz5Ag1332Hyn16+2zJv3MReOGFRN11Fynz5pG+br3TcURERMSHCiy4rbUzrLXXAb2stdfle91mrZ1fghm9p0JFuLQnbPoYDu9yOo2cI+PnR5Wxz2Czs0kaOeqkqSU2K4vkf/6TXYP/QVC9eiTMn09Iq1YOpi1Y9IC/46pcmeSxY4u18oqIiIiUbkWZw73MGPO2MWYRgDHmEmNMXx/n8p0WfcHmwqp3nU4ixVChZk1iBz1K2pIlHFm4EICsvXvZ1qs3B2e8R2TPntSaMZ2AuJJZ8u9cuEJDiR08mPSff+bwgg+djiMiIiI+UpSC+13gM6Cq5/2vwKM+S+RrUQlQtx2smg7Zx51OI8UQeffdBDdtSvLYcRz57HO2dO5MxsaNVH3xReKfHI6pUMHpiGcUfvttBDdpwt4JE8hJTXU6joiIiPhAUQruaGvtXCAXwFqbDeT4NJWvtboPju6DDQucTiLFYFwuqjw7jtz0dHY98giuiiHUnjOb8Js7OB2tyIyfH3EjRpBz8CD7X3vd6TgiIiLiA0UpuI8aYyoDFsAY0xo47NNUvnbBdVD5IliumyfLusALLiB+9Ggiunal9rz3Capb1+lIZy24UUMiunTh4H/+w/Hff3c6joiIiHhZUQruwcBHQB1jzDLgPeBhn6YqRLY37i3z83OPcu9aBTtXeaFBcVJEx9upMmY0rtBQp6Ocs5hBj+JXsSJJ48YVaX1xERERKTvOWHBba1cD1wJXAA8ADay1a30drCC7j+aS6Y2qu0kPqFDJ/SAcEYf5R0URM3Agx77/gdTPv3A6joiIiHhRUR7tfgcQbK3dANwOzDHGXOrzZAXItfD1L3uL31BQGDS9EzbMh7R9xW9PpJgiu3cjsG5dkp9/jtz0dKfjiIiIiJcUZUrJU9baVGPMVcD1wNvAZN/GKpjLwLxVO73TWMv7ICcTVk/3TnsixWD8/Ykb8STZu/dwYOo0p+OIiIiIlxSl4D6xIkkHYKq19hPAsfXWQgIMX2/ey4E0LyzpF1PXfQPlincgJ6v47YkUU0irVoTddBMHpk0jc8cOp+OIiIiIFxSl4N5ljHkL6Ab8zxgTWMTzfCK0giE71/Lhmt3eabDV/ZC6GzZ/4p32RIop9rGh4HKR/NzzTkcRERERLyhK4dwV94NvbrTWpgBRwFCfpipEgB80rh7uvWkldW+EiJq6eVJKjYD4eKIffJC0r74i7f++dTqOiIiIFFNRVik5Zq2db639zfN+j7X2c99HK1iX5tXZuOcIG3Z7YTlwPxe07AfblkHS+uK3J+IFUb3upUKtWiSPG4fNzHQ6joiIiBSDz6aGGGPeMcbsNcactoo1xtxljFlrjFlnjPnOGNOkqG3f0rgqFVx+fLBql3fCNusJ/kEa5ZZSw69CBeKGDyNz61YO/utfTscRERGRYvDlXOzpQLtC9m8BrrXWNgKeAYpc7UaGVOCGS2JZsGaXd9bkrhgFje6AtXMh/VDx2xPxgkrXXkulxET2v/4GWcleWApTREREHFGUdbj/cufW6badylq7FDhYyP7vrLUnqtsfgOpnajO/Ls2rc/BoJku8sSY3wGUPQHY6/PRv77Qn4gVxw57AZmWxd8KLTkcRERGRc1SUEe42p9nW3ss5+gKLzuaEay6KIbpSoPdunoxvBDWvgBXTIDfnzMeLlIAKtWoR1acPRz76mGOrVjkdR0RERM6BsdaefocxDwIPARcAf+TbFQoss9befcbGjakNLLTWNizkmOuAN4CrrLUHCjjmfuB+gJiYmOZz584FYPbmTL7YlsVL11UkrII5U5wzitn7LQ02jmddwxEciG5Z7PbKgrS0NCpVquR0jHLDJ/15/DjRo0aTWymEg8OGgZ9jq3KWKP1uepf607vUn96jvvQu9ad3XXfddaustS2K245/Iftm4h51/ifwRL7tqdbaAqeKnA1jTGNgGtC+oGIbwFo7Bc8c73r16tnExEQAqlycyqeTlrK/Ym1uvSqh+IFyroQd/6ZR+neQ6NjKhyVqyZIlnOhPKT5f9eeRnBx2Df4HTZKTiezRw+vtl0b63fQu9ad3qT+9R33pXerP0qnAoTJr7WFr7VZgBJBkrd0GJAB3G2MiinthY0xNYD7Q01r767m0US8+lEbVvLgmtysAWvSBPxbD/t+806aIF4S2b0/FVq3YN+llsg/pxl4REZGypCh/m/4AyDHGXIh7lLkG7tHvQhljZgHfA/WMMTuNMX2NMf2NMf09hzwNVAbeMMasMcasPJcP4NU1uQGa9wK/AFg+1TvtiXiBMYa4EU+Sk5bGvpdfdjqOiIiInIWiFNy51tpsoBPwqrV2KFDlTCdZa3tYa6tYawOstdWttW9ba9+01r7p2d/PWhtprW3qeZ3T/Jhbm1QlwGW8tyZ3pVho0BHWzITjqd5pU8QLgurWJfLOO0mZM5eMjRudjiMiIiJFVJSCO8sY0wO4B1jo2Rbgu0hnJzKkAjfUj/PemtzgXiIwMxV+nu2d9kS8JObhAbgiI0kaO46CbngWERGR0qUoBXdv4HJgnLV2izEmAShVj77z+prc1ZpD1WbuJ0+qqJFSxBUWRuzgQaSvXs2Rjz92Oo6IiIgUwRkLbmvtRuBxYLXn/RZr7RkffFOSrqnr5TW5jYFWD8D+X+HPJd5pU8RLwjt1IqhRI5LHjycnLc3pOCIiInIGRXnS5C3AGuBTz/umxpiPfB3sbAS4/OjYrCqLN+/lQNpx7zTaoCNUjHaPcouUIsbPj/gRT5Kzbz/735jsdBwRERE5g6JMKRkFtAJSAKy1a3A/DKdU6dy8Otm5lg/X7PZOgwFB0Pxe+GURHNrmnTZFvCS4SRPCO3Xi4HvvcfzPP52OIyIiIoUo0k2T1tpT19zz0t2J3nNxfJh31+QG95rcxs/9uHeRUiZ28CD8goNJHvesbqAUEREpxYpScG8wxtwJuIwxFxljXgW+83Guc+L1NbnDq8PFHWD1e5B5zDttiniJf3Q0MQ8P4OiyZaR99ZXTcURERKQARSm4HwYaAMdxP/DmMPCIL0OdK6+vyQ3uJQIzUmD9PO+1KeIlkT16EHjRhST/8zlyMzKcjiMiIiKnUZSCu4O19klrbUvPawRwq6+DnQufrMld60qIvURLBEqpZAICiHtyBFm7dnHg7bedjiMiIiKnUZSCe1gRt5UKXl+T2xhodT8krYPtP3inTREvCml9GaHt2nFgylQyd3rxrzsiIiLiFQUW3MaY9p752tWMMa/ke00Hskss4Vk6sSb3B6u9ePNk464QFK4lAqXUintsKBjD3udL1RL5IiIiQuEj3LuBlUAGsCrf6yPgRt9HOzcn1uT+apMX1+SuEALNesKmj+DIHu+0KeJFAVWrEv3A/aR+8QVHvyuV9zSLiIictwosuK21P1trZwAXWmtn5HvNt9YeKsGMZ+3Emtwf/eylNbkBWvaF3BxY+Y732hTxoqg+fQioUYPdw5/k+G+/OR1HREREPIryaPeskgjiTT5ZkzvqArioLax6F7K9NHIu4kV+gYFUf3kSNiebrT3u1Ei3iIhIKVGUmybLpC7Nq7Nh9xE27j7ivUYvux+O7oONH3qvTREvCrrkEhLmzCGgShW23/8Ah95/3+lIIiIi571yW3DnrcntzZsnL/gbRNXRzZNSqgVUrUqtmf8h5LLLSHrqafZOmIjNLXUPhxURETlvnLHgNsbUNcZMNcZ8boxZfOJVEuGKI29N7p92kZXjpWLDz8+9RODOFbBzpXfaFPEBV2goNd6cTETXrhyYOpVdg/+hB+OIiIg4pCgj3O8Dq4ERwNB8r1KvS/PqHDiayZJf9nmv0aY9ICQGPugLR/d7r10RLzMBAcSPHkXs0KGkfvop2+/tRfaBA07HEhEROe8UpeDOttZOttYut9auOvHyeTIvOLEm97xVO7zXaFA4dJ8FqUkwqwdkpXuvbREvM8ZQuW8fqr38MhmbN7O1W3eO//GH07FERETOK0UpuD82xjxkjKlijIk68fJ5Mi/wyZrcADVaQse3YOdy+G9/0PxYKeXCbmxLrfdmkJue7l7B5IcfnY4kIiJy3ihKwX0v7ikk3/H/H35TZiYw+2RNboAGt0ObMbBxAXw12rtti/hAcJMm1J4zB//YGLb360fK/P86HUlEROS8UJR1uBNO87qgJMJ5g0/W5D7hioHQvDcsmwSrpnu/fREvq1C9GrVnzqRiyxbsGT6cvS+/jLXW6VgiIiLlWlFWKQkwxgw0xszzvAYYYwJKIpy3+GRNbgBj4KYX4cIbYOFg+P1L77Yv4gOusDBqTplCeOdOHJj8JruHDCX3uB7mJCIi4itFmVIyGWgOvOF5Nfdsc8TR3KNnfY5P1uQ+weUPd0yH2Etgbi9IWu/9a4h4mQkIoMrYscQMGsSRTz5he5++ZB865HQsERGRcqkoBXdLa+291trFnldvoKWvgxXkYPZBth3Zdlbn+GRN7vwCQ+HOORBYCWZ2hSN7vH8NES8zxhD9wP1UmziBjHXr2Nq9O8e3bHE6loiISLlTlII7xxhT58QbY8wFQI7vIhXOYHhjzRv/j737Dm+yauM4/j1Jk3Skky5oC6V00TLLUPYogqggAjJUhoAoS3zdWxS3iIOhIiiKskGWiAKCDEWUKVAoo+wNpQs687x/pBTKLJA0bbk/15XrbZLzJHdC3/rLyXnuc8PH2aUn98U8g+Ch6ZCZYg3dWen2eR4hbMzjnnuoOHEiltQ09nXrztl//nF0SUIIIUSZUpTA/RywTCm1XCn1B/A78Ix9y7o6s97ML0m/kJiceEPHWXtyG23bk/tS5WtA52/h2BbrxjgWh30uEeKGuMbVJnTaVPQ+Puzr05eUefMcXZIQQghRZhSlS8lSIAJ4EhgCRGmatszehV2Nh94DN4MbozeMvqHjDHodHWoF2b4n96UiW8M9H0HiIlj0IkgHCFFKGCtWJHTqFFxr1+bw8y9wYvQY6WAihBBC2EBRZrjRNC1L07TN+ReHtjPQoaN3bG+WHVjG5hObb+hYu/XkvlS9ftBgMKwdB2scdn6pEDdM7+lJxfFf49mhAydHj+bwCy9gyc52dFlCCCFEqVakwF3S0FU6ygAAIABJREFUPBLzCN4mb0ZtGHVDx1Ut70G1IA/79OS+1F3DoWo7+PVlSJhv/+cTwkaU0Uj5997Fb+iTpM6bz4E+fck7c8bRZQkhhBClVqkM3G4GN/pV78eaI2tYe2TtDR3bOc5OPbkvpdPBA+MgKA5mPQYH19n3+YSwIaUUvgMGUOGjjzi3aRN7u3Une/9+R5clhBBClEpF2fhmtlLqXqVUiQrnXaO74u/qz+cbPr+hdabtawXZryf3pYyu0H0qmP1gSldIvrF2hkI4mme7+6g48VvykpPZ27UbZ9evd3RJQgghRKlTlBA9FngI2KmUel8pFWXnmorEpDfxRM0n2HRiEysOrijycT5uRuKj7diT+1Jmf3h4JuRlw48Pwjn5al6ULq516hA6bSo6D3f2936UlJ9/dnRJQgghRKlSlC4lSzRNexiIA/YCS5RSfyqlHnX0Fu8dwjsQ4h7CqA2jsGhFD89278l9Kb8o6PoDnN4D03tArpyEJkoXY2gooVOn4ly9OoefeZaTX34pHUyEEEKIIirSMhGlVDmgN9AP2AB8hjWAL7ZbZUVg0BkYWGsgO5J38Nve34p8XLMoa0/uWcVx8uR5lZtC+88haQUseEraBYpSx8nbm4rffoNHu3ac+PQzjrz8Cpp0MBFCCCGuqyhruH8CVgKuQDtN09prmjZN07QhgNneBV5P29C2hHuFM2bjGHItuUU6pqAn9/ZjnM4oxsBQ6yFo9gJs/BFWjCi+5xXCRnRGIxU+/ADfQYNI+ekn9j/Wn7yUFEeXVSJpmkbmtm0kT52G5exZR5cjhBDCgYoyw/25pmkxmqa9p2nakYvv0DStrp3qKjK9Ts/g2oPZm7qX+buL3n6vU51gcvI05m08ZMfqrqD5S1CjKyx7GzbPKN7nFsIGlFL4DRlM+fff4+z69ezt/hDZB4vx26ISTMvOJuPPPzk6/G12xceT1LETR4cN48ATA7CcO+fo8oQQQjhIUQJ3jFLK6/wVpZS3UmqgHWu6YS1DWlKtXDW+2PQF2XlFm7Eu6MldHN1KLqYUtB8FlRrB3IGw78/ifX4hbMSrQwcqThhP7qlT7O3SlXMbNzq6JIfIS0sj5eefOfT0MyQ2asz+Pn05M2sWzlVjKP/OOwS+9SZn//2XAwMHYsnMdHS5QgghHKAogfsxTdMKWmtompYMPGa/km6cUoohcUM4knGEGYlFnzXuHBfMlkOpJByxc0/uSzmZrCdRelWCqQ/ByV3F+/xC2Ihb/fqETpmCzs2Nfb16k7roV0eXVCxyjhzh9I8/sr9PXxIbNuLwM8+SsWYN7q3vInjsGCL/+pOQMaPx6tQR7y5dqPDeu5xd8zcHBw3GkuXQzXqFEEI4QFECt14ppc5fUUrpAaP9Sro5Dco3oG5AXb7e/DVnc4q2XrKgJ3dxnjx5nqsPPDwdlA5+7AwZJ4u/BiFswBRWmdBpU3GOieHQU09xavz4MtfBRNM0MhMSODF6DEkdO7GrRUuODX+bnCNHKNerJ5UmTyZi5QoqvPMO7i1bonNxKXS85/33U/7tt8lYvZqDQ4ZgkZNNhRDitlKUwL0ImKaUildKxQNT8m8rUZRSPBn3JKcyTzF5++QiHVPQk3tjMfXkvqyAMOvGOKmHrTPdOfJ1syidnHx8qDjxWzzuacvxER9z9PXXydyxo1QvodBycsj46y+Ovv0Ou+NbkfRAR06OGYMymfB/9hnCFi6kyi8L8X/2WVzjaqP0+