Bayesian_neural_network.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Bayesian-neural-network.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "toc_visible": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9vw14nzqO6dW"
      },
      "source": [
        "# First glance at Bayesian neural network\n",
        "In this notebook we compare the non-Bayesian and Bayesian neural network which share similar architecture. Bayesian NN provides uncertainty in its prediction while non-Bayesian NN only provides point esitmates. MNIST dataset is used."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "p0dEROjC9tes"
      },
      "source": [
        "## 0.0 Import libraries and set plot format"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "iKcc7Dd0WP7R"
      },
      "source": [
        "import numpy as np\n",
        "import tensorflow as tf\n",
        "from tensorflow import keras\n",
        "from tensorflow.keras import layers\n",
        "import tensorflow_datasets as tfds\n",
        "import tensorflow_probability as tfp\n",
        "import pandas as pd\n",
        "from sklearn.model_selection import train_test_split\n",
        "import matplotlib.pyplot as plt\n",
        "import seaborn as sns\n",
        "import time\n",
        "import sklearn\n",
        "from sklearn import datasets\n",
        "import scipy.stats as ss\n",
        "from tqdm import tqdm\n",
        "\n",
        "tfd = tfp.distributions"
      ],
      "execution_count": 36,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "jpVlnjANWTZN",
        "outputId": "4837df3a-1901-4220-95f9-4f0aca06e92f"
      },
      "source": [
        "print(tf.__version__)\n",
        "print(tfp.__version__)"
      ],
      "execution_count": 37,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "2.4.1\n",
            "0.12.1\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MAmKZ_Yv0v9l"
      },
      "source": [
        "plt.style.use('ggplot')\n",
        "\n",
        "SMALL_SIZE = 12\n",
        "MEDIUM_SIZE = 15\n",
        "BIGGER_SIZE = 18\n",
        "\n",
        "plt.rc('font', size=SMALL_SIZE)          # controls default text sizes\n",
        "plt.rc('axes', titlesize=SMALL_SIZE)     # fontsize of the axes title\n",
        "plt.rc('axes', labelsize=MEDIUM_SIZE)    # fontsize of the x and y labels\n",
        "plt.rc('xtick', labelsize=SMALL_SIZE)    # fontsize of the tick labels\n",
        "plt.rc('ytick', labelsize=SMALL_SIZE)    # fontsize of the tick labels\n",
        "plt.rc('legend', fontsize=SMALL_SIZE)    # legend fontsize\n",
        "plt.rc('figure', titlesize=BIGGER_SIZE)  # fontsize of the figure title"
      ],
      "execution_count": 38,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "izn9ZKLOrjhG"
      },
      "source": [
        "## 1.0 Load datasets"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "JxAArbk2Wb3e",
        "outputId": "4c5b8196-b03b-4609-b2fa-d71d28527fab"
      },
      "source": [
        "# the data, split between train and test sets\n",
        "(X_train, y_train), (X_test, y_test) = keras.datasets.mnist.load_data()\n",
        "X_train, X_test = X_train.astype(np.float32)/255., X_test.astype(np.float32)/255.\n",
        "\n",
        "X_train = np.expand_dims(X_train,-1)\n",
        "X_test = np.expand_dims(X_test,-1)\n",
        "print(f\"Training input data size is {X_train.shape}\")\n",
        "print(f\"Training output data size is {y_train.shape}\")"
      ],
      "execution_count": 39,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Training input data size is (60000, 28, 28, 1)\n",
            "Training output data size is (60000,)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "sNhPczF7fhHD",
        "outputId": "eac74ca6-74d5-42e2-e005-bc9a0801c7da"
      },
      "source": [
        "X_train.dtype"
      ],
      "execution_count": 40,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "dtype('float32')"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 40
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "rR3C_CvCphV2",
        "outputId": "f235b867-cc87-4ea6-93fa-3001e0a3c736"
      },
      "source": [
        "input_shape = X_train[0].shape\n",
        "print(f'input shape is: {input_shape}')\n",
        "print(f'outputs can be:  {set(y_train)}')"
      ],
      "execution_count": 41,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "input shape is: (28, 28, 1)\n",
            "outputs can be:  {0, 1, 2, 3, 4, 5, 6, 7, 8, 9}\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "eaEbA1qsRt5D"
      },
      "source": [
        "## 2.0 Non-Bayesian neural network model"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "vtYQWWtP5B2d"
      },
      "source": [
        "### 2.1 Build non-Bayesian model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bS7SXRfrHZbt"
      },
      "source": [
        "mymodel = keras.models.Sequential([\n",
        "                                   layers.Conv2D(kernel_size=(3,3),filters=32,activation='relu',input_shape=input_shape),\n",
        "                                   layers.MaxPooling2D(pool_size=(2,2),strides=None, padding='valid'),\n",
        "                                   layers.Flatten(),\n",
        "                                   layers.Dense(100,activation='relu'),\n",
        "                                   layers.Dropout(0.2),\n",
        "                                   layers.Dense(10, activation='softmax'),\n",
        "            \n",
        "])"
      ],
      "execution_count": 42,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hrmT-wUO-Bw6"
      },
      "source": [
        "def run_experiment(model,loss, X_train,y_train,X_test,y_test, epochs, patience):\n",
        "  '''compile, fit, evaluate in one function''' \n",
        "  early_stop = tf.keras.callbacks.EarlyStopping(monitor='val_loss',mode = 'min', patience=patience)\n",
        "  model.compile(\n",
        "        optimizer=keras.optimizers.Adam(learning_rate=1e-3),\n",
        "        loss=loss,\n",
        "        metrics=['accuracy'],\n",
        "    )\n",
        "  print(\"training the model...\")\n",
        "  history = model.fit(X_train,y_train,validation_split=0.1,epochs=epochs,verbose=1,callbacks=[early_stop])\n",
        "  print(\"evaluating model on train data\")\n",
        "  loss_t, acc_t = model.evaluate(X_train, y_train)\n",
        "  print(f\"Trainning loss: {loss_t :.3f}, training acc: {acc_t:.3f}\")\n",
        "\n",
        "  print(\"evaluating model on test data\")\n",
        "  loss_tt, acc_tt = model.evaluate(X_test, y_test)\n",
        "  print(f\"Testing loss: {loss_tt:.3f}, testing acc: {acc_tt:.3f}\")\n",
        "  return history"
      ],
      "execution_count": 43,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "MmcGUVaR5Gff"
      },
      "source": [
        "### 2.2 Run non-Bayesian model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Px846tFgEmqz",
        "outputId": "302a07d6-1159-444c-9ab5-9d92e6aeecb4"
      },
      "source": [
        "loss='sparse_categorical_crossentropy'\n",
        "start= time.time()\n",
        "history = run_experiment(mymodel,loss, X_train,y_train,X_test,y_test,200,10)\n",
        "time_cost = time.time()- start\n",
        "print(f'Training cost {time_cost :.2f} seconds.')"