ms+nlenjgS+9SYZK1Zy6Mmh0uFFCCFuI05FGPMC8DgwIP/6YmC83Sq6BbX9a9MkqAnfbvmWLlFd8DB6XPeYznWCWbT1KH/sOEGrmIBiqPISIfWh41cwozfMeQI6fWPdFl6IUkZnMlFhxAgMIRU59dVXnJkxE5TCUKECxrAwjJVDMYWFYQytjLFyZZz8/bjoy7MSIS89nYwVK0hb+jvpK1ZgSUtDmUy4NWqE78ABmJs3x8nX96Yf37tLF7BYODrsTQ4+/TTBn3yCMjh0OwMhhBDF4LqBW9M0C/BF/qXEG1J7CF0WdOG7rd8xpPaQ644/35N75rqDjgncALEPWLd9X/IGeIdCq2GOqUOIW6R0Ovz/9xSe7e4jc8cOsvckkZ2URFZSEmf//Rftok4dOjc3jJUrYwyrjKlyZYyVw6zXQyuhM5mKreaco0dJ+/130pf+TsbatZCTg97bG/e77sI9viVuDRtetkTkVnh364aWm8ext9/m0DPPEvTxCAndQghRxl03cCulIoD3gBjA+fztmqaF2bGum1a1XFVaV2rNpG2TeCj6Icq5lLvm+PM9ub/7ay+nM7LxcXPQ8vRGQyE5CVZ9Yg3ddXo7pg4hbMAUHo4pPLzQbZrFQu6xY2Tt2UN20l6y9+whe28SZ//5l9R5F7X0VApDUBBenh4c++svawivbJ0hd/K79VlxTdPI2rGDtKVLSV/6O5nbtgFgrFQJnx49cI9viUutWtddInIrfB55GPJyOfbe+xx+4QUqfPghyqkoXzgKIYQojYryF/5b4A3gE6AF8ChF3KHSUQbVHsSS/UuYsGUCz9d7/rrjO9UJZvyqJOZtPETvRpWLocIrUAru+RjOHIAFT4NnCITHO6YWIexA6XQYypfHUL48NGpU6D7L2bNk791LVv6MeHZSEmf/20zy9BmFZ8XN5vwAftHylLDKGCtde1Zcy8nh7Lp11qUiv/9OzqFDoBQuNWvi98zTuMfHYwor3jkEn1690HLzOP7RR6DTU+GD9+0a8oUQQjhOUQK3i6ZpS5VSStO0fcAwpdQ64HU713bTwjzDaBfWjmnbp9EzpieBboHXHH9xT26HBW4AvRM8OBG+uRum94K+v0JArOPqEaKY6FxdcY6JwTkmpuC2ncuX06xpU3KPHiUrKanwrPjaf644K37p8pS806dI+30Z6X/8gSU11boeu0EDyj3xOO4tWtzSemxbKNe3D1peHidGjkTpdZR/910J3UIIUQYVJXBnKaV0wE6l1GDgECVgS/frGVBrAD8n/cxXm7/ijQZvXHd857hghs3fRsKRVKqWv/7Jlnbj7GFtFzi+FfzYBfotAY/yjqtHCAdSOh2GChUwVKhQhFnxPWQl7eXsP4XXiuu9vXGPj7+wHtvVtbhfxjX59n8M8nI58dnnoHei/NvDUXLitBBClClFCdxDAVfgSWA41mUlvexZlC0EmYPoHNGZmYkzeTT2USp6VLzm+Pa1gnhnYQKz1h3k1ftirjnW7jyD4aFp8E1bmNIVei8EU4n/jCNEsbrSrDjkrxXPnxXXubjgUrNmiZ819h0wAC03z9p2UK8n8M1hErqFEKIMueZf9PxNbrpqmpauadpBTdMe1TStk6Zpa4qpvlvSv0Z/nHROjN009rpjHd6T+1Lla8KD38LR/2BWP7DkOboiIUqF87Pi5kaNcI2LK/Fh+zzfwYMo98TjnJkxg6PDh5e5zYOEEOJ2ds3ArWlaHtC4mGqxOT9XP7pX7c7CPQvZmbzzuuM71wnmZHo2f+w4UQzVFUFkG2j7IST+AoteBPkPsBBlllIKv6FDKdevL2emTOXYO+9K6BZCiDKiKN9ZblBKzVNK9VBKdTx/sXtlNtIntg9uBjdGbxh93bEX9+QuMeo/BncOgrXjYE2paIUuhLhJSin8nnkGn969Sf7hB46//76EbiGEKAOKsobbGTgFtLzoNg2YbZeKbMzL2YuesT0Zu3EsW05uoZpvtauOLTE9uS/Vejic2Qe/vgxeFaHqfY6uSAhhJ0op/F94Hi0vj9PffQ9OTvg/+2yJ25VTCCFE0V13hjt/3fallz7FUZyt9IzpibfJm8/Xf37dsZ3qBJOTpzFv46FiqKyIdHro+DVUqG1dz31onaMrEkLYkVKKgJdfwvuh7pye8A0nPvlUZrqFEKIUu27gVkp9q5T65tJLcRRnK24GN/pW78tfR/7in6P/XHPsxT25SxSjq7VzidkPJnezbgUvhCizlFIEvPoqXl26cGrcOE6Ouv6yOCGEECVTUdZwLwB+zr8sBTyAdHsWZQ9do7ri7+LP5+s/v+5MUee4YLYcSiXhSGoxVVdEZn94aAbkZsHkLnDujKMrEkLYkdLpCBz2Bp6dOnJy7FhOjBnj6JKEEELchKIsKZl10eVHoAtQ1/6l2ZazkzOP13ycjSc2svLQymuObV8rCINeMasknTx5nn80dJ0Ep3bB9B7W8C2EKLOUTkf54cPx7NCBk6NGc/LLrxxdkhBCiBt0MzsrRAD+ti6kODwQ/gDB5mBGbRiFRbt6r20fNyMto/1LTk/uS4U1g/ajIWkF/PQEWEpgjUIIm1E6HeXfeRuPdu048emnnJowwdElCSGEuAFFWcOdppRKPX8B5gMv2L802zPoDQysNZDtp7ezeN/ia47tXCeEk+nZrEgsIT25L1WrO7R6E7bOlh7dQtwGlF5PhffexeOeezj+0QhOTZzo6JKEEEIUUVGWlLhrmuZx0SVS07RZxVGcPdxT+R6qeFZh9IbR5FpyrzqueZQf5dxKWE/uSzUamt+j+ytY+bGjqxFC2JlycqLChx/g3qYNx9//gNOTfnB0SUIIIYqgKDPcDyilPC+67qWU6mDfsuxHr9MzuPZg9qbuZcGeBVcdZ9Dr6FA7iCUJx0jOyC7GCm+AUtD6bajeBX4fDuu+c3RFQgg7U05OBI34CPe7WnHsnXc4PXmyo0sSQghxHUVZw/2Gpmkp569omnYGeON6B+W3DzyulNpylfuVUupzpdQupdRmpVRc0cu+NfEV44kpF8MXG78gO+/qYbrz+Z7cmw4XV2k3TqeD+8dAlXhY8BRs/9nRFQkh7EwZDAR9/DHmFi049tZwkqdNd3RJQgghrqEogftKY4qyQ+VE4O5r3N8W6wmYEUB/oNj2LVdK8WTtJzmccZhZO6++OqZqeQ9iK3iU7GUlAE5G6PK9dWOcmX1g35+OrkgIYWfKaCTos09xa9aUo2+8wZlZpXalnxBClHlFCdz/KqVGKqWq5F9GAtfd6lDTtBXA6WsMuR/4XrNaA3gppcoXrexb17BCQ+oE1GHc5nGcyz131XGd6wTz36EUth8tYT25L2UyW3t0e4ZYN8Y5ttXRFQkh7ExnNBL8+ee4NW7MkVdf48ycOY4uSQghxBUUJXAPAbKBacBUIBMYZIPnDgIOXHT9YP5txeL8LPfJcyeZsn3KVcfdX5J7cl/KrRz0mG3dlXJSR9mNUojbgM5kInj0KNwa3MmRl14mZf58R5ckhBDiEup6uy7e0oMrFQos0DSt2hXuWwC8r2naqvzrS4EXNE379wpj+2NddoKfn1+d6dNtt17xi2NfsDd7L8OChuGic7nimFEbMtmZbGFkcxecdMpmz20vbun7qLXxJXIMnmyo/T45Rs+rjk1PT8dsNhdjdWWbvJ+2I+/lDcrOxnv0GAw7d5LStw9ZdQvvTybvp23J+2k78l7alryfttWiRYt1mqbd8oaP112LrZRaDDyYf7IkSilvYKqmaW1u8bkPASEXXQ/Ov+0ymqaNA8YBREVFac2bN7/Fp77A/5Q/XRd0ZY/PHgbVuvLEfY7/MR77/l9U+RiaVw2w2XPbVfVIDJM60Gjvp9BrvnXJyRUsX74cW76ftzt5P21H3ssbZ2ncmP39+6O+nUhQtep43H3hz7S8n1aWc+dAp0NnMt3S48j7aTvyXtqWvJ8lU1FOfvQ9H7YBNE1LVkrZYqfJecBgpdRU4A4gRdO0IzZ43BsSUy6Guyrdxfdbv6d7dHd8nH0uG3NxT+740hK4KzWAzt/CtIetW8B3n2Y9uVIIUWbpXF0J+fIrDvTvz6Fnn0U56XFv1crRZTlMbnIymdu2kZWQQOa2BDK3byc7KQn0ekwRETjHxuASG4tztWqYIiNvOYQLIcTVFCVwW5RSFTVN2w+glKoEXHcdilJqCtAc8FVKHcTaStAAoGnal8BC4B5gF3AWePRmXoAtDK41mKX7l/LNf9/wbL1nL7v/fE/u7//aS3JGNt5upSS4Rt8D7T6DeUNg7kB4YJy1jaAQoszSm90IGfcVB/r24+D/nib4s89wb9nC0WXZlaZp5B4+TOb5YJ1gveQePVowxqlCeZyrxuDRti1abi6ZW7eSvmQpKTPzu7s4OWGKiMClWizOsdaLKSoKnbGU/L0XQpRoRQncrwCrlFJ/AApoQv566mvRNK37de7XsM3Jl7cszCuM+8LuY8r2KfSI6UGA2+Wz2J3rBDNhVRLzNh2mV8PQ4i/yZsX1hIwTsPQtcPODNu9aN8wRQpRZerOZkPFfs79PXw4NHUrw6FGOLslmtNxcspOS8kP19oJwbUnJ3y5Cp8NYuTKudeviXLUqzjFVMUVH4+Ttfflj5Qf1c1u3krllK5lbt5K2eAlnZsy0DjAYMEWE4xJb7aIQHikhXAhxw64buDVNW5S/Kc2d+Tc9pWnaSfuWVfwG1BzAwqSFjNs8jtcavHbZ/Rf35C5VgRug8dOQfhzWjAWzPzT+n6MrEkLYmd7dnYrjv2b/o304OORJXNu3Iy03F72PD3pvb5x8fNB5eKBK8AdwS2YmWYmJhWatsxIT0TIzAWsvclNUFB5t2uAcUxXn6GjrrLTLlU+Av5RSCkNQEIagIDxatwasITzn0GEyt1oDeOaWLaT99htnZsywHmQw4BwRgbuPD8nHjluDeGQESkK4EOIaijLDDZAHHAecgRil1Pk+22VGsHswnSI6MStxFr2r9SbEPeSyMZ3rBPPm/G1sP5pKdKCHA6q8SUpBm/cg4yQsGWad6a79iKOrEkLYmd7Tk4rfTGB/3364z5zFwZmXbI7j5ITeywsnb29rEPfxxsnbGsj1PtZQrvf2ufCzlxfKqaj/2bgxeSkpF81aW9ddZ+1Jgrw8AHTu7jhXrYp3167WWeuqVTGFhdm8HqUUxuAgjMFBeLS5OIQfKpgFz9y6Bef16zm6apX1IIMB58jIgllw52qxOEdICBdCXFCULiX9gKFYu4hsxDrT/RfQ0r6lFb/+NfozZ9ccvtj4Be82efey+++vFcS7CxOYte4gr9wb44AKb4FOBx2+gLMnYd6T4OoLUdfaCFQIURbovbwInT6NlXPnUi8qirzTyeQlnyb39OkLPycnk3c6mayE7WQkJ19YnnEFOk/PSwK6d+FQ7n3R7T4+6JydCx2vaRq5x47lz1pvs85ab0sg5/DhgjFOAQE4R0djbtUqf1lIDIagIIfNxltDeDDG4OCCzi/Lly2jYUQEmVu2FMyGpy5axJn8trXKYMAUGYlztWo4x8ZYg7iEcCFuW0WZGhgK1APWaJrWQikVDVyeRssAf1d/Hop+iIlbJ9KnWh/CvcML3e/jZqRltD8/bTjM83dHY9CXshMQnYzQ9Qf4rh3M6A095zq6IiFEMVB6PRYfH1xiY4s0XsvJIS8lpXAoL/g5mdxk6885+/ZzbuMm8pKTC2aiL3tuV9cL4dtkImv3but4AKUwVqqES62aeHXvhnPVGJyrRuNUrpytXrr9FArh1skLTdPIOXiwYCnKua1bSf3lF85Mm2Y9xGDAFBWFc2wsLjWq4x4fj97Ly5GvQghRTIoSuDM1TctUSqGUMmmatl0pFWX3yhykT7U+TE+czpiNY/ikxSeX3d+5Tgi/bj3G0oRj3F2t2Haitx2Tu3UL+G/awOQuuFYf7uiKhBAljDIYcPL1xcnXt0jjNYsFS1qaNZQnJ5N3Oj+gJ5+x/pwf0C0ZGZhbtrAG65iqmCKj0Jvd7Pxqio9SCmNICMaQkMIh/MABMrdu5dyWLWRu3UbqwoWcmTaNo8PexBwfj1fHB3Br1Ail1zv4FQgh7KUogfugUsoLmAMsVkolA2V2z3AvZy96xfRi7KaxbD25lVjfwjNCzaP8qOLnxpvzt9EgzBdPV4ODKr0FZj/rFvATWlNz0zBo1AK8Ll+zLoQQRaF0OvSenug9PaFyZUeXU6IopTBWrIixYkU82rYFrCE8c9s2UubMJXXePNIWLcLJ3x/PDh3wfKADJnkPhShzrrsmQtO0BzRNO6Np2jDgNWAC0MHehTlSj5geeJm8GLXh8lZaBr2OT7rW4kRaFq/N3eKA6mzEOxQemY0+7xz80BHOnnZ0RUIIcVtQSuESG0vgKy8TvnIFQZ9+iqlqNKfGj2dP23vY+9DDnJk5k7z0DEeXKoSwkRtahKxp2h+aps3TNC3bXgWVBGajmb7V+rL68Gr+PfrvZffXCPZiaHwE8zYdZu7GK+5GXzoEVuO/6q9C8j748UHIlj/uQghRnHRGIx53t6HiV18RvmwZfs88TV5yMkdefY2dTZpw+IUXyfh7LZrF4uhShRC3oJSd9Vd8ukV3w9/Fn1EbRmHdo6ewAc2rEFfRi9fmbOFIyjkHVGgbKV6x0PkbOLwepveCvBxHlySEELclQ4A/vo89RtjCn6k0ZTKe991H2pIl7O/Vi91t7ubE2LGFurkIIUoPCdxX4ezkTP8a/Vl/fD2rDq267H4nvY6RXWqRa9F4dsYmLJbr7nZfclW9D+4dCbsWw9zBIDMpQgjhMEopXGvXpvzwt4hYtZIKH36AoUIFTn4+il3xrdjfpw8p8xdgyd8ASAhR8kngvoaOER0JMgcxasMoLNrlITTU143X7oth9a5TTPxzb/EXaEt1H4UWr8DmqbDkdUdXI4QQAtC5uODZvj2VvptIlSWL8R04kOy9+zj83HPsbNKUI28M49ymTVf8JlYIUXJI4L4Gg97AwFoDSTidwJJ9S644plu9EOKj/Xl/0XZ2Hksr5gptrOlzUO8x+HMUrP7c0dUIIYS4iDE4GL8hg6myZDEVJ36LuUVzUubOZW/Xbuxp145TEyaQe+KEo8sUQlyBBO7ruLfyvYR5hjF642jyLJdv7KCU4v1ONXA3OfHUtI1k55bi5RhKQdsPIKYDLH4NNk11dEVCCCEuoXQ63O68k6APPyRi5QoC33oTvdmd4x+NYGfzFhwYMJDUxYvRsst0fwMhShUJ3Neh1+kZXHswSSlJLNiz4Ipj/NxNvNexOlsPp/LpksRirtDGdHroOA4qN4W5g2DnYkdXJIQQ4ir07u54d+lC6NQphC38mXJ9HuXclv84NORJdjZrzrH33iNzxw5HlynEba8oG9/c9lpVbEVMuRi+2PQF91S+B4P+8s1uWscG0rVuCF/+sZsW0f7UC/VxQKU24mSCrj/Cd/fB9J7Qcx6E1HN0VUIIIa7BFBaG/zPP4Dd0KOmrVpEy+ydOT57C6e++xzkmBs+OHfG8795i205es1iwnD2LJT294JKXlo4lI//n9HQs6Rm47d5FSkoKpogIjFWqoDOZiqW+ksCSnU327t1k7thB1o5E8k6fxq1JE8zNm5epXViFBO4iUUoxpPYQBiwZwKyds+gW3e2K415rF8Nfe07x9PSN/DK0KWZTKX57nT3g4ZkwoTVMfhD6/Ap+UY6uSgghxHUoJyfcmzfHvXlzcpOTSZ2/gDM//cSxt9/m+AcfYG4Vj1fHjrg1bHjF7eQLgnJaWqFgbMlIJy8tzfrz+QCdkX/f+bHnr6enY8nIgCKczOmmFId/WWS9otdjrFQJU2QkpsgInCMjMUVGYggORulK75fymqaRe+wYWTt2kLkjkawdO8hK3EHWniTIsy5XVUYjOldXUubORZlMmJs2xaPt3ZibNUPnJuG7tCvFibB4NarQiDj/OMZtHsf94ffj4uRy2RizyYmRXWrS5au/eGv+Vj7sXNMBldqQ2R96/GQN3ZM6Qt/fwDPI0VUJIYQoIidvb3x69sCnZw8yExI4M/snUufPJ+2XRTgFBGAKD78oVKdfCMpFoHNzQ2c251/c0LuZcQoIQOduRm82o3O76L6Cce6Fr7u58cfy5TSoVImsnTvJTEwkK3EnmVu3krZoUcFzKVdXTOHhhUK4KTISJ5+S922y5exZsnbtKpi1ztqxg8zERCwpKQVjnCqUxzkyCnPLeJyjIjFFRWGsVAl0Os5t2EDqL4tI/XURaYsXo5ydMTdrZg3fTZuic3V14KsTN0sCdxEppXgy7kl6L+rN1O1TebTao1ccVzfUhwHNqzBm2W7iqwbQJjawmCu1MZ/K8Mgs+PYe6xbwj/4CriXvD5wQQohrc65alcBXquL/3LOkL1tOyty55J06hc5sxikw0BqUC0LyxUH5kpBsNqNzdb3i7PhN0eutYTo8HI+2bQtutmRkWINrfgjPSkwkfenvpMycdeFQX1+cIyMwRUQUhHBTeDg6l8snxWxNs1jIOXQof9b6QrjO3r+/YGZfubriHBGBR5s2mKIicY6KwhQZid7D46qP61qnDq516hDw0oucW7/eGr5/+420X39FubhYw/fdd2Nu1rRYXqewDQncN6BOQB0aBTViwpYJPBj5IGaj+YrjhsZH8kfiCV6a/R9xFb3xcy/l69HK14Duk+GHTjClG/SYA0b5hC2EEKWRzmjEo01rPNq0dnQp16Rzc8OlZk1cal74tljTNPJOnSIrMbFQEE+eNh3t/EZASmGoGGKdCY+4MBturFTxpj8k5KWlkbVzpzVcb9+RvyQkEcvZs5c8ZxQe7doVhOtbWQqj9Hpc69XDtV49Al55mbP/riN10S+k/baYtEWLUC4uuLdojvvd+TPfzs439TwllZadjTIaHV2GzUjgvkFDag+h24JuDFw6kA+afEB5c/nLxhiddHzSpRb3jVrFC7M2M6FXXZRSDqjWhio3hY5fw4zeMPNR60mVevn1EUIIUXyUUjj5+uLk64tbw4YFt2t5eeQcOFAohGclJpK29PeC3ZOVyYSpShVrAL9oRtzJ36/gv9FaXh7Z+/aTlVh41jrn0KGC59J5eOAcGYnnAw9cmLUOD7frOmul1+N2R33c7qhP4KuvcvaffwvCd+rCX1Curri3aIFH27uLtG6+pLFkZJC5YweZW7eRuW0bmQkJZO/ZQ+Tfa8rMLL4kphsUWy4yGEeQAAAgAElEQVSWD5t+yJt/vUmn+Z0Y3nA48ZXiLxsXEeDOi22jeXP+NqasPcBDd1R0QLU2FtsBzn4MPz8N84fC/aOtvbuFEEIIB1J6PcbQUIyhodD6wsy9JTOTrN27C4XwjNWrSZkzp2CM3tMTU2QklnPnyNq5Ey0rK/8OPcbKobjUrIlXly4F4dopMNChk2hKr8ftzjtwu/OO/PD9D6m/LCLtt99I/fln/EwmDt11Fx5t78atceMS1/Ul78wZMhMSyNyWYA3X27aRvXdvwQcFfblyOMfEYG7WzNpLXgL37att5bZUK1eN51Y8x1PLn6JrVFeerfsszk6Fv87p1SCUpQnHGb5gGw2rlCPUtwycZVyvL6Qfhz/eB7MftBrm6IqEEEKIK9I5O+MSG4tLbGyh23OTkwuF8KydO9G5m/Hu3h1TVBTOUZGlokWhcnLCrUED3Bo0IPD118j4+292fPMtGStXkrpgATo3N8zxLfG4uy1ujRuhK+YlGrknThSE6vMB++JvC5wqlMe5agwe992Lc0wMzjGxhb5xKEskcN+kEI8QJrWdxGfrP+O7bd+x/vh6RjQdQZhXWMEYnU4x4sGatPl0Bf+bvpEZjzfASV962xoVaP4iZByHVZ+Amz80GOjoioQQQogic/L2xil/iUZZoZycMDdqRFpODnGNGpHx91rrspPFS0idNx+d2Yx7fDzube/G3LChTddHa5pGzqHDZCZsKwjYWdsSyD1xomCMsVIlXGrWwKtb1/xwHYOTt7fNaijpJHDfAoPewLP1nuWO8nfw6upX6bqgKy/Wf5GOER0LPp0FejrzdodqDJmygbHLd/NkfISDq7YBpeCeEZBxEn59Cc6egqbPgqFsfO0jhBBClGbKYMDcuBHmxo3Q3niDjDVrrMtOliwhZe5cdB4euMfHW5ed3HnnDYVvzWIhe+++/LXWF2avC9oe6vWYqlTBrWFDnGOtwdoUHY3efOVGE7cLCdw20CS4CTPbzeSlVS8x7K9hrDmyhtcbvI670R2AdjUrsCThGJ8t3UmzSD9qhhTPLl92pdNbT6Jc4AYrR8CWWXDfSKjS0tGVCSGEECKfMhgwN2mCuUkTtGFvkPHXXxfC908/ofP0LBy+DRd209Zycsjas6fQyYxZCQkF3VmUwYApKgqPNm1wjqlqDdeRkWWuY4otSOC2ET9XP75q9RXfbv2W0RtG89/J//io6UdU96sOwFv3V2Nt0mn+N30jPw9pgovRRv1LHcngDA98CTW7wYKnYdIDUK0ztHkX3AMcXZ0QQgghLqKMRszNmmFu1gxLdjYZf/5JWv4JlymzZ6P39MTcKh6ld7IuC0lMtJ64SH5P8ehoPDt2tC4JiY3BFBZWKKCLq5PAbUN6nZ5+1ftRN6AuL6x4gZ6/9GRI3BB6x/bG08XAxw/W5KHxf/PeLwm8dX81R5drO2HNYcCf1jXdq0bCzsVw1zCI6w2leCteIYQQoqzSGY24N2+Oe/Pm1vC9arV1zfeiX8HJCeeYqng/8kjBeutb6WMuJHDbRS3/WsxoP4Nhfw7jk3Wf8PeRv3mn8Ts0DPelb+PKTFiVRMtof5pH+Tu6VNsxOEOLl6B6Z1jwP+tl4xS47xMILEMfLoQQQogyRmc04t6yBe4tW6Dl5YFOVyY7hTiSTD/aiYfRg4+bfczrDV5n3bF1dJ7XmT8P/8lzbaKIDDDz/MzNJGdkO7pM2/ONgF7z4YGv4PRu+Kop/PYaZGc4ujIhhBBCXIfS6yVs24EEbjtSSvFg5INMuXcKXiYvHl/8OF9s/pyPHqxG8tlsXpnzH1op3BHqupSyruse/C/Ufhj+/BzG3Ak7Fjm6MiGEEEKIYieBuxhEeEcw5b4pdI7szDdbvuHDTU/Sr7kXC/87yk8bDl3/AUorVx9oPwoeXQRGN5jSFaY9Aill+DULIYQQQlxCAncxcXFy4Y0GbzCi2QiSUpKYc+IZoqvs5o25WzmYfNbR5dlXpQbw+AqIf916QuWY+rDmC7DkOboyIYQQQgi7k8BdzNqEtmFG+xmEeYVxyPg1mu8M/jd9LXmWMri05GJORmjyDAxcAxXvhEUvwtct4NB6R1cmhBBCCGFXErgdIMgcxMS7J9Kvej+Ux1q2quF88Pvvji6rePhUhodnQudvIe0ojI+Hhc9DZqqjKxNCCCGEsAsJ3A5i0BkYGjeUca2+wtmUxeQDz/LZ2u/K5kmUl1IKqnWEwf9A3b6wdpx1mcnWOXA7vH4hhBBC3FYkcDtYg6AGzLhvJrqscMYnjGDosv+RkpXi6LKKh7Mn3DsC+i0FN1+Y0Qsmd4XkfY6uTAghhBDCZiRwlwBVygXySbNRZB67h+UHlvPg/AfZcHyDo8sqPsF14LHl0Pod2LsKxt4Jqz6FvBxHVyaEEEIIccskcJcQ8VUD6RLxCBl7nyAnT/HookcZt3kcebdLJw+9EzQcDIP+hrAWsOQN+KoZHFjr6MqEEEIIIW6JBO4S5JV7q1LRLZqsvUNpGXIXozaMov/i/hw/e9zRpRUfrxDoPhm6TYbMFJhwF8wfCueSHV2ZEEIIIcRNkcBdgrganRjZpSbHUkA79hBvNXyL/07+R+d5nVlxcIWjyyte0fdaZ7sbDIb1k2B0Pdg8Q06qFEIIIUSpI4G7hKld0ZvBLcL5aeNhjOfuZOq9U/Fz9WPQ0kF8+M+HZOdlO7rE4mMyQ5t3oP9y8KoIs/vBpA5warejKxNCCCGEKDIJ3CXQ4Jbh1Az25JU5/+Gmq8DkeyfTLaobk7ZNoscvPdifut/RJRav8jWg72K4Z4R1o5yxDeCPDyE3y9GVCSGEEEJclwTuEsig1/FJ11pk5uTx3MzNGHVGXrnzFT5t8SkH0w7y4PwHmb97vqPLLF46PdR/zNq7O/peWPYOfNEIklY6ujIhhBBCiGuSwF1ChfmZeeXeGFYknmDSGmtf6viK8cxqP4ton2heXvUyL658kVPnTjm40mLmHggPfgsPz4K8bPjuPvhpAGScdHRlQgghhBBXJIG7BHvkjoo0i/Tj3YUJ7DqeDkCgWyAT2kxgQM0B/Jr0K+3mtGNywmRyLbkOrraYRbSCgWug8dPw33T4JBam94LtP0PubbTOXQghhBAlngTuEkwpxUeda+Bi0PP09I3k5FkAcNI5MbDWQGbdP4vYcrG8t/Y9ui3oxvpj6x1ccTEzukKrN2DAnxDX07ppztSHYEQEzHvSet1icXSVQgghhLjNSeAu4fw9nHmvY3U2H0xh1NKdhe4L8wxj3F3j+LjZx6Rkp9BrUS9eXvkyJ8/dZssr/KLgno/gme3w8EyIbAP/zYSJ98Kn1WHx63D0P2kpKIQQQgiHkMBdCtxdrTyd4oIZvWwX6/cX3gBGKUXr0NbMvX8u/ar345e9v9Dup3ZM2jaJHMtttjW63gARd0HHcfDcTug0AQJi4a8x8GVja3eTlR9D8j5HVyqEEEKI24gE7lLijfYxlPd04elpG8nIuny9tqvBlaFxQ/mp/U/U9K/Jh/98SJf5Xfjn6D8OqLYEMLpB9c7w8HR4JtHaUtDZE5a+BZ/VgAlt4J/xkHGbnXQqhBBCiGIngbuU8HA2MLJLTfadPsvbPydcdVyoZyhfxH/Bpy0+5WzOWfr82ofnVzx/e20Pfym3ctaWgn1/haGboOVrkHkGfn4GPo6EH7tYl6BkZzi6UiGEEEKUQRK4S5E7wsrRv2kYU9buZ2nCsauOU0oRXzGeOR3m8ETNJ1i6byntfmrHt1u+JSfvNltmcinvUGj6rLXDyROr4M6BcGwLzOoLH0XA7P6wczHk3WZdX4QQQghhNxK4S5mn74qkankPXpi1mVPp195p0cXJhUG1BjHn/jnUC6zHyHUj6TS/E2uOrCmmakswpSCwOrQeDk9tgV4LrEtQEhfBj53h4yhY+Bwc+EdOthRCCCHELZHAXcqYnPR82rUWqedyeWbGJjJz8q57TIhHCKPjRzO65Why8nJ47LfHeGb5MxzNOFoMFZcCOh1UbgLtP4dnd0LXHyG0Eaz7Dia0gs9rwe9vw4lER1cqhBBCiFJIAncpFBXozhvtY1i+4wRdx63hWGpmkY5rFtKMOR3mMKjWIP44+Aft57Tnt5TfyM6TjWIKOJmg6n3Q5Xtrp5P7x1qXoaz8GMbUg6+awp+jIfWwoysVQgghRCkhgbuUeviOSozrUYedx9JoP3oVmw6cKdJxJr2JJ2o+wdwOc2lQvgHzz8yn47yOrD602s4Vl0LOnlD7Yeg5F55OgDbvgtLBb6/AyBj4rh2snwTnivbeCyGEEOL2JIG7FGsdG8jsgQ0x6HV0+eov5m48VORjg8xBfNbyMwb4DwDgiSVPMPT3oRxKL/pj3FbcA6HBIOi/HAb/C82ehzMHYN5gGBEJ0x6BbXNxykl1dKVCCCGEKGEkcJdy0YEezB3UiJohXgydupGPft2OxVL0k/xiXGKY3X42Q+OG8teRv