      ],
      "execution_count": 44,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "training the model...\n",
            "Epoch 1/200\n",
            "1688/1688 [==============================] - 5s 3ms/step - loss: 0.3881 - accuracy: 0.8815 - val_loss: 0.0619 - val_accuracy: 0.9822\n",
            "Epoch 2/200\n",
            "1688/1688 [==============================] - 5s 3ms/step - loss: 0.0724 - accuracy: 0.9792 - val_loss: 0.0462 - val_accuracy: 0.9875\n",
            "Epoch 3/200\n",
            "1688/1688 [==============================] - 4s 3ms/step - loss: 0.0511 - accuracy: 0.9843 - val_loss: 0.0456 - val_accuracy: 0.9877\n",
            "Epoch 4/200\n",
            "1688/1688 [==============================] - 4s 3ms/step - loss: 0.0410 - accuracy: 0.9870 - val_loss: 0.0476 - val_accuracy: 0.9883\n",
            "Epoch 5/200\n",
            "1688/1688 [==============================] - 4s 3ms/step - loss: 0.0290 - accuracy: 0.9900 - val_loss: 0.0531 - val_accuracy: 0.9872\n",
            "Epoch 6/200\n",
            "1688/1688 [==============================] - 5s 3ms/step - loss: 0.0217 - accuracy: 0.9932 - val_loss: 0.0445 - val_accuracy: 0.9898\n",
            "Epoch 7/200\n",
            "1688/1688 [==============================] - 4s 3ms/step - loss: 0.0175 - accuracy: 0.9942 - val_loss: 0.0534 - val_accuracy: 0.9863\n",
            "Epoch 8/200\n",
            "1688/1688 [==============================] - 5s 3ms/step - loss: 0.0178 - accuracy: 0.9942 - val_loss: 0.0474 - val_accuracy: 0.9888\n",
            "Epoch 9/200\n",
            "1688/1688 [==============================] - 5s 3ms/step - loss: 0.0142 - accuracy: 0.9953 - val_loss: 0.0603 - val_accuracy: 0.9900\n",
            "Epoch 10/200\n",
            "1688/1688 [==============================] - 4s 3ms/step - loss: 0.0117 - accuracy: 0.9964 - val_loss: 0.0458 - val_accuracy: 0.9900\n",
            "Epoch 11/200\n",
            "1688/1688 [==============================] - 4s 3ms/step - loss: 0.0101 - accuracy: 0.9963 - val_loss: 0.0512 - val_accuracy: 0.9892\n",
            "Epoch 12/200\n",
            "1688/1688 [==============================] - 4s 3ms/step - loss: 0.0090 - accuracy: 0.9969 - val_loss: 0.0479 - val_accuracy: 0.9907\n",
            "Epoch 13/200\n",
            "1688/1688 [==============================] - 4s 3ms/step - loss: 0.0088 - accuracy: 0.9971 - val_loss: 0.0662 - val_accuracy: 0.9880\n",
            "Epoch 14/200\n",
            "1688/1688 [==============================] - 4s 3ms/step - loss: 0.0080 - accuracy: 0.9974 - val_loss: 0.0538 - val_accuracy: 0.9900\n",
            "Epoch 15/200\n",
            "1688/1688 [==============================] - 5s 3ms/step - loss: 0.0064 - accuracy: 0.9979 - val_loss: 0.0572 - val_accuracy: 0.9888\n",