7h/zv18uelLsvKufULmbc03Alq8DE9ugH6/Q91HYf8amN6Txqt7WHe3nN4TVn0Cu5fBueTrP6YQQgghyiwnRxcgbl05s4kf+t7BG/O2MmbZbnYcTefTbrUwm4r2z2vUG+lXvR/3Vr6Xj/79iDEbxzBv9zxerP8iTYOb2rn6UkwpCK5jvbR+B/b/xe6VM6jikgqHN8C2uRfGeodChdrWS/laUL4muHg5rHQhhBBCFB8J3GWE0UnHuw9UIzrQnbcWbKPT2D8Z36suIT6uRX6M8ubyjGw+kr8O/8V7a99j0NJBNAtuxgv1XyDEPcSO1ZcBeieo3IQD+/Ko0ry59bazp+HIJmv4PrwBDq2DrT9dOMYn7PIQ7uzhkPKFEEIIYT8SuMsQpRS9GoYS7m9m4I/raT96FWMfrkODKuVu6HEaVGjArHaz+CHhB77Y9AUd5nSgT/U+9K3WF2cnZztVXwa5+kCVFtbLeRmn4MgGOLzRGsL3/w1bZl24v1wEVKh1IYgH1gCTufhrF0IIIYTN2DVwK6XuBj4D9MB4TdPev+T+isB3gFf+mBc1TVtoz5puB43CfZk7qBH9vv+XHhP+Zlj7WB65s9INPYZBb+DRao9yT+V7+Hjdx3y56Uvm757Pc/Weo2VIS5RSdqq+jHMrB+GtrJfz0k/AkfwAfngj7F0N/83Iv1OBb2R+AM8P4oHVwehW6GEtmoXF+xYz/r/xHMk4Qpx/HPUD61O/fH3CvcLRqdvvdI1f/jvC0dRMejUIRaeT31chhBCOY7fArZTSA2OAu4CDwD9KqXmapm27aNirwHRN075QSsUAC4FQe9V0Own1dWP2wIYMnbKBV+dsYcfRNF5vF4NBf2PBK8AtgA+bfsiDkQ/y7t/v8tSyp2gU1IiX6r9EJY8bC/HiKsx+EHGX9XJe2rHCIXzPMtg81Xqf0oFvFFSoTV75Gvyqz2XcwcXsTk0i1COUZsHNWHdsHcsOLAPA2+RN3cC61gAeWJ/KnpXL/AemoymZPD19E+dy8li96xSfdK2Ju7PB0WUJIYS4Tdlzhrs+sEvTtD0ASqmpwP3AxYFbA84vWvUEZDcRG/JwNjC+Vz0+/HU7X/2xh13H0xn7cBzebsYbfqx6gfWY3m46UxKmMHbTWB6Y+wC9Y3vTr3o/XA1FXycuisg9ANzbQGSbC7elHikI4bmH1rPw4O98fXIZe40GwrNz+NDiTmuPEPTmWKj3EodzUll7dC3/HP2HtUfXsnjfYgB8XXypF1CPeuXrUT+wPhXdK5a5AP7hr9vJs2gMaRnO2OW76TBmNeN61qWKnyzPEUIIUfzsGbiDgAMXXT8I3HHJmGHAb0qpIYAb0AphU3qd4qW2VYkKcOfF2f9x/5jVjO9Vl8gA9xt+LIPOQM/YntwTdg8j/x3J1/99zbzd82gW3IzKnpULLoFugbflEga78yhPjtmXBbpMxp1eycE8I1EeYYz0b0B8Zg66Ixth5yLY9CMsfI4KsQ/QIa4nHRq9jQYcTDvI2qNrC0L4L3t/AcDf1b9g9rteYD2C3YMd+zpv0eaDZ5i9/hCPNwvjmdZRNKziy6DJ6+kwejWfdK1Fq5gAR5cohBDiNqM0reg9m2/ogZXqDNytaVq//Os9gDs0TRt80Zin82v4WCnVAJgAVNM0zXLJY/UH+gP4+fnVmT59ul1qLut2n8nj8w1ZZOVqPFHTRC1/J9LT0zGbb27Wb3fmbhamLORg9kHOWs4W3G5URvwN/gQ4BRBgCCDQEEiAIQA/gx8GVba/1r+V9/NacrQc/k7/m8Upizmdd5oQYwhtPdtSzaVa4dlpTcMzJYHAo0vwP74KvSWLDNdgjga24mhgC3KMXvnDNI7nHmdn5k4SMxPZlbmLNEsaAD56HyKcI4hwjiDSORJvJ2+bv56iuJn3UtM03v07k2NnLXzQ1BUXJ+t7c+qchc83ZLEv1cID4QbaVTGgK2Oz+tdjr9/N25W8n7Yj76VtyftpWy1atFinaVrdW30cewbuBsAwTdPa5F9/CUDTtPcuGrMVayg/kH99D3CnpmnHr/a4UVFR2o4dO+xS8+3gSMo5+n+/ji2HU3i+TTTR2n5atGhx/QOvQdM0krOS2XNmD0mpSSSlXLgcTj+MhvV3TKEIMgcR5hVGZY8LM+JhnmF4OZeNntTLly+n+fm2gDaQlZfFrMRZfLPlG46dPUYN3xo8XvNxmgQ1uf4ykKw0axvC9ZPg4FrQOUFEG4jrAeF3WVsZ5tM0jd1ndhfMfv9z7B9SslIACHEPKZj9rh9YHz9XP5u9vmu5mfdywebDDJ68gfc6Vqd7/YqF7svMyePl2f8xe8MhWscE8HGX22tdt61/N2938n7ajryXtiXvp20ppWwSuO25pOQfIEIpVRk4BHQDHrpkzH4gHpiolKoKOAMn7FjTba+8pwvTH2/A87M288Gi7TSooKdB4zycDfqbfkylFD7OPvgE+lA3sPDv5Lncc+xP3U9SShJ7UvYUBPG/j/xdaDdLb5N3oWUp5y8V3Cqg1918baXVudxzzNgxg4lbJ3Li3Ani/ON4q9FbNCjfoOjrrU3uENfTejmxAzb8AJumwI6fwRwANbtD7UfANwKlFOHe4YR7h/NQ1YewaBZ2Ju8sWILy297fmLXT2r4w1CPUGsDL16NeQD3KudxY20l7yczJ472F24kOdKdL3cv7xjsb9HzcpSbVgjx5Z2ECD4z9k3E96hAm67qFEELYmd0Ct6ZpuUqpwcCvWFv+faNp2lal1FvAv5qmzQOeAb5WSv0P6wmUvTV7TbmLAi5GPZ93q0VUgJkRvyXSddwaxvWoQ4CH7Xtsuzi5EOUTRZRPVKHb8yx5HMk4UhDAz4fxZQeWFQQ7AJPeRCWPShdCuEdlwrzCqORRCRcnF5vX62hnc84ybcc0Jm6dyOnM09QPrM8HTT+gbkDdWzux0S8KWg+H+Ndh52/WWe8/R8HqT6FiA2vwjulQ0PNbp3QF/249YnqQZ8lje/J2/jliPQHz56SfmZ5oXdoV7hVeMPtdN6Cuw76tmLAqiUNnzjG53x3or9IGUClFn8aViS7vzqAf13P/mNV81q0WLaPts647Ky+LPw78gZfJi8qelfF18S1zJ6gKIYS4Prv24c7vqb3wkttev+jnbUAje9YgrkwpxeCWEWSf2Mf4rWm0H72KcT3qUjOkeMKSXqcn2D2YYPdgmgQ3KXTfmcwzly1N2XZqG4v3LcZy0fL+Cm4VCoJ4pHcktf1rU8mjUqkMNOnZ6UzdMZXvtn7HmawzNKzQkMdrPE5cQJxtn0hvgOh7rZe0o9YZ7w0/wNxB8MsLEPuAdUY8uJ516/rzh+n0xJaLJbZcLL2r9SbXksu2U9sKlqDM2TWHKdunoFBEekdyR/k76BnTkwC34jlB8XhqJmOX7eKumAAahvted3zDKr7MH9KYxyeto+93//LMXZEMahFu09+dXcm7eGHlCyQmJxbcZjaYC35nQz1CC36u6F4Rg/72Wd4ihBC3G9lp8jYXF+DE7Gb16ffdv3T56i8+7FyD+2sFObQmL2cvajvXprZ/7UK3Z+VlXXF5yvrj6zmXew4AH2cfavtbj60TUIconygMupIbZFKzU/kx4Ud+2PYDqdmpNAlqwuM1H6emX037P7l7IDT+HzR6CvavsQbvLbNhwyRrn+/aj0DNbmD2v+xQJ50TNfxqUMOvBv2q9yMnL4ctp7aw9og1gE/ZPoWfdv3Eq3e8StvKbe3+IWjEbzvIzrPw8j1Vi3xMsLcrM59oyEuzNzPit0S2HEplRJeamE239mdR0zSmbJ/CyHUjcTO4MbL5SNyN7oU+QP595G/m7Z5XcIxeWT+AXnxuw/mLp8nzluoRQgjheBK4BdGBHswd1IgBP65n6NSN7DiaxrOto0rc7nwmvYkI7wgivCMK3W7RLOxN3cuGYxtYf3w964+tZ+n+pYB1SUsN3xrUDrCG8Jp+NXEzuF3p4YvVmcwzTEqYxOSEyaTnpNMipAWP13icWN/Y4i9GKajUwHpp+/6FEy0XvwZL37zqiZYXM+gNBR90Hq/5OPtS9/Hyqpd5YeUL/H7gd16941W7LTXZciiFGesO0q9xZSr73ti/rYtRzydda1EtyJN3FybwwJh0vu5Zl9AbfJzzTp47yeurX2floZU0DmrM8EbD8XWxzrjfWf7OQmMzcjLYm7q3UBBPSkli9eHV5FhyCsb5OPsUmg0/v7Sqgvn2PL9BCCFKIwncAoByZhM/9L2DYfO3Mnb5bhKPpfFpt9q3PNtXHHRKR5hnGGGeYXSK7ATA8bPH2XB8A+uPrWfD8Q2M2zwOi2ZBr/RE+UQR5x9Hbf/axAXEFQSi4nA68zTfb/2eKduncDb3LHdVuov+NfoT7RNdbDVc0w2caHktlTwq8d3d3zFx60TGbBzDumPreLPhmzQNbmrTcjVNY/iCbXi7Ghnc8to1XY1Sin5NwogO9GDwlPW0H72Kz7vXpnnU5TP717Li4ApeW/0a6dnpvFT/JbpHd7/mzL6bwa1gmc7F8ix5HM44fFkQv/T8BoPOUPj8hvwgHuoZWiI+VAohhLig5KcpUWyMTjre6VCN6EB33py/jY5jVzO+Zz0qlit9O0n6u/rTJrQNbUKtOzWmZ6ez+cRm1h+3BvCZiTP5IeEHACq6VywI37X9axPqEWrzJRAnz51k4paJTE+cTmZuJneH3s1jNR67bLa+RLnBEy0v5aRzol/1fjQJasJLq15i0NJBdIroxHP1nrNZIPx161H+TjrN8A7V8HS5taVDjSN8mT+4Mf0nrePRif/wbOsoBjavct3fhczcTD5Z9wmTt08mwjuC8a3H39K/q16nJ8Q9hBD3kMs+oJzJPHPZrHhiciK/7/+dPC2vYJy/q39BAD8fxi/ulS+EEKJ4SeAWhSil6NkglCp+Zgb+uJ77x6xi7MN1aFClZLR+u1lmo5mGQQ1pGNQQgJy8HBJOJxTMgq84uIK5u+cC1q/wa/nVIi4gjjj/OKLLRd/0OvBjGcf4duu3zEycSY4lh3sr30u/Gv0I8wyz2Wuzu6KcaFm7B4TUL3Si5XlRPlFMvXcqYzaOYeLWiaw5soa3G719WQvJG5WVm8c7CxOIDDDTvd7lbQBvRoiPK7MHNOT5WZv56NcdbD2cwkeda+J2lW96dpzewYsrX2TXmV08UvURnqrzFCa9ySa1XImXsxe1nGtRy79Wodtz8nI4kHbAGsIvOuF4wZ4FpOekA+CsnNm3cR89YnpgNkorRCGEKE4SuMUVNQr3Ze6gRvT7/l96TPibYe1jeeTOSo4uy2YMekPBSX+9YnuhaRpJqUkF68A3HN/A7wd+B8BZ70wNvxoFs+BFWQd+JP0IE7ZMYPbO2Wiaxn1V7uOx6o9R0aPiNY8r8a55omVk/omW3S870dKoN/K/Ov+jeUhzXln1Cn1+7UPPmJ4MiRty0wH129V7OXD6HJP61sdJr7PFqwMutM2sHuTB+79sZ/fxDMb1rEOlchf+zS2ahckJk/lk3Sd4mDz4stWXNApyXMMlg95AmFcYYV6FP8hpmsbJcyfZdWYXY1aPYeymsfy4/Uf6VOtDt6huuBpK37dXQghRGtltp0l7kZ0mbet6O1KlZuYwdMoGlu04QY87K/F6uxgMNgw3Jdn5deDnZ8F3JO/AolmsPaq9owqWoMT5xxXsvjhz8Uy2uG0pmC3vEN6BvtX6Euwe7MiXYl/nd7Tc8AMc+Nu6o2X4XVChFnhVAu9K1v91Lw86HWdzzjJy3Uim7ZhGFc8qvNvkXWLKxVz2sNf63TyRlkWLEcu5o7IPE3rXs9tLW7nzBIMnbwDg8+61aRbpx4mzJ3ht9WusPrya5sHNebPRm/g4+9itBltZvnw5ftX9GL1hNKsOraKcczkeq/EYD0Y+iFFvdHR5pY7s5mc78l7alryftmWrnSYlcN/mivJ/zDyLxoe/buerP/bQIKwcYx+Ow9vt9vsPdEZOBv9v787jo6ru/4+/zmxZZrLvKyFkg4QlYV9kVRYRUFyqKGK17trWtn7Vaqu1rbW2tb+quINLXeqCCAIKbqggIJAAYSeEJYSEQDYSsif398edrCSBQCbr5/l43MfcuffO5MzJzeQ9Z849Z3vO9roW8B0nd1BWXQbo05+Hu4ezIXMDRoORq6Ov5taEWwmyBXVyqTtY7YWWuz6Fwgz0+azsjBbwDK8L4estRv546kfyqkq4c8ACfpF4H6YGY1G3dm4+8kkqH23JYPUD4+nn4Jkij+aWcMd/t7D/RBHXjj/NhsKXKK0q5cHhD3JtzLXdZtz3hvWZfCKZ51OeZ8uJLQRaA7lr0F3MjprdpYfQ7Gok1LQfqcv2JfXZviRwi3bRlj/MT5KP8fAnqQS6O/P6gmHEBLg5tnBdXGVNJXtz99YNRbg/fz9RKorHpj3WYRO+dGlV5VCQAfmHoeAw5B+BgiP1t6X5FBoUT/l4s8pmJaGiir9We+rdIjz7cOBUBXRNPR4AACAASURBVNHDL61vIbfo3R92Hz/NFc//wIIxETw+q2OGUcwrKeL6JY+SVfMtVsJZdPm/ifeL6ZCf3V6a/q1rmsbGrI08n/I8qadSCXcL5+4hdzMjYoYMN3geJNS0H6nL9iX12b7aK3BLH25x3uYmhdLX18od/93K3Bd/5D/XD2FK/94bLM0GMwP9BjLQbyAL4hcA+hudhG07kxP4RulLc8oK8Sg4yt/zDzP56Df85cR3XKed4YHTe7gh7Wuiq0oh7bX6463+aF59yM+18nsnL+b5XgLpJ/Uw7hGqX9zpAHty9/DQDw+RVXOIIe5Xsm7zMH6bn82r80O75Qg+tZRSjA4ezaigUXx37DteSHmBR354hEWpi7h3yL1MCZ/SbVrvhRCiq5PALdokMdyL5feN5Y63t/KLt7fwf9PiuGtCpPxjFm3n7AGBAyFwINP6z2Jo6Sme+PEJnj72Hd8Mv5wra8YzKyHe3ip+CPKPkJd5gLAzuxhtzMOwemn9cykjeIQ06DMeod8GJIB//2ZHTzmXGq2Gt3e9zX9S/oO3kzevXvYqo4NH8138Se5/L5nZC9fxwg1JjIvuuHHcHUEpxcSwiYwPHc+aI2tYmLKQB9Y+QH/v/tyfeD/jQsbJ37cQQlwkCdyizYI8XPjortE8+PEO/v7FXt7ZeIQrBgcxa1Aw8cHu8s9ZXBBfF1+en/w8S9OW8vef/s6O6lRqyh9jdsLVKKWoqKrh6n9/h9FD8cUvx2AozqrvopJ/uH79wJdQfKL+id1DIPoyfcbMvuNbHDe8oRNnTvDo+kfZlLWJKeFTeGL0E3UzZU6I8WP5feO4479buHnxJh6eEcftl3T/D50GZWB6xHQuDb+UlekreWn7S9zz9T0k+idyf+L9DA903MWpQgjR00ngFhfE2awPnXZpf3+WpmSy6IdDvPJdOpG+Vq4YHMzswUFE+ffuPt6i7ZRSzI2ey4jAEdy/8n4eW/8YXx/9msdHP87SLYUczi3hjZ8Px2y26C3YXn2gbzNPVFmqh+9jP+mT9qQuga1v6hdu9hkL0VP1xaffWa3fXx/5msc3PE5FdQVPjH6CudFzzwrTEb5Wlt4zlgc/3s5Tq/ayM/M0f796EC6W7t/32WQwMSdqDpf3vZylaUt5Zfsr3Lr6VkYGjeT+xPsZ7De4s4sohBDdjgRuccGUUswZEsKcISHknangi53ZfLb9OM9/c4Dnvj5AXKAbswYHM2tQcLfu6yo6XqhbKPcH3E+GfwbPJT/HnE+vpDBjDhNixjPpfKZcN7uAf5y+JN0MVRWQsRH2r9ZbwFc/oi9efevCd0noUJ5J+Q9LDixhgM8Anr7kafp6NJfmdVYnEwvnJfHi2oP8c80+0nKKeWX+UMK8e8a5bjaauS72Omb3m82H+z5k0c5F3LTqJiaETuC+xPuI847r7CIKIUS3IYFbtAtvq4V5I8OZNzKcnNNlrErN4rMdWfxj9T7+sXofg8M8mTUoiJmDggjycOns4opuwKAMLIhfwLiQcSz47AEIeAv34BxOV8ThbnFv25OZLHp3kr7jYdpf9S4oB77Ul+S32bVtMQ/7+3HEbOI2n2HcO/pRzK2E7VpKKe6dFMWAYHd++X4Ks19Yxwvzkhgb1b37dTfkbHLm5vibuSbmGt7d8y5v7HqDaz+7lql9pnLvkHvPmmxHCCHE2XrHDCaiQ/m7O3PL2L4suXsM6x6axCMz4qiuqeEvK/cw+m/fcN3LG/jvhsOcKi7v7KKKbqCqzJ/ju39BnNNcfshaw9xlc9lwfMPFPalXBIy4neob3mfRrCe4KTSEUicbi4oVv97yCebnh8LCkbDmMTj0vd5C3opJsf4sv28cvjYn5i/axOs/pNPdhlw9F1ezK7cPup0vrv6COwbdwbrMdVy1/CoeXfcoGUUZnV08IYTo0iRwC4cK9XLlzgn9WHH/JXz7u4n89rIY8ksq+MOyXYz461fMX7SJDzdnUFhS2dlFFV2Qpmn8ZeVubE5OvDLrUd65/B1cza7c8eUd/G3T3yitKr3g584+k83tX97O/9u2kEnhU1hy3VcMv2c73LcFpj2lT2O/8WV4axY8Ewkf3ATJ/4XTWc0+X19fK0vvHctlAwL4y8o9PPDBNkorqi+4fF2Vu8Wd+xPv5/OrP2d+//msPrya2Utn8+SGJ8k+k93ZxRNCiC5JupSIDtPX18r9U6K5f0o0+7KL+Gz7cT7bcZz/W7KDRz9NZUKMH7MGB3Np/wCsTnJqCvh2Xw4/HDjFH64YgLfVgrc1gQ+v+JD/JP+Hd/a8w4/Hf+Sv4/7KIL9BbXreNYfX8KcNf6KyppI/j/0zc/rNqb8w0jdaX0bfC+XFcOg7/cLLA1/Cns/0YwIH1V94GToM7BPF2JxMvHTjUF5cm8a/vtzPAXu/7lCvntGvuyFvZ29+N/x33Bx/M6/ueJUlB5awLG0Z18Vex20Db8PXped0qxFCiIslqUZ0ithAN2IDY/nt1BhSMwv5bPtxVuzI4qs9OTibDUyJC+CKQUFMivPH2dz9R34QbVdVo/HUyj1E+lqZP6pP3XZnkzMPjXiISWGTeGz9Y8z/fD63JdzG3YPvxnyOyW/OVJ7h6Z+e5tO0TxnoO5CnL3macPfwlh/gZIO4mfqiaZCzuz58r/s3/PBPcPGCflMgZhr0m4LB6sN9k6MZEOzOr97fxuwX1vPuL0bSP6iN/c67CX9Xfx4b9Rg/T/g5L29/mff2vseSA0u4sf+N3BJ/Cx5OHp1dRCGE6HQSuEWnUkoxKNSTQaGePDKjP1uP5rNi+3FWpmaxMjULq8XI1PhAZg0OYlyUHxaT9IJqzaFTZ6iu0YjyP/dY013dt0erSD9ZwaIFw5r9vY8IGsGS2Ut4ZvMzvJb6Gj9k/sBT454i2iu62edLPZnKQz88RGZxJncMuoO7Bt+F2dCG2SmVgoB4fRn3AJQWwMFv9PCd9iXs/BhQeot39FQmR1/GsntHM+/1zdz25mY+vW8s/m7OF1gbXV+ILYQ/j/0ztybcykvbXuL11Nf5YO8H3Bx/M/MHzMdqtnZ2EYUQotNI4BZdhsGgGB7hzfAIb/5wxQA2Hcrjs+3H+XxnNktTMvFwMTMjIZBZg4MZFemD0dC9JxppL+VV1XyxM5v3Nh1l06E8LCYDL9yQyNT4wM4u2gUrKKng04MVjIvyZXJcy8MAulnc+PPYPzM5bDJPbHiCn634Gfcn3s/NA27GaO/mUV1TzaKdi3hx24v4u/qzeNpihgYMvfhCunhCwlx9qamBrG32kU/WwLdPwbd/JdLqz+fh4/hgv8aKl1cxf9JgzDYfvVW84WJyuvjydBF9PfryzIRnuG3gbSzctpCF2xby7p53uS3hNq6Pux5nU8/90CGEEC2RwC26JJPRwNgoX8ZG+fLknATWpZ3ks+1ZfLb9OP/bnIGvzYmZAwO5YnAwQ8O9MPTC8J2WU8z/fjrKkuRj5JdUEu7tyoPTYlmz+wR3vbOVv80dyM+Gt9Jdogv7f18doKQSHrui/3nN4DgpfBKD/Qfz5w1/5tmtz7I2Yy1/GfcXjMrIIz88QnJOMjMiZvDY6MfaPqTg+TAYICRJXyY+BGdOQdrXcGANXofXcYfxJIYz1bCihcebXcHZs0EI9zw7lDe33WK7oGnrO0KsdyzPTX6O1JOpvLDtBf619V+8tfstboi7gaujr8bHxaeziyiEEB1GArfo8iwmA5PjApgcF0BZZTXf7s1hxY4s/rc5g7c2HCHIw5mZA4OYnhBIUg8P32WV1azeVd+abTIopsYHMG9EH8b088FgUNwyJoK7303moSWpnCqu4J6J/brVtONpOUX8d+MRJoaZiAs8/3Ds7ezNsxOfZUX6Cv626W9cvfxqTMpEDTU8Ne4proi8ouPqweoLg3+mL4BB03j96x28+XUKd4/w5sZBblBWAKX5TRb7trz0+vutjcRiMNWHb+eWQrq+uJ7J1lviDR3bLWug30BeuewVtmRv4dUdr/J8yvO8vP1lZvSdwby4ecT7xndoeYQQojNI4BbdirPZyIyBQcwYGERxeRVf7znB8m3HeXvDEV5fdwg/NyemDghgWnwgo/v5YDb2jD7fzbVmPzQ9jmuGhuLn1rg7gtXJxOs3D+PBj7fzj9X7OFVczh9mDug2H0T+unIPrmYjV0Vb2vxYpRSz+s1ieOBwntzwJGXVZfxpzJ8IcwtzQEnbVDBumzKIvfnw6KZj2PrGMGdIyPk9trK0Pog3tzQM7sXZcHKPfnz56UZPMwIg9fcQnKT3Mw8Zpt9aO2Y0kWGBwxgWOIz0wnTe3/M+yw4uY/nB5QzyG8S8uHlM7TP1nBe9CiFEdyWBW3RbNidT3dTyRWWVfLvvJKvt/b3f3XQUd2cTl/YPYFpCIOOj/XCxdK/RTmpbs9/ddJSf7K3Z0+IDuWFEeF1rdkssJgP/vm4IPlYnFq8/RG5xBf+8dnCXv+j0u/0n+XbfSX5/eRzuNRc+mUqgNZAXL32xHUt28ZRSPHXVQI7mlfDgxzsI9XJlaB+vcz/Q7KIv7kFt+4HVlVBWqAfxkjz2rv+MOFsxZG6BH54FzT5GuGefxgE8cBCYHdfPOtIjkkdHPcovk37JsrRlvL/3fR7+4WH+ueWfXBdzHdfGXitDCgohehwJ3KJHcHM2M3twMLMHB1NWWc0PB06xelc2X+05wScpmbiYjUyI8WN6QiCT4vzxcOm6LWlpOcW8b2/NLjhHa3ZrDAbFH67oj6+bhWe+2Ed+SQUv3zS0y45xXlVdw19W7KaPjysLxkSwYV3Pm73QYjLw8k1DuerF9dz53y18eu9Yx43RbTTrrdf2Fuzs9FLiJk7U91WcgaztcGyLHsCPboKdS/R9BjMEJtQH8JBh4NOv3fuKu1ncuGnATczrP4/1met5b+97vLj9RV5NfZWpfaYyr/88BvkO6lbdoYQQoiVd8z+vEBfB2WzksgEBXDYggMrqGn46lMcXO7NZszubL3ZlYzYqxvTzZVp8IJcNCGhTiHWUskr7SCM/tb01uzVKKe6ZGIWv1YmHP9nBvNc2sviW4fjYOv81N/X+T0frJopxMnWvbyPawttqYdGC4Vz14npue3MLH989GjfnDv4AaLFCnzH6Uqsouz6AH9sC29+Hza/p+5w9IWRo45ZwV+92KYpBGbgk9BIuCb2EI6eP8L+9/+PTtE9ZdWgV8T7xzOs/j+kR07EY297FSIjzUVVTxebszZgMJhJ8E3AxuXR2kUQPJIFb9GjmBqOd/Gl2PNuOFbB6px68f780lUc/TWV4H2+mJQQyLT6gw2cETMsp4v2fMupas/v4XFhr9rlcNzwML6uF+95L5tqXN/D2bSO61OyHhSWVPPvlfkZH+jB1QEBnF8fhovxtvHTjUBa88RO/fD+F1xcM7/xhLt0Cof8V+gJQUw0n99UH8Myt8P0/QKvR93v1bdIVZeBFD2/Yx70PD414iPsS72P5weW8v/d9Hl33KP/a8i+uibmG62KuI8Da888P0TH25+9nedpyVh5ayanSUwCYlIkBPgNI9E8kKSCJRP9EvJzPo+uXEOcggVv0GgaDIinci6RwLx6eEcfe7CJW78rmi53Z/HnFbv68YjcJIe5Mjw9kekIgUf5uDilH09Zss1ExNT6QeSPCGR154a3Z53LZgAD+e9tIbntrM1e/9CNv3zqS2EDHvMa2eu6bAxSUVp73MIA9wbho/UPgY5/u5K8r9/DHWQM6u0iNGYwQMEBfkm7Wt5UX6+ON17aEH14PqR/p+4wWPXTXdUUZCt6RF9QVxWq2ckPcDVwfez0bsjbw/p73eW3HayxOXcyUPlOYFzePRP/EXnOuiPaTV5bHqvRVLD+4nD15ezApE5eEXsLsfrOxGC2k5KSQfCKZ9/a+x1u73wL0seWT/JPqAnioLVTOPdFmErhFr6SUon+QO/2D3Pn1pTEcPnWG1buyWb0rm3+u2c8/1+ynn5+VafbwPTDE46LfYJtrzX54ht6a7dtBXTxG9PXmo7tGc/Oin7j25R9ZfMtwhkW0T9eAC5V+spi3fjzMz4aFER/cu6YBv2lUHw6eLGbx+kNE+lm5qcEU9l2Skw0ixulLrdPHG3RF2Qop78BPr+j7XLwbd0UJSWpTVxSlFGOCxzAmeAwZRRl8sPcDPkn7hNWHVxPnHce8uHnM6DtDJtMRraqoruD7Y9+z7OAy1h1bR5VWxQCfATw84mEu73t5oxbs8aHjASivLmd37m62nthKSk4Ka46sYckB/ToHfxd/EgMS9VZw/yRivGLqJtoSoiUSuIUAInyt3DmhH3dO6Ed2YRlf2vt7v/J9Oi+uPUiwh7O920kgwyO8z/vr/7rW7E1H+elwx7VmtyYu0J0ld49hweKfuPH1TSycl8SlndiN46lVe3E2G/nt1NhOK0NnemzmAA6fOsPjy3cR4WNlXHQ3G6HDPRgGzNYXgOoqOLm3cVeUtV8Bmr7fI1y/KDMgQW8RD0wAz4hzjg8e5hbG74b/jnuG3MPKQyt5b897/PHHP/Ls1meZGz2X62OvJ8jWxpFcRI+laRq7cnexLG0Znx/+nMLyQvxc/Jg/YD6z+s0i2iu61cc7GZ1I9NdDNUCNVsPBgoMkn0gmOSeZlJwUVh9eDejfyAzxG1LXDUX6gYvmSOAWoolAD2fmj45g/ugI8s9U8PXeHL7YqQ/P98b6w/hYLVw2QB9ucEw/n2Yv8EvLKeK9TRl8ktJ5rdmtCfN25aO7RvPzNzdzp31WyuuGdfxY1esOnOKrPSd4aHpcl7h4tTMYDYrnbkjkmpc2cPe7W1l6z1ii/G2dXawLZzTpITowAYbeom8rL4LjKXr4zt4JJ3bC/i/q+4NbbBAQrwfw2iDuPwAsZ19n4Gp25dqYa7km+ho2Z2/mvb3v8eauN3lz15tMDpvMvP7zGBYwTL7y76VOnDnBivQVLD+4nPTCdJyMTkwOm8zsqNmMChqFyXBhscegDER7RRPtFc3P4vQJrbKKs+rCd3JOMgu3LURDw2TQ+4En+SfVhXbpBy6UpmmdXYY2iY2N1fbt29fZxegx1q5dy8TaocJEq86UV7F230m+2JXNt3tzKC6vwuZkYnKcP9MTAhkd6cNLy75n22lro9bsG0eEM6qTWrPPpbi8irvf2coPB07x8Iw47hwf2WFBpbpGY+ZzP1BcXsVXv5mAs7nxB5fedm5m5JVw1YvrsTqZ+PSesXhZ23dUji5Xn5WlkLNHD9/ZqfVBvG7CHgU+UQ1awwfp625BZ/ULP158nA/2fcCSA0soLC8k2iuaG+JuYGbfmbiaHXNxcJerz27sYuuytKqUr49+zfK05WzM2oiGRqJ/IrP7zWZaxDTcLB1zrUpheSHbT26vawXfeWonlTWVgD7+fKJ/IkMDhpLon0iILcRh77VybrYvpdRWTdOGXezzSAu3EOfJ6mRi5qAgZg4Koryqmh/TcvliZzZf7jnB8u3H646L8DHyyIw4ru4irdmtsTmZWLRgOL/9aDtPf76XU0Xl/P7y/h3y4eCDzRnszS7ixRuTzgrbvVGYtyuvzB/GDa9t5M53tvLObSO7/ERFF8XsovfpDkmq36ZpUHCkPnxnp0JmMuxaWn+Mi7e9BX2QPYgnEOwbywNDH+DuwXfz+aHPeW/vezy54Un+vfXfzI2ay/Vx1xPqFtrxr7EJTdOoqKmgpLKE8upyTAYTZoMZs8GMxWi54NbX3qhGqyH5RDLLDy5nzZE1nKk8Q4gthDsH38nsyNmEuXf8N3YeTh6MDx3fqB/4rlO7SM5JJvlEMmsOn90PvPZizGjPaOkH3sPJX7cQF8DJZGRSnD+T4vz5a3UNW47kszE9F0thBnddNbFLtma3xGIy8J+fDcHHauH1dYfIPVPBM9cMwmx0XNg7XVbJv9bsY0SENzMSAh32c7qboX28+Mc1g/jV/7bx6NJUnrmml038ohR4RehL7fCEoM+YeWK3HsBP2FvDN78OVWX6foMZ/OJwDkzgqoAErhxwLylmeDf9M97Z8w5v736bCWETmBc3j1FBo86rTitrKimtKqWksoSSqhJKK0v1W/u2lKIUju0+pt+vKqk7rqSypNG22vXax1fXzvDZDIMy6OHbYMFsNDcK4w1vzUb7Ma3ta3CM2djguCb7PJ09CbGF4OHUPS5YzijK4LODn7H84HIyizNxNbkyNWIqs/vNZmjAUAyq63xIdTI6kRSgB2oG6h8S0grSSDmRwtacrSSfSK7rB24z2xjsP5ghfkOI844j1iuWQGtg7/r77+EkcAtxkUxGA6MifRgV6cPatce7VdiuZTAoHp81AF+bhX+u2U/emQpeuikJV4tj3iIWfpNGXkkFb14xQP6hNDFnSAgHT57hua8P0M/fxl0T+nV2kTqfswf0Ga0vtaqrIO+gvTtKqt4ifvBb2P4+CkgCktyCyQ6I40NXM0uyNrM2Y23dV/sNQ3CjkGy/X9sVoFV5+o1JmXA1u+JicsHV7IqryRVXsyv+rv6NttWuu5hccDY6U1VTRWVNJRU1FVRW199W1lRSUV1Rt692vXZfeVU5xTXFzR9fXUFFTQVVNVVtqmI3sxshbiGE2PQl2BZMqC20bt1RXXPOR3FFMWuOrGFZ2jKSc5JRKEYGjeTeIfcyJXxKp5atLQzKQIxXDDFeMWf1A6/thrIwc2Hd8W5mN6K9oon1jq17XJRnVLd5vRerpLKkR71WCdxCCEAfgu2+ydH42Jx4dGkq817bxBu3DG/3vsRHcs+weP0hrkkKZWBo92hV62gPXBpN+sli/v7FXiJ8rEyXbwHOZjSBX6y+DLymfnvxyfpW8BM7CcxO5Zfp+7lTq2K11coH5ZV8X3AIV4MFV5MLLhY3PJw8CLKF4OrsdVY4rg3PLiaXunVXkyvbt2xn8iWTcTW5YjZ28Eyh50HTND2kNxPGG4b03LJcMosyOVZ8jOPFxzlUeIh1mesory5v9Hzezt51ATzETQ/hIbYQQm2hBFmD2r0Oqmuq2Zi1keUHl/PN0W8oqy4jwj2CXyX9iisiryDQ2jP+JoJsQcy0zWRm5EwAzlSe4UD+Afbn72d//n725e1jWdoySqpKAFAowt3D6wJ4jFcMsd6xBFuDu13jhaZp5JblcqzoGEeLjpJRlFG3HCs6Rl5ZHpvmbeoxoVsCtxCikRtGhONttXD/+ylc+8oG3r51BMGe7TfE1d9W7cVsNPDgtN45DOD5UErxz2sHk5FfygMfbCPUazQJIfLh5LzY/MA2GfpNrt9WVY7Tyb3Mzt7J7OxUOLVfbx0vOFg/UgqAkzt49wXvfvqkPT79wBaqr1t9G12sedR0tEt3w1BKYTFasBgtWM3WNj22YRDKLM5stKSeSuXLI19SpdW3oBuUAX9X/7rW8VBbqB7KrcGEuoXi5+J33v2TDxYcZNnBZaw8uJKc0hzcLe7MiZrD7H6zGeg7sNuFyraymq0M8R/CEP8hddtqtBoyizPrQvj+PD2If3XkKzT7cJs2s41or2hivGKgCLxOehHtGd3pYbWqpoqsM1l1IbphqM4oyqC0qrTuWIUiyBpEmFsYk8ImEeYWRk3Dv89uTkYp6eXkaub21ZPqc2N6Lre/tQWbs4m3bx1BdMDFX+m/4WAuN7y2kd9NjeG+ya2Pg9uT6vJC5RSVceUL66nRYNl9Ywlwv/AJXqQ+m1FVAQVHIS9dD+B56ZBrvy04Cg37WzcJ43tOVtJ/zOX6/SZhvKerqqkipySHzOLMulB+vPi4fr/4GCdLTtYFQQCTwUSwNbiudbw2lNe2km/4cQOng0+z/OByduXuwqiMXBJyCbOjZjMhdAIWY/t+y9ZTlFSWcKDgQF0Irw3kxZXFdceEuYUR61XfJSXGO4YQW0i79nUvrSptNkxnFGWQVZzV6MOZxWAh1C2UMLcwwtzCGq2H2EK65O+6vUYpkcDdy8k/4fbV0+pz9/HTLHjjJyqqalh8y3CG9rnwsWSrazRmPb+OwtJKvv7t2cMANtXT6vJC7ck6zTUv/Uikn40P7xyNi+XCRjKQ+myj6ko9dNcG8IaBvMUwHqkHcJ9+9eu9LIyDPrNjbQCvDeHHi4+TWaTfzy/Pb/Zxcd5xzO43m8v7Xo6Pi08Hl7pn0DSNpV8vxSvGS++Skr+PA/kHOHL6SN2HIFeTq9433B7EY71jifKMwmZpfvx/TdMoKC84K0zXhuyTpScbHe9mcasL0WFuYYS7hdcFa39X/y51Yev5kGEBhRAONyDYnU/uHsP8RZu48fWNvHTjUCbF+V/Qc328NYPdWad5/oZEGQawDfoHufPcDYn84u0t/ObDbSycl9QtL8ztdoxmPTj7NHPRanUlm1Z/xMhov8aBPGs77F7eehiv7arSg8O4xWghwiOCCI+IZvefqTyjh/GiTI6fOU7qvlR+Pv7nxHpLN7OLpZTC2+TNxPCJTAqfVLe9tKqUtPy0+r7h+fv4/PDnfLj/w7pjQmwhxHrFEu0VTWVNZaNQ3bDVHPRhDcPcwxgbMrZRuA5zC+vSXa06kwRuIUSrwrxd+fjuMdzyxk/84u0tPHP1IK4e2rYxjYvLq/jH6v0M7ePFFYNk+u22mtI/gEcv789fVu7hX1/u48FpcZ1dpN7NaKbUNRiiJ0L0ZY331baMN+ye0lIYd/YA3xj7Eg0+0fq6d1898PdQVrO1rosDwNoTazs/bFeW6XXeQ8fCdjG5MNBvIAP9BtZt0zSNEyUn2Je3r1EQX3tsLQZl0Lv+uIUy2G9wo0Ad6haKs+nCu7f1VhK4hRDn5Gtz4v3bR3HXO1v57UfbyT1Tzh3jz3+4uhe/TeNUcTmvL5Apty/UbeP6cvBkMQu/PUikr63NH3pEB2nYMn6uMJ57QL+A8+A3sO3d+uMMJvDqaw/iUY1DuYtMEX5RNE3/HZzYZV9S9dvcg+Dqo4//3n829B3foz/0gN4aHmgNJNAayISwCXXby6vLMSmTxex6OwAAHbVJREFUTMTTziRwCyHOi5uzmcW3DOc3H27nqVV7yS2u4OEZcecM0Bl5Jby+7hBzE0MYEubZQaXteZRSPDkngSO5JTz8yQ7CfVwZHuHd2cUSbdFaGC87bQ/gaXoIP7UfTh2AtC+huqL+OKufvSU8unEQ9wzvsa2zF6y8GHL26GO0n9hZH7LLT9cf49UXAuIhfq5e/zs+gq1vgrMnxF4OA2ZD5CQw954WXSdj154hubuSwC2EOG9OJiPPXZ+Ij9XCK9+nc6q4gqevHtjqrJRPf74Xo1I8OF36Z14ss9HASzcO5aoX13Pnf7fy6T1jCffpGWPU9nrO7hAyVF8aqq7Sp7vPbRLE966Aktz644xOepCvDeJ1oTwanC5+hKEuraZGr6MTuxqH67xDUDtaisVND9YDr4XABAhIAP/+Z9dNZan+jcPu5bB3JWx/Dyw2iJ6qh+/oqWBp2zCLQoAEbiFEGxkNij/NjsfX5sSzX+4nv6SChfOSmh09Y/PhPFamZvHrS6MJ8mi/sbx7Mw9XM4tuGc6VC9dz61ub+eSeMbg79+yvvns1o6m+VTxmWuN9JXl6+G4YxE/sgj0rGvcVdwuuD9+1LeK+MeAe0v0u2iw7DTm7G7dYn9gFFbUX9Sn9wtSABBh8gx6yA+LBs8/5vVazC8TN1JeqCjj0PexZpofvXZ+AyRmiLtW7ncRO1/vhC3EeJHALIdpMKcUvp0TjY7Pwh093cuPrG1l8y3A8XevHUK2p0Xjys90EeThzZxv6e4tz6+tr5eWbhjJ/0SbufTeZN24ZjqmVbxlED+XqDeEj9aWhqgrIP9Q4iJ86ADs+bNydwmwFt0C9ldfJTR9RxclNb22v21a7vck2Zw/91lH9nGtq9NdQG6ztM4dScKT+GCcPvbV6yDx7sB4I/nHt1wJtskD0pfoy899w9Ee95XvPZ/o3DAYzRE7UW75jZ4JVhjIULZPALYS4YDeO7IOP1cIv39/GtS9v4O3bRtS1ZH+SkklqZiH/72dDLnjsaNGy0f18+OtVCTy0JJUnV+zmyTkJnV0k0VWYLPXT3jekaVCc0ziInzmph/DyIj3M1q6XnW7cSt7iz3I+O7A3DOeNwnsz+53cMVUWwZEf7cHafhFjzm6o1KczRxnAJwpCkiBpvh6sA+LBI7TjWuiNJv1Cyr7jYcYzcGwz7FmuB/Dl94P6NUSM1Vu++8/SP8gI0YAEbiHERZmeEMSbt5q54+2tXP3ij7x920iCPJx55ou9DAnzZPbg4M4uYo/1s+HhHDx5hle/T6efn40FYyI6u0iiK1MK3AL0pe8lrR+raXp/5vIi+3K6yW1R43Bet61IHwWkvPC8g/s4gPX2O86eEDgQkm7Wu4UExINfHFi60LUKBkP9NwtT/wJZ2/RW793LYdXvYNWDEDZSb/nuP0u/oFX0ehK4hRAXbUw/X/53xyhueWMz1778I2OjfMkpKuelm4bKJC0O9tD0ONJPnuFPn+2ij48rE2MvbGIiIRpRSg+5Flc9oF8oTYOqsvowXlbYOJyXF5G2fw9RI2fo4do9uHv1K1cKghP1ZfIf4ORee7eT5bD69/oSnKi3fA+Y0/xESqJXkMAthGgXCSEeLLl7NDcv/okVO7KYPTj4oqaCF+fHaFD85/ohXPPyBu5/L4Ul94whJqCHj0ohug+l9AsRzS5ga/7D4LHStUTFTOzYcjmCUvrIJ/79YeJD+tjetd1Ovv6TvvjH21u+Z+vHdacPF+KiyFU2Qoh208fHysd3jeHOCZE8dkX/zi5Or2F1MrFowTCcLUZufXMzucXlnV0kIYRPPxj3ANzxLfx6J0z7m96nfe3T8NJoeGEYfPUnOJ6ifxMgejQJ3EKIduXn5sQjM/rj79Z7JoroCoI9XXj95mGcLCrnjv9upazyPC54E0J0DM8wGH0P3PoF/HYvzPyXPizj+v/AqxPhP4Ng9aOQ/p0++VFJnj5Si+gxpEuJEEL0EIPDPHn2uiHc+14yj3ySyrPXDT7nTKBCiA7mFgjDf6EvZ3Jh3yq968mmV2DDC/XHKYN+Eamrjz4EZO2ti3fjbQ3vu0g3vq5KArcQQvQgMwcFkX4yhn99uZ9+flbumxzd2UUSQrTE6qMPdZg0X7+g9NhmPYSX5EJpnn5bYr8tOKp3PynJheqKFp5QMdZkgx3+TUK5Vwsh3b7PKHHQ0aSGhRCih7lvchQHTxbzzzX76etrY+agoM4ukhDiXJw99Fksz0XToOJMk0CeV3c/50AqIV7O+v3TmfqkQSW5UFXa+s9u2FJu8we3IL013i2ofrH6STi/QFJrQgjRwyilePrqQWTkl/Lbj7YR6uXS2UUSQrQXpcDJpi/NjPF9QK0lZOLEsx9XUWIP5XkNWtCb3s+F4hP6BETFJ0Br0o9cGcDq3yCIB+pDOTa87xakh3eDXCbYkARuIYTogZzNRl6ZP5QrF67nF29v4YHBBkoqqnA2GWVsdCF6o9px1T1Cz+/4mmp9JtKiLCjKhtPH9dva+4XH4NhPekhvymA+O4Q3vXUP0mcd7SXXmUjgFkKIHsrX5sTiW4Yz98Uf+f26cn6/bjUATiYDLhYjLmZ9cTYb6+7Xrxv0+w2Oc7HY9zd3v+HzWQxYjAa5YFOI7sxgtIfjc0xTX1Wut4Y3DOMNw/nJffroK+WFZz/W7NpKMA+C0GFgcnLM6+tgEriFEKIHiwlwY+k9Y3jri42ERURSWllNaWU1ZRXV9vUaSiuqKbNvzztTUbdeWllNaUU15VVtH57MoDgrlLs5mxgc5smoSB9G9vXG09XigFcshOhQJie9a8u5prAvL7YH86z6MH46q/5+ZrK+XlVW/5j/OySB+3wopaYD/wGMwOuapj3dzDHXAU8AGrBd07R5jiyTEEL0NtEBblzax8zECRc2rXRNjUZZlR6+Syvt4byiplEoL2uwXn9MfXAvq6wmt7iC9zYd5Y31h1EK+ge6MyrSh9H9fBgR4Y2Hq7mdX7kQosuo7Xfe2vT2mqaP1lJkD+I9aJhDhwVupZQRWAhcBhwDNiullmuatrvBMdHAI8BYTdPylVLNz/sqhBCi0xgMCleLCVfLxf/LKK+qZntGIRvTc9mYnsu7m46weP0hlIIBQe6MjvRhVKQPw/t64+EiAVyIXkUpcPHUF/+eNVuxI1u4RwBpmqalAyil/gfMAXY3OOZ2YKGmafkAmqblOLA8QgghOpmTyciIvt6M6OvNL6dEU1ZZzfaMAjam57Eh/RRvbzzC6+sOYVAQH+zBqEjvugDu7iwBvKfRNI2Csho0TZM+/6JHc2TgDgEyGtw/BoxsckwMgFJqPXq3kyc0TfvCgWUSQgjRhTibjYyM9GFkpA+/Qg/g2zIK2HBQbwF/68cjvPaDHsATQjwYFenDqEhvhkd44yYBvFvSNI2UjAJW7sji89QsjheW8aefvmRImCdDwrxIDPdkcJinfMMhehSlaZpjnlipa4Dpmqb9wn5/PjBS07T7GhyzAqgErgNCge+BgZqmFTR5rjuAOwD8/PyGfvjhhw4pc29UXFyMzWbr7GL0GFKf7Ufqsn111/qsqNY4WFDD3rxq9uZVc7CghioNFBDhYSDO20ict4EYLyMupo5rIe2u9dlZNE3jYGENm7Or2JxdTV6ZhknBQD8jEdYqcivNpBdUk1msUZtKgq2KSE8j/TwM9PM0EGIzYJQhLc9Jzs32NWnSpK2apg272OdxZAt3JhDW4H6ofVtDx4BNmqZVAoeUUvuBaGBzw4M0TXsVeBUgNjZWm9jcgO7igqxduxapz/Yj9dl+pC7bV0+pz7LKapKP5Nv7gOfx1dF8Pj+kYTQoewu4N6MjfRgW4Y3NyXH/4npKfTpSbUv2qh1ZfL4zm8yCMixGA+Nj/Jg5KIgp/QNwdzY3qsuiskp2HCsk5Wg+2zIKSDlawLpMfRpzV4uRgSEeJIbrreCJYZ74uzt34ivsmuTc7JocGbg3A9FKqb7oQft6oOkIJJ8CNwBvKKV80buYpDuwTEIIIboxZ7ORMVG+jInyBaC0oprko3oA33Awl8XrDvHKd+kYDYqBIR6M7qdfhDmsjxdWBwZwodM0jW0ZBaxKzWJVajaZBaX2kO3Lb6fGcOmAgFb74rs5mxkb5ctY++9X0zQy8kpJycgn5WgBKRkFLFqXTmW13g4e4unCEHv4Tgz3Ij7YHWezsUNeqxBt4bB3H03TqpRS9wGr0ftnL9Y0bZdS6klgi6Zpy+37piqldgPVwIOapjUzZZEQQghxNheLsVFAK6moIvlIARvST7ExPY/Xvk/npbUHMRkUA0M9GG2/ADMpzEuGIWwnLYXsS6LPL2S3RilFuI8r4T6uzBkSAujfcuw6ftreAq4H8ZU7sgAwGxUDgtxJDPdiSJgnieGehHu7dpkLMjVNo6RCH+8+90wFeWfKyS2uIO+MvpRUVBPg7kSQhwtBns4Ee7gQ6OEsHyJ6AId+3Nc0bRWwqsm2PzZY14Df2BchhBDiorhaTIyL9mVcdH0A33okv+4izFe/T+fFtQcBiPK3MTTci6Q+ngzt40Wkr02mvT9Pmqax/VghK3ccrwvZZqNifLQfv7lMD9mOuujR2WxkaB8vhvbxAvoCkFNUxjZ7C3jK0Xw+3JLBmz8eBsDbaiExzNMewL0YFObRbiPeaJrG6bIqe2CuD8+5Z+pDdG2wzivW11uaSMpi0md3LSytPGufj9VCkKczQR4uBHs4E+zpQpCnvh7k6UKAmxMmo6FdXpNwDPl+TQghRI/lajFxSbQfl0T7AXCmvIrtGQUkH81n65F8vtiVzQdb9AG13J1NJIbrQS4p3Ish4Z4O7Qfe3dSG7FWpWazckVUXsi/pgJB9Lv5uzkyND2RqvD4NeVV1DQdyivVuKEfzScko4Ou9+sjDSkG0v60ugCeGexLt74bRoKip0SgorTxneK7dl19SUde9pSlXixFvqwUfqwU/mxOxAe742Cx4Wy112/VbJ7xtFqwWI0opyiqrySosI6uglOMNbwtLOZpbwsb0XIrKqhr9LIPS6yDI0xlTRRnrincT5OlCiD2kB3k642t1kg+UnUjeSYQQQvQaVidToz7gNTUa6afOkHw0nxR7CP/3VyfRND3ExAS41QXwoX286OPTdbondARN09hxrJCVzYTsBy6L4bJODNmtMRkN9A9yp3+QO/NG6lOOF5ZWsuNYQV0I/3L3CT7ccgwAq8WIs9lIfkkFNS0M3ubmbKoLyaFergwO9cTbVh+cG4ZnH6vlgruBOJuN9PW10tfX2uIxRWWVZBWWcbygtHE4Lyzl4KkaUjcdoayycUu62agI9KhvJa9rIbcH8hBPFzxczL3q/O5IEriFEEL0WgaDIsrfRpS/jeuG6QNrFZZWsi2jgOQj+SQfzWf5tuO8u+kooHdPSAr3xLO6AufwXAaHeuJi6Vn9a2tD9qrULFamZnEsXw/Z46J89ZDdP6Bb9n/3cDE3+rZD0zSO5JaQkpHP9oxCKqprzg7PVgs+NgterhYspq7TZcPN2Yybs5mYALez9q1du5YJEyaQX1JZH8gLSzleoN9mFZSx5Ug+2TuyqGry6cLFbCTI05nhfby5MjGEkX29pVW8nUjgFkIIIRrwcDEzIcaPCTF6MKuu0UjLKWarPYAnH8kn/VQlH+/fiMmg6B/kTlK4J0n2lvBQL5du10rYXMg2GRSXRPvyqynRTB0Q2C1DdmuUUkT4WonwtXJVYmhnF6ddKaXqPjgkhHg0e0x1jcap4vK6UF57m5FXwoodx/lgSwbBHs7MSQxhbmII0c2Ee3H+JHALIYQQrTAaFLGBbsQGutV1T1ix5ltcwwfoIfxIAR9uOcZbG44A4O/mVNcFJamPJ/HBHl1ylAlN00jNLGTljsYhe1wPDtmintGgCHB3JsDdmcQm+0oqqvhy9wk+TcnkVftIP/HB7lyVGMLswcEy/vkFkMAthBBCtJHNopgYF8DkuABAv0hvb3ZRXQt48tECvtiVDYDFaCA+xL0+hId7EejROLDU1GiUV9VQVllNeVUN5VXVlFXqt3XbK2soq9Jvz3lsVY39uKaPqz+2tFJfrw3Zv5wSzdQBAXi6Wjq8PkXX4moxMWdICHOGhHCyqJwVO46zNCWTv6zcw1Or9jA2yperEkOYFh8o49ufJ6klIYQQ4iKZjAYSQjxICPHg5tERAJwsKm8QwPN5Z+MRFq07BICvTQ+1tWG4pZEuzpfFaMDJZMDJbLTfGnA2GXEy69u9rRacTAaca/eb9NuYQDcJ2aJVfm5O/HxsX34+ti9pOcUs25bJ0pRMfvPhdlzMO5kWH8BVSaGM7ecjQxO2QgK3EEII4QB+bk5Miw9kmn2ouoqqGnZnnSb5SD77soswGtVZIdjZ3Pi2br89ODd3rJPJIBe2iQ4R5W/jt1NjeeDSGLYezWdpSiYrth/n023H8bU5MXtwMHOTQogPdu921zE4mgRuIYQQogNYTAaG2CdgEaI7MxgUwyO8GR7hzeOzBvDt3pN8mpLJOxuPsHj9IaL8bVyVGMKcIcGEerl2dnG7BAncQgghhBDigjiZjExPCGR6QiAFJRWsSs1macox/rF6H/9YvY8Rfb25KjGEywcGdckx2zuKBG4hhBBCCHHRPF0tzBsZzryR4WTklbBsWyafpGTyyCepPL5sF1P6+3NVYggTY/271LjmHUECtxBCCCGEaFdh3q7cNzmaeydFkZpZyCfJmXy2/Tif78zG09XMzIFBzE0KISncq1f095bALYQQQgghHEIpxaBQTwaFevLozP6sSzvF0uRMliQf491NRwn3duXKIcFcmRhCpJ+ts4vrMBK4hRBCCCGEw5mNBibF+jMp1p/i8ipW78xmaUomz3+bxnPfpDE4zJO5iSFcMSgIH5tTZxe3XUngFkIIIYQQHcrmZOLqoaFcPTSU7MIyPtt+nE9SMnl8+S6eXLGbCTF+/G3uQAJ6yKyWEriFEEIIIUSnCfRw5vbxkdw+PpK92af5NOU43+0/iVcPmpBJArcQQgghhOgS4gLdeXiGOw/PiOvsorSr3jUmixBCCCGEEB1MArcQQgghhBAOJIFbCCGEEEIIB5LALYQQQgghhANJ4BZCCCGEEMKBJHALIYQQQgjhQBK4hRBCCCGEcCAJ3EIIIYQQQjiQBG4hhBBCCCEcSAK3EEIIIYQQDiSBWwghhBBCCAeSwC2EEEIIIYQDSeAWQgghhBDCgSRwCyGEEEII4UASuIUQQgghhHAgCdxCCCGEEEI4kARuIYQQQgghHEgCtxBCCCGEEA6kNE3r7DK0iVKqCNjX2eXoQXyBU51diB5E6rP9SF22L6nP9iX12X6kLtuX1Gf7itU0ze1in8TUHiXpYPs0TRvW2YXoKZRSW6Q+24/UZ/uRumxfUp/tS+qz/Uhdti+pz/allNrSHs8jXUqEEEIIIYRwIAncQgghhBBCOFB3DNyvdnYBehipz/Yl9dl+pC7bl9Rn+5L6bD9Sl+1L6rN9tUt9druLJoUQQgghhOhOumMLtxBCCCGEEN1Glw3cSqnpSql9Sqk0pdTDzex3Ukp9YN+/SSkV0fGl7B6UUmFKqW+VUruVUruUUr9q5piJSqlCpdQ2+/LHzihrd6CUOqyUSrXX01lXLyvdc/Zzc4dSKqkzytkdKKViG5xz25RSp5VSv25yjJybrVBKLVZK5SildjbY5q2U+lIpdcB+69XCYxfYjzmglFrQcaXuulqoz38opfba/56XKqU8W3hsq+8NvU0LdfmEUiqzwd/z5S08ttUM0Bu1UJ8fNKjLw0qpbS08Vs7NBlrKRQ5979Q0rcstgBE4CEQCFmA7MKDJMfcAL9vXrwc+6Oxyd9UFCAKS7OtuwP5m6nMisKKzy9odFuAw4NvK/suBzwEFjAI2dXaZu8Ni/7vPBvo02S7nZuv1Nh5IAnY22PYM8LB9/WHg7808zhtIt9962de9Ovv1dPbSQn1OBUz29b83V5/2fa2+N/S2pYW6fAL43Tked84M0BuX5uqzyf5/AX9sYZ+cm43ro9lc5Mj3zq7awj0CSNM0LV3TtArgf8CcJsfMAd6yr38MTFFKqQ4sY7ehaVqWpmnJ9vUiYA8Q0rml6tHmAG9ruo2Ap1IqqLML1Q1MAQ5qmnakswvSnWia9j2Q12Rzw/fHt4Arm3noNOBLTdPyNE3LB74EpjusoN1Ec/WpadoaTdOq7Hc3AqEdXrBuqIVz83ycTwbodVqrT3v+uQ54v0ML1U21kosc9t7ZVQN3CJDR4P4xzg6IdcfY3wgLAZ8OKV03Zu96kwhsamb3aKXUdqXU50qp+A4tWPeiAWuUUluVUnc0s/98zl9xtutp+Z+FnJttE6BpWpZ9PRsIaOYYOU8vzK3o32A151zvDUJ3n717zuIWvrKXc7PtLgFOaJp2oIX9cm62oEkucth7Z1cN3MIBlFI2YAnwa03TTjfZnYz+Vf5g4Hng044uXzcyTtO0JGAGcK9SanxnF6i7U0pZgNnAR83slnPzImj6d6AyHFU7UEo9ClQB77ZwiLw3nNtLQD9gCJCF3g1CXLwbaL11W87NZrSWi9r7vbOrBu5MIKzB/VD7tmaPUUqZAA8gt0NK1w0ppczoJ9W7mqZ90nS/pmmnNU0rtq+vAsxKKd8OLma3oGlapv02B1iK/vVnQ+dz/orGZgDJmqadaLpDzs0LcqK2G5P9NqeZY+Q8bQOl1C3AFcCN9n/EZzmP94ZeT9O0E5qmVWuaVgO8RvN1JOdmG9gz0Fzgg5aOkXPzbC3kIoe9d3bVwL0ZiFZK9bW3fF0PLG9yzHKg9srQa4BvWnoT7O3sfbsWAXs0TXu2hWMCa/vAK6VGoJ8b8gGmCaWUVSnlVruOfjHVziaHLQduVrpRQGGDr6hE81psnZFz84I0fH9cACxr5pjVwFSllJf9a/2p9m2iCaXUdOD/gNmappW0cMz5vDf0ek2uZ7mK5uvofDKAqHcpsFfTtGPN7ZRz82yt5CLHvXd29pWirVxBejn6VaMHgUft255Ef8MDcEb/+jkN+AmI7Owyd9UFGIf+tcgOYJt9uRy4C7jLfsx9wC70q8E3AmM6u9xdcUG/an67fdnV4NxsWJcKWGg/d1OBYZ1d7q68AFb0AO3RYJucm+dff++jfzVfid6X8Db061m+Bg4AXwHe9mOHAa83eOyt9vfQNODnnf1ausLSQn2moffZrH3/rB0hKxhYZV9v9r2hNy8t1OV/7e+LO9DDTVDTurTfPysD9Palufq0b3+z9v2ywbFybrZely3lIoe9d8pMk0IIIYQQQjhQV+1SIoQQQgghRI8ggVsIIYQQQggHksAthBBCCCGEA0ngFkIIIYQQwoEkcAshhBBCCOFAEriFEKIXU0pNVEqt6OxyCCFETyaBWwghhBBCCAeSwC2EEN2AUuompdRPSqltSqlXlFJGpVSxUurfSqldSqmvlVJ+9mOHKKU2KqV2KKWW2mdDQykVpZT6Sim1XSmVrJTqZ396m1LqY6XUXqXUuw1m9nxaKbXb/jz/7KSXLoQQ3Z4EbiGE6OKUUv2BnwFjNU0bAlQDN6LP0rlF07R44DvgcftD3gYe0jRtEPqsfrXb3wUWapo2GBiDPmsdQCLwa2AA+qx0Y5VSPuhTb8fbn+cvjn2VQgjRc0ngFkKIrm8KMBTYrJTaZr8fCdQAH9iPeQcYp5TyADw1TfvOvv0tYLxSyg0I0TRtKYCmaWWappXYj/lJ07RjmqbVoE9xHAEUAmXAIqXUXKD2WCGEEG0kgVsIIbo+BbyladoQ+xKradoTzRynXeDzlzdYrwZMmqZVASOAj4ErgC8u8LmFEKLXk8AthBBd39fANUopfwCllLdSqg/6e/g19mPmAes0TSsE8pVSl9i3zwe+0zStCDimlLrS/hxOSinXln6gUsoGeGiatgp4ABjsiBcmhBC9gamzCyCEEKJ1mqbtVko9BqxRShmASuBe4Awwwr4vB72fN8AC4GV7oE4Hfm7fPh94RSn1pP05rm3lx7oBy5RSzugt7L9p55clhBC9htK0C/0GUgghRGdSShVrmmbr7HIIIYRonXQpEUIIIYQQwoGkhVsIIYQQQggHkhZuIYQQQgghHEgCtxBCCCGEEA4kgVsIIYQQQggHksAthBBCCCGEA0ngFkIIIYQQwoEkcAshhBBCCOFA/x9SHOPFAieARQAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 864x576 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "text": [