            "Epoch 16/200\n",
            "1688/1688 [==============================] - 5s 3ms/step - loss: 0.0065 - accuracy: 0.9975 - val_loss: 0.0497 - val_accuracy: 0.9908\n",
            "evaluating model on train data\n",
            "1875/1875 [==============================] - 4s 2ms/step - loss: 0.0058 - accuracy: 0.9989\n",
            "Trainning loss: 0.006, training acc: 0.999\n",
            "evaluating model on test data\n",
            "313/313 [==============================] - 1s 2ms/step - loss: 0.0450 - accuracy: 0.9895\n",
            "Testing loss: 0.045, testing acc: 0.989\n",
            "Training cost 77.14 seconds.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "4_Gu4ZpR4oU7"
      },
      "source": [
        ""
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "_w0qxMJvT1H1",
        "outputId": "4f2e7a2e-2968-4139-b15a-51d5aaf07fe9"
      },
      "source": [
        "mymodel.summary()"
      ],
      "execution_count": 45,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential_1\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "conv2d_1 (Conv2D)            (None, 26, 26, 32)        320       \n",
            "_________________________________________________________________\n",
            "max_pooling2d_1 (MaxPooling2 (None, 13, 13, 32)        0         \n",
            "_________________________________________________________________\n",
            "flatten_1 (Flatten)          (None, 5408)              0         \n",
            "_________________________________________________________________\n",
            "dense_2 (Dense)              (None, 100)               540900    \n",
            "_________________________________________________________________\n",
            "dropout_1 (Dropout)          (None, 100)               0         \n",
            "_________________________________________________________________\n",
            "dense_3 (Dense)              (None, 10)                1010      \n",
            "=================================================================\n",
            "Total params: 542,230\n",
            "Trainable params: 542,230\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "AD_WSjd4XtXh"
      },
      "source": [
        "mymodel.save('Normal_CNN.h5')"
      ],
      "execution_count": 46,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 393
        },
        "id": "fRVUe0ajI01F",
        "outputId": "b7c24894-f286-42bf-ce1e-a9c90ff50027"
      },
      "source": [
        "pd.DataFrame(history.history)[['loss','val_loss']].plot(figsize=(10,6))\n",
        "plt.title(\"loss change of Non-Bayesian NN\")\n",
        "plt.show()"
      ],
      "execution_count": 47,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAmUAAAF4CAYAAAAVL47/AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nOzdeXhTVf4G8PcmaZuGpC00dEnapiwiiyACtSBQwKk4OigiCgyLK3XcRmZwBjeGVRZRUdSCM/AbVFxGHEQUl5HasaBCYURka9lb2qZ035s0TXJ+fxRSQoFuaZO07+d5fJ7k3pvcc78m6cu5554rCSEEiIiIiMitZO5uABERERExlBERERF5BIYyIiIiIg/AUEZERETkARjKiIiIiDwAQxkRERGRB2AoI/JQ0dHRSEpKcmsbMjIyIEkSrFarW9vRWuvWrUNoaCjUajWKiorc3RyPc9ttt+Hdd991dzOIOj2GMiLq0GprazF37lx8++23qKysRHBwsNP6C8Hz9ttvd1o+c+ZMLFq0yOXteeeddyCXy6FWq6FWq9GzZ0+sW7fO5ftpjq+//hr333+/y9/3nXfegSRJWLVqldPyiIgIfP/99wCARYsWQZIkbN682bHearVCkiRkZGS4vE1EnoyhjIg6tLy8PJjNZgwYMOCq26WmpuKnn35qlzaNGDEClZWVqKysxJYtWzBv3jz88ssv7bLv9tatWzesWrUKFRUVV91m4cKFsNls7dgyIs/DUEbkBWpqavCnP/0JOp0OOp0Of/rTn1BTUwMAKCwsxIQJExAUFIRu3bph9OjRsNvtAICXXnoJer0eGo0G1157Lb777rvLvr/JZMLTTz8Ng8GAwMBAjBo1CiaTybH+gw8+QFRUFLRaLZYtW+ZYvnfvXowYMQJBQUEIDw/Hk08+CYvF4lgvSRLefvttXHPNNQgKCsITTzyBCzcRsdlsePrpp6HVatGjRw+89dZbTqdKy8rK8PDDDyM8PBx6vR7z58+/4h/tK9Xn+PHjuPbaawEAQUFBuPnmm69Y43nz5uGFF1644vr169ejd+/e6NatG+68804YjcYmHWdjbrjhBvTr1w9paWmOZffeey/CwsIQGBiIuLg4HDlyBACwb98+hIaGOtXh008/xfXXXw8AsNvtWLlyJXr16oXg4GBMmTIFxcXFAACz2YyZM2ciODgYQUFBiImJQV5eHgBg7Nix2LBhAwDg1KlTuPnmmxEcHAytVosZM2agtLTUsb/o6Gi88sorGDRoEAIDAzF16lSYzeYrHl+/fv0wYsQIrF69+orb/Pa3v4Wvry/ef//9JtWMqKNiKCPyAsuWLcOePXtw4MAB/Prrr9i7dy9efPFFAMCrr76KiIgIFBQUIC8vD8uXL4ckSTh27Bjeeust7Nu3DxUVFfjPf/6D6Ojoy77/X/7yF/z888/46aefUFxcjFWrVkEmq/95+OGHH3Ds2DF89913WLJkiSNAyOVyvPbaaygsLMTu3bvx3XffYe3atU7vvX37duzbtw8HDx7E5s2b8Z///AdAXcj5+uuvceDAAezfvx+fffaZ0+seeOABKBQKnDx5Er/88gu+/fZbR3Boan369OnjCDSlpaVITk6+Yo0ff/xxHD9+/LLj+JKTk/Hcc89h8+bNyM3NhcFgwLRp05p0nI3Zt28fjh8/jmHDhjmW3XbbbThx4gTy8/MxZMgQzJgxAwAQExOD4OBgfPvtt45tN23ahPvuuw8A8Oabb+Kzzz5DSkoKjEYjunbtiieeeAIA8O6776KsrAxZWVkoKirC22+/DX9//wbtEULgueeeg9FoRFpaGrKyshqcxt28eTO++eYbnDlzBgcPHsQ777xz1WNcunQpXn/9dUdAvJQkSVi6dCkWL16M2traRmtG1GEJIvJIBoNB7NixQwghRM+ePcWXX37pWPfNN98Ig8EghBDib3/7m7jzzjvFiRMnnF5/4sQJ0b17d7Fjxw5hsViuuB+bzSaUSqU4cOBAg3VnzpwRAERWVpZjWUxMjPjoo48u+16vvfaauOuuuxzPAYhdu3Y5nt97771ixYoVQgghxo0bJ95++23Huh07dggAora2Vpw7d074+vqK6upqx/oPP/xQjB079rL7vVp9LhxDbW3tZV978frExEQRGxsrhBBixowZYuHChUIIIR566CHx17/+1fGaiooKoVAoxJkzZxo9zktt3LhRyOVyERgYKNRqtQAgnnzySWG32y+7fUlJiQAgSktLhRBCrFy5UkyfPl0IIURRUZHw9/cXRqNRCCFE3759RVJSkuO1RqNRKBQKUVtbK/7v//5PjBgxQvz6668N9jFmzBixfv36y+5/69atYvDgwY7nBoNBbNq0yfH8r3/9q/jDH/5wxWMdOXKkoybz5s0TQgih1+vFf//7XyGEEAsXLhQzZswQQghx4403irVr14ra2loBwFFfos6CPWVEXsBoNMJgMDieGwwGx+mzv/71r+jduzfGjx+Pnj17YuXKlQCA3r174/XXX8eiRYsQEhKCadOmOZ1yu6CwsBBmsxm9evW64v7DwsIcj1UqFSorKwEAx48fx4QJExAWFoaAgAA8//zzKCwsbNJrjUYjIiMjHesufpyZmYna2lqEh4cjKCgIQUFB+MMf/oD8/Pxm16c5Zs+ejby8PHzxxRdXfX+1Wo3g4GDk5OQ0epwDBgxwDOrftWsXAGD48OEoLS1FRUUFzp07hyNHjuD5558HUHda99lnn0WvXr0QEBDg6N28UNeZM2fiiy++QFVVFTZv3ozRo0cjPDzcUbdJkyY5atavXz/I5XLk