            "Classification Report of Proposed Model\n",
            "                   precision    recall  f1-score   support\n",
            "\n",
            "Oligodendroglioma       0.94      0.97      0.95     10357\n",
            "     Glioblastoma       0.95      0.94      0.94      8251\n",
            "      Astrocytoma       0.94      0.90      0.92      3779\n",
            "     Unidentified       0.97      0.95      0.96      4249\n",
            "\n",
            "         accuracy                           0.95     26636\n",
            "        macro avg       0.95      0.94      0.94     26636\n",
            "     weighted avg       0.95      0.95      0.95     26636\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 513
        },
        "id": "1GLtVE_2dsIP",
        "outputId": "d4e13294-10e4-49e4-e0d9-12382dbea5ce"
      },
      "source": [
        "fig = plt.figure(figsize=(12,8))\n",
        "plt.plot(range(epochs),history.history['val_accuracy'],label='Adaptive_Momentum')\n",
        "plt.plot(range(epochs),history2.history['val_accuracy'],label='SGD')\n",
        "plt.plot(range(epochs),history4.history['val_accuracy'],label='RMSprop')\n",
        "plt.plot(range(epochs),history5.history['val_accuracy'],label='Adam')\n",
        "\n",
        "plt.legend(loc=0)\n",
        "plt.xlabel('epochs')\n",
        "plt.xlim([0,epochs])\n",
        "plt.ylabel('accuracy on test set')\n",
        "plt.grid(True)\n",
        "plt.title(\"Comparing Model Accuracy\")\n",
        "plt.show()\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 864x576 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "c9jHA7ErbEJX",
        "outputId": "33283400-ef4b-4c84-ac00-8ef7f9afb9ec"
      },
      "source": [
        "m = [model,model2,model3,model4,model5,model6]\n",
        "\n",
        "for i in m:\n",
        "  predictions = i.predict_classes(x_test)\n",
        "  print('Classification Report of Proposed Model')\n",
        "  print(classification_report(y_test,predictions,target_names=target_names))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Classification Report of Proposed Model\n",
            "                   precision    recall  f1-score   support\n",
            "\n",
            "Oligodendroglioma       0.94      0.97      0.95     10357\n",
            "     Glioblastoma       0.95      0.94      0.94      8251\n",
            "      Astrocytoma       0.94      0.90      0.92      3779\n",
            "     Unidentified       0.97      0.95      0.96      4249\n",
            "\n",
            "         accuracy                           0.95     26636\n",
            "        macro avg       0.95      0.94      0.94     26636\n",
            "     weighted avg       0.95      0.95      0.95     26636\n",
            "\n",
            "Classification Report of Proposed Model\n",
            "                   precision    recall  f1-score   support\n",
            "\n",
            "Oligodendroglioma       0.95      0.97      0.96     10357\n",
            "     Glioblastoma       0.94      0.95      0.95      8251\n",
            "      Astrocytoma       0.95      0.90      0.93      3779\n",
            "     Unidentified       0.98      0.96      0.97      4249\n",
            "\n",
            "         accuracy                           0.95     26636\n",
            "        macro avg       0.96      0.94      0.95     26636\n",
            "     weighted avg       0.95      0.95      0.95     26636\n",
            "\n",
            "Classification Report of Proposed Model\n",
            "                   precision    recall  f1-score   support\n",
            "\n",
            "Oligodendroglioma       0.91      0.94      0.93     10357\n",
            "     Glioblastoma       0.92      0.92      0.92      8251\n",
            "      Astrocytoma       0.92      0.86      0.89      3779\n",
            "     Unidentified       0.94      0.93      0.94      4249\n",
            "\n",
            "         accuracy                           0.92     26636\n",
            "        macro avg       0.92      0.91      0.92     26636\n",
            "     weighted avg       0.92      0.92      0.92     26636\n",
            "\n",
            "Classification Report of Proposed Model\n",
            "                   precision    recall  f1-score   support\n",
            "\n",
            "Oligodendroglioma       0.84      0.84      0.84     10357\n",
            "     Glioblastoma       0.78      0.88      0.83      8251\n",
            "      Astrocytoma       0.84      0.71      0.77      3779\n",
            "     Unidentified       0.91      0.84      0.87      4249\n",
            "\n",
            "         accuracy                           0.83     26636\n",
            "        macro avg       0.84      0.82      0.83     26636\n",
            "     weighted avg       0.83      0.83      0.83     26636\n",
            "\n",
            "Classification Report of Proposed Model\n",
            "                   precision    recall  f1-score   support\n",
            "\n",
            "Oligodendroglioma       0.85      0.92      0.89     10357\n",
            "     Glioblastoma       0.87      0.87      0.87      8251\n",
            "      Astrocytoma       0.89      0.79      0.84      3779\n",
            "     Unidentified       0.96      0.86      0.91      4249\n",
            "\n",
            "         accuracy                           0.88     26636\n",
            "        macro avg       0.89      0.86      0.88     26636\n",
            "     weighted avg       0.88      0.88      0.88     26636\n",
            "\n",
            "Classification Report of Proposed Model\n",
            "                   precision    recall  f1-score   support\n",
            "\n",
            "Oligodendroglioma       0.95      0.96      0.96     10357\n",
            "     Glioblastoma       0.95      0.95      0.95      8251\n",
            "      Astrocytoma       0.95      0.90      0.93      3779\n",
            "     Unidentified       0.96      0.96      0.96      4249\n",
            "\n",
            "         accuracy                           0.95     26636\n",
            "        macro avg       0.95      0.94      0.95     26636\n",
            "     weighted avg       0.95      0.95      0.95     26636\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    }
  ]
}