5eVh1qxZuPXWWzFt2jTodDrMmzfvsr1SeXl5mDZtGvR6PQICAjBz5swm/z+9miVLlmDdunWOU6aX8+KLL2LZsmVXPR1K1JExlBF5AZ1Oh8zMTMfzs2fPQqfTAQA0Gg1effVVnD59Gp9//jlWr17tGDs2ffp0/PDDD8jMzIQkSXjmmWcavLdWq4VSqcSpU6ea3a7HHnsMffv2xYkTJ1BeXo7ly5c3eSxVeHg4srOzHc+zsrIcjyMjI+Hn54fCwkKUlpaitLQU5eXljlORl7pafZrD19cXCxcuxN/+9jen47j0/auqqlBUVAS9Xt/oex45csQxqH/06NEN1oeGhmLy5MmOIPjhhx9i27ZtSEpKQllZmeMKxAvt0ev1GDFiBD799FNs2rQJs2bNcrxXZGQkvv76a0fNSktLYTabodfr4ePjg4ULF+Lo0aP46aefsH37drz33nsN2vP8889DkiQcOnQI5eXleP/995v8//Rq+vbti7vvvttpTOKlbrnlFvTu3bvBKXCizoKhjMgL/P73v8eLL76IgoICFBYWYsmSJZg5cyaAurFMJ0+ehBACgYGBkMvlkMlkOHbsGJKTk1FTUwOlUgl/f3+ncWIXyGQyPPTQQ5g7dy6MRiNsNht2797tuJDgaioqKhAQEAC1Wo309PRmTe0wZcoUrFmzBjk5OSgtLcVLL73kWBceHo7x48fj6aefRnl5Oex2O06dOoWUlJRm16e5Zs2aBbPZjG+++cbp/Tdu3IgDBw6gpqYGzz//PGJjY684Rq85ioqKsHXrVsfVoRUVFfDz80NwcDCqq6sdPWgXu++++7Bq1SocOnQId999t2P5o48+ihdeeMERIAsKCrBt2zYAwH//+18cOnQINpsNAQEB8PHxueznoaKiAmq1GoGBgcjJycHLL7/c6mO8YOHChdi4caPThQOXWrZsWYMpNIg6C4YyIi8wf/58DBs2DIMGDcLAgQMxZMgQzJ8/HwBw4sQJxMfHQ61WY8SIEXj88ccxbtw41NTU4Nlnn4VWq0VYWBjy8/OxYsWKy77/K6+8goEDByImJgbdunXDM88847iC82peeeUVfPjhh9BoNEhISMDUqVObfEwJCQkYP348Bg0ahBtuuAG33347FAoF5HI5AOC9996DxWJB//790bVrV9xzzz3Izc1tdn2aSy6XY8mSJU6D0uPj47F06VJMnjwZ4eHhOHXqFP71r3+16P0BYPfu3Y5Tmv369UP37t3x5ptvAqgLXAaDAXq9Hv3798fw4cMbvH7SpEmOU5UqlcqxfM6cObjzzjsxfvx4aDQaDB8+HKmpqQCAc+fO4Z577kFAQAD69euHMWPGOPWyXbBw4ULs378fgYGB+N3vfucU+lqrR48emDVrFqqqqq64zciRI3HjjTe6bJ9E3kQSruiXJiJqpa+//hqPPvqo02lCurJevXrh73//O+Lj493dFCJyEfaUEZFbmEwmfPXVV7BarcjJycHixYsxadIkdzfLK2zZsgWSJF113jUi8j7sKSMit6iursaYMWOQnp4Of39//O53v8OaNWsQEBDg7qZ5tLFjx+Lo0aPYtGkTbr31Vnc3h4hciKGMiIiIyAPw9CURERGRB2AoIyIiIvIADGVEREREHkDh7ga4Qktup9JcWq22wa1GOivWwhnrUY+1cMZ6OGM96rEWzjpTPa52txH2lBERERF5AIYyIiIiIg/AUEZERETkARjKiIiIiDxAhxjoT0RERE0jhIDZbIbdbockSe5uDgAgLy8PNTU17m6GSwghIJPJoFQqm11fhjIiIqJOxGw2w8fHBwqF50QAhUIBuVzu7ma4jNVqhdlshr+/f7Nex9OXREREnYjdbveoQNYRKRQK2O32Zr+OoYyIiKgT8ZRTlh1dS+rMUEZERETkARjKiIiIyCPExsZi586d7m6G2zCUEREREXkAhrJGCHM1xKH/wV5W4u6mEBERUQfGUNaYvFzY31gCy5Ff3N0SIiKiTqGmpgYLFizAkCFDMGTIECxYsMAxj1lxcTHuu+8+9OvXDwMGDMCkSZMcVzomJiZi6NCh6NOnD0aPHo1du3a58zCarUnXxFZWVmLdunU4ePAgNBoNpk+fjlGjRjXY7vPPP0dKSgoKCgqg0Whw66234s4773Ssz8/Px7p163DixAlotVo89NBDGDRokGP99u3bsW3bNlgsFsTGxiIhIQE+Pj4uOMxWCI8AJAnWzNNAn0GNb09EROQl7P9aD5F1pk33IUX2gGxaQrNe88Ybb2D//v349ttvIUkSHnzwQaxZswbz5s3D3//+d4SHh+PgwYMAgP3790OSJJw8eRIbN27El19+ibCwMGRlZcFms7XFIbWZJvWUbdiwAQqFAuvXr8dTTz2F9evXIysrq8F2Qgg88cQT2LhxI1544QV88803+PHHHx3r16xZg+joaPzzn//EtGnTsHr1apSXlwMADhw4gG3btmHBggVITExEfn4+Nm/e7KLDbDnJ1w/oHg7r2dPubgoREVGnsHXrVvz5z3+GVqtFcHAw5s6diy1btgComwMsPz8f2dnZ8PHxQWxsLCRJglwuh8ViwfHjx1FbW4vIyEhER0e790CaqdGeMrPZjNTUVLz66qtQKpXo27cvhg0bhp07d2LGjBlO206cONHxWKfTYdiwYTh27BhGjhwJo9GIM2fOYP78+fD19cXw4cPx1VdfYc+ePRg/fjxSUlIwbtw4REZGAgAmT56MN954o8E+3EIfxVBGREQdTnN7sNpLXl4eIiIiHM8jIiKQl5cHAHjsscfw6quvYvr06QCAGTNm4Mknn0SPHj2wePFirF69GsePH8eYMWOwcOFChIWFueUYWqLRUJabmwu5XA6dTudYZjAYcPTo0au+TgiB9PR0xMfHAwCys7MRGhrqdMsBg8GA7Oxsx/qYmBindWVlZaioqIBGo3F676SkJCQlJQEAVq5cCa1W29hhtErlNf1R9etehGg0kPz82nRf3kChULR5zb0J61GPtXDGejhjPeq5sxZ5eXkeOaO/QqFw9HiFhoYiNzcXAwYMAACcO3cOYWFhUCgUCAoKwtKlS7F06VKkpaVh8uTJGDJkCOLi4nDvvffi3nvvRUVFBf7yl79gxYoVSExMdMvx+Pn5Nfv/cZN6yi69d5NKpYLZbL7q6z755BMIITBu3DjH+6hUqgbvU1xcfNn1Fx6bTKYGoSw+Pt4R9gCgsLCwscNoFdFVC9jtKDxyAFJUrzbdlzfQarVtXnNvwnrUYy2csR7OWI967qxFTU2Nx91nUqFQwGq1QggBm82GiRMnYvXq1Rg4cCAkScIrr7yCSZMmwWq1YseOHejduzeio6OhUqkgl8sdHUHnzp1DTEwM5HI5/Pz8YLPZYLVa3XJMNTU1l/1/fHEn16UaDWVKpRImk8lpmclkglKpvOJrvvnmG6SkpGDJkiWOgfpKpRLV1dUN3udC4Lt0/YV9Nvdmnm1CbwAAiJyzDGVERERtbM6cOaisrHR0wEyYMAFz5swBAMdQqKKiIgQGBuK+++7DyJEjcfToUaxYsQInTpyAj48Phg4dilWrVrnzMJqt0VAWHh4Om82G3NxchIeHAwAyMzMdY78ulZycjM8++wyLFy9GcHCwY3lERATy8/OdglhmZiZGjhzpWJ+ZmYmbbrrJsS4wMLBBL5lbdA8HFD5AToa7W0JERNRhpaamOh5fOEV5qUceeQSPPPJIg+X9+/fHl19+2abta2uNXn2pVCoRGxuLjz/+GGazGenp6di3bx/i4uIabLtr1y589NFHmD9/PkJDQ53W6XQ6REdH45NPPoHFYsHevXuRmZmJ4cOHAwDGjBmD5ORkZGdno6qqClu2bMHYsWNdc5StJCkUUEQYIHLOurspRERE1EE1aaTf7NmzsXbtWiQkJECtViMhIQGRkZFIS0vD8uXLsWnTJgDAv/71L1RWVuK5555zvHb06NGORDtnzhysXbsWDz74ILRaLebOnYuAgAAAwODBgzFx4kQsXrzYMU/ZlClTXH28LaaI6gnr4f3ubgYRERF1UJIQQri7Ea1lNBrbfB/+KV+h8v23IVvzESRVlzbfnyfjYF1nrEc91sIZ6+GM9ajnzlpUV1c3uPDO3S4M9O9IrlTnqw30522WmkhxYYC/kacwiYiIyPUYyppIYegJABA5mW5uCREREXVEDGVNJNOGAn7+AEMZERERtQGGsiaSZDJAFwnB05dERETUBhjKmkGKiAZyMtABro0gIiIiD8NQ1hy6KKCyAqgodXdLiIiIqINhKGsG6fztlsBJZImIiDzCTz/9hKFDhza6XWxsLHbu3NkOLWo5hrLm0EcBAARvt0REREQuxlDWHJogQB3AnjIiIiJyuSbdZonqSJIE6A2cq4yIiDqEDf/Lw5kSc5vuo0dXJWYPC210u8TERBw4cADr1693LFuwYAGEELjuuuuwdu1a5ObmIjg4GI8//jhmzZrV4jbV1NRg2bJl2L59OwBgwoQJeOGFF+Dn54fi4mL86U9/wr59+yCTydCnTx9s2bIFMpkMiYmJ+Oc//4mKigqEhoZi+fLlGD16dIvbcSmGsmaS9AaIH7+DsNvrpskgIiKiVps4cSJWr16NyspKqNVq2Gw2fPHFF9iwYQNKSkrw7rvvwmAwYM+ePZg5cyYGDx6MgQMHtmhfb7zxBvbv349vv/0WkiThwQcfxJo1azBv3jz8/e9/R3h4OA4ePAgA2L9/PyRJwsmTJ7Fx40Z8+eWXCAsLQ1ZWFmw2mytLwFDWbPoooMYEFBcA2saTPxERkadqSg9We4mIiMDAgQPx9ddf495778WPP/4If3//BoP4R4wYgTFjxiA1NbXFoWzr1q1YunQptFotAGDu3Ll45plnMG/ePCgUCuTn5yM7Oxs9evRAbGwsAEAul8NiseD48eMIDg5GZGRk6w74MtjV00ySjldgEhERtYW77roL27ZtA1AXnO666y4AQHJyMiZMmIABAwagX79+SE5ORnFxcYv3k5eXh4iICMfziIgI5OXlAQAee+wxREdHY/r06RgxYgTeeustAECPHj2wePFirF69Gtdffz0ee+wxnDt3rsVtuByGsubS8QpMIiKitnDHHXdg9+7dMBqN+OabbzBp0iTU1NQgISEBjz76KA4cOIC0tDTcfPPNrZrIPTQ0FNnZ2Y7nOTk5CA2t6zVUq9VYuHAhdu/ejY0bN+If//gHdu3aBQCYNGkSPvvsM6SmpkKSJCxbtqx1B3wJhrJmklRdgG5a9pQRERG5WHBwMEaMGIG5c+ciMjIS11xzDWpra2GxWBAcHAyFQoHk5GSkpKS0aj8TJ07EmjVrUFRUhOLiYrz22mu4++67AQA7duzAmTNnIISARqOBXC6HTCbDyZMn8cMPP6CmpgZ+fn5QKpWQuXhsOceUtYQ+GsLIKzCJiIhc7a677sKcOXMwf/58AHU9V0uXLsWjjz4Ki8WC+Ph4jB8/vlX7mDNnDiorKxEfHw+g7urLOXPmAADOnDmD+fPno6ioCIGBgbjvvvswcuRIHD16FCtWrMCJEyfg4+ODoUOHYtWqVa072EtIogPcyNFoNLb5PrRaLQoLCwEA9n+/A/Hd55C9uRmSovPl2otrQazHxVgLZ6yHM9ajnjtrUV1dDZVK5ZZ9X4lCoYDVanV3M1zqSnXW6XRXfA1PX7aE3gBYrUBBrrtbQkRERB1E5+vmcQFJHwUBQGRnQgp3/SWxRERE1Dw5OTkYO3bsZdd9//330Ov17dugFmAoa4mwCECSAcZMAKPc3RoiIqJOT6/X48SJE+5uRqvw9GULSL5+QEg4b7dERERepwMMJfcKLakzQ1lL6Q2cFoOIiLyOTCbrcIPqPY3Vam3RdBk8fdlCkj4K4pfdEJaaup4zIiIiL6BUKmE2m1FTUwNJktzdHACAn58fampq3N0MlxBCQCaTQalUNvu1DGUtJOkNdV2TudmAoZe7m0NERNQkkiTB39/f3c1wwulS6vD0ZUudvwcmb7dERERErsBQ1lIh4YDCh+PKiIiIyCUYylpIksuB8AjebomIiIhcgqGsFSRegUlEREQu0qSB/pWVlVi3bh0OHjwIjUaD6dOnY9SohpOmHj58GFu2bMHp06ehVquRmJjoWFdYWIg///nPTtvX1NRg1qxZuOOOO3DkyBEsWbIEvr6+jvUPP/zwFWfn9Qg6A7Dne4jqSkgqtbtbQ0RERF6sSaFsw4YNUCgUWL9+PTIyMrBixQoYDAZERjrfYkipVGLcuHEYOXIktm7d6rROq9Vi06ZNjuf5+fn44x//iNjYWMeyrl274u23327N8bSrC7dbgvEs0Lu/u5tDREREXqzR05dmsxmpqamYOnUqlEol+vbti2HDhmHnzp0Ntu3duzfi4uIQEhLS6I5TUlLQv3//Jm3rsfTnr8DM5rgyIiIiap1Ge8pyc3Mhl8uh0+kcywwGA44ePdrinQohsHPnTkyePNlpeVlZGRISEuDr64uYmBhMmzbtspOvJSUlISkpCQCwcuVKaLXaFrelqRQKRYP9iOBgFPiroCzOR0A7tMFTXK4WnRnrUY+1cMZ6OGM96rEWzliPOo2GMrPZ3GCSOZVKBbPZ3OKdpqeno7S0FMOHD3cs0+v1ePnll6HT6VBYWIjExES89957eOSRRxq8Pj4+HvHx8Y7n7THh3JUmthO6KJhOHYOlE016x0n+nLEe9VgLZ6yHM9ajHmvhrDPV4+JOrks1evpSqVTCZDI5LTOZTC26fcAFKSkpiI2NdXqPoKAgREREQCaTISQkBDNmzEBqamqL99FeJF0UYMzkDV6JiIioVRoNZeHh4bDZbMjNzXUsy8zMbDDIv6ksFgt2797d6FWVkiTBbre3aB/tSm8AKiuA8lJ3t4SIiIi8WJN6ymJjY/Hxxx/DbDYjPT0d+/btQ1xcXINt7XY7LBYLbDYbhBCwWCwN7kS/d+9eqNVqDBgwwGn54cOHUVBQACEECgsL8eGHHyImJqaVh9f2JF1U3YMcDvYnIiKilmvSlBizZ8/G2rVrkZCQALVajYSEBERGRiItLQ3Lly93THWRlpaGxYsXO143c+ZM9O/fH4sWLXIsS0lJwejRoxvcmT4jIwNvvvkmqqqqoNFoEBMTg9///vcuOMQ2duEKzJxMSP0Hu7kxRERE5K0k0QEGQxmNxjbfx9UGIdrmzoI0KAayB55q83Z4gs40ILMpWI96rIUz1sMZ61GPtXDWmerRqoH+1AR6A4SRt1siIiKilmMocwFJbwCMZyG84cIEIiIi8kgMZa6giwJqzEBRvrtbQkRERF6KocwFpPOD/XkFJhEREbUUQ5krnJ8WQzCUERERUQsxlLmA5K8CgkMADvYnIiKiFmIocxVdFHvKiIiIqMUYylxE0huAczkQl9zBgIiIiKgpGMpcRR8F2KxAfttPZEtEREQdD0OZi0i6+tstERERETUXQ5mrhEcAMhmnxSAiIqIWYShzEcnHFwjRQeTwCkwiIiJqPoYyV9JHAUb2lBEREVHzMZS5kKQzAAXnIGpq3N0UIiIi8jIMZS4k6Q2AEMC5LHc3hYiIiLwMQ5kr6c/fbimbpzCJiIioeRjKXCkkHFD4cFwZERERNRtDmQtJMjmgi+RcZURERNRsDGUuJukMAKfFICIiomZiKHM1fRRQWgRRVenulhAREZEXYShzMUlfd7slzuxPREREzcFQ5moX7oHJwf5ERETUDAxlrtZNC/irOK6MiIiImoWhzMUkSQJ0UewpIyIiomZhKGsDkr7uCkwhhLubQkRERF6Coawt6AxAVQVQVuLulhAREZGXYChrA9L52y3xCkwiIiJqKoaytnB+WgzO7E9ERERNpWjKRpWVlVi3bh0OHjwIjUaD6dOnY9SoUQ22O3z4MLZs2YLTp09DrVYjMTHRaf0TTzyB0tJSyGR1WfDaa6/F/PnzHeu3b9+Obdu2wWKxIDY2FgkJCfDx8WnN8bmFpAkEAoJ4D0wiIiJqsiaFsg0bNkChUGD9+vXIyMjAihUrYDAYEBkZ6bSdUqnEuHHjMHLkSGzduvWy7/XMM89g0KBBDZYfOHAA27Ztw4IFC9C1a1e88sor2Lx5M2bMmNGCw/IAegMEp8UgIiKiJmr09KXZbEZqaiqmTp0KpVKJvn37YtiwYdi5c2eDbXv37o24uDiEhIQ0uyEpKSkYN24cIiMjoVarMXnyZHz//ffNfh9PIemiAONZCLvd3U0hIiIiL9BoT1lubi7kcjl0Op1jmcFgwNGjR1u0wzfffBN2ux09evTAzJkzER0dDQDIzs5GTEyM00zxROkAACAASURBVD7KyspQUVEBjUbj9B5JSUlISkoCAKxcuRJarbZFbWkOhULRrP1UXzsAFd99ga72WihC9G3YsvbX3Fp0dKxHPdbCGevhjPWox1o4Yz3qNBrKzGYz/P39nZapVCqYzeZm7+yPf/wjevbsCSEEvvrqKyxbtgyvv/46unTpArPZDJVK5bQPADCZTA1CWXx8POLj4x3PCwsLm92W5tJqtc3ajwjoBgAoOfwLJIVfWzXLLZpbi46O9ajHWjhjPZyxHvVYC2edqR4Xd3JdqtHTl0qlEiaTyWmZyWSCUqlsdkP69u0LX19f+Pn5YdKkSejSpQvS0tIc+6murnbaB4AGgdBrnJ8Wg+PKiIiIqCkaDWXh4eGw2WzIzc11LMvMzGwwyL8lJElyPI6IiEBmZv3VipmZmQgMDGzQS+YtJKUKCA7hXGVERETUJE3qKYuNjcXHH38Ms9mM9PR07Nu3D3FxcQ22tdvtsFgssNlsEELAYrHAarUCqDvFmJ6eDqvVCovFgs8//xzl5eW49tprAQBjxoxBcnIysrOzUVVVhS1btmDs2LGuPdr2pjdAGNlTRkRERI1r0pQYs2fPxtq1a5GQkAC1Wo2EhARERkYiLS0Ny5cvx6ZNmwAAaWlpWLx4seN1M2fORP/+/bFo0SKYTCZs2LABeXl58PHxQXR0NJ5//nlHT9jgwYMxceJELF682DFP2ZQpU9rgkNuPpI+COLIfwloLSeF9860RERFR+5FEB7hrttFobPN9tGQQon3P9xD/txqyRW/V33qpA+hMAzKbgvWox1o4Yz2csR71WAtnnakerRroTy0nOW63lOHehhAREZHHYyhrS2ERgEwG8ApMIiIiagRDWRuSfHyAUD0E74FJREREjWAoa2OSLorTYhAREVGjGMramt4AFOZB1DT/DghERETUeTCUtTFJHwUIARiz3N0UIiIi8mAMZW1NHw0AHFdGREREV8VQ1ta6hwI+vhxXRkRERFfFUNbGJJkcCI/kjcmJiIjoqhjK2oGkjwJ4+pKIiIiugqGsPegNQGkxRFWFu1tCREREHoqhrB1IurrbLXFcGREREV0JQ1l7cNwDk+PKiIiI6PIYytpD12DAvwvHlREREdEVMZS1A0mSAH0UBE9fEhER0RUwlLUTSWcAcs5CCOHuphAREZEHYihrL/oooLoSKCt2d0uIiIjIAzGUtRPp/GB/ZPMUJhERETXEUNZezk+LwXtgEhER0eUwlLUTSRMABHYFOC0GERERXQZDWXvS8QpMIiIiujyGsnYk6Q1A7lkIu93dTSEiIiIPw1DWnnRRgMUCFOa5uyVERETkYRjK2pEUEV33gKcwiYiI6BIMZe0pPBIAOK6MiIiIGmAoa0eS0h/QhgJGXoFJREREzhjK2pvewJ4yIiIiaoChrJ1JuiggLwfCWuvuphAREZEHUTRlo8rKSqxbtw4HDx6ERqPB9OnTMWrUqAbbHT58GFu2bMHp06ehVquRmJjoWFdWVoaNGzciLS0NZrMZUVFRuO+++3DNNdcAAI4cOYIlS5bA19fX8ZqHH34YY8eObeUhehi9AbDZgHM5wIWB/0RERNTpNSmUbdiwAQqFAuvXr0dGRgZWrFgBg8GAyMhIp+2USiXGjRuHkSNHYuvWrU7rzGYzevfujfvvvx+BgYFITk7GypUrkZiYCKVSCQDo2rUr3n77bRcdmmeS9AYI1A32lxjKiIiI6LxGT1+azWakpqZi6tSpUCqV6Nu3L4YNG4adO3c22LZ3796Ii4tDSEhIg3WhoaGYMGECunbtCplMhvj4eFitVhiNRtccibcI0wNyOQf7ExERkZNGe8pyc3Mhl8uh0+kcywwGA44ePdqqHWdkZMBqtSIsLMyxrKysDAkJCfD19UVMTAymTZvm6EW7WFJSEpKSkgAAK1euhFarbVVbmkKhULhsP4W6KCgKchHUDu1uC66sRUfAetRjLZyxHs5Yj3qshTPWo06jocxsNsPf399pmUqlgtlsbvFOq6ur8eabb+Kee+6BSqUCAOj1erz88svQ6XQoLCxEYmIi3nvvPTzyyCMNXh8fH4/4+HjH88LCwha3pam0Wq3L9mMP1aMm42S7tLstuLIWHQHrUY+1cMZ6OGM96rEWzjpTPS7u5LpUo6cvlUolTCaT0zKTyXTZHqymsFgseOmll3DNNddg0qRJjuVBQUGIiIiATCZDSEgIZsyYgdTU1Bbtw+Ppo4CCcxA1LQ+2RERE1LE0GsrCw8Nhs9mQm5vrWJaZmdlgkH9T1NbW4uWXX0ZwcPBle8AuJkkS7B30xt2SPrruAceVERER0XlN6imLjY3Fxx9/DLPZjPT0dOzbtw9xcXENtrXb7bBYLLDZbBBCwGKxwGq1AgCsViteffVV+Pj44IknnoBM5rzrw4cPo6CgAEIIFBYW4sMPP0RMTIyLDtPD6KMA8HZLREREVK9JU2LMnj0ba9euRUJCAtRqNRISEhAZGYm0tDQsX74cmzZtAgCkpaVh8eLFjtfNnDkT/fv3x6JFi3D8+HHs378fvr6+eOCBBxzbPP/88+jXrx8yMjLw5ptvoqqqChqNBjExMfj973/v2qP1FNpQwNcXyGFPGREREdWRhBDC3Y1orfaYVsPVgxBtL84Fuqgh//MSl71ne+lMAzKbgvWox1o4Yz2csR71WAtnnakerRroT21D0kWxp4yIiIgcGMrcRW8AyoohKsvd3RIiIiLyAAxlbiLpDXUP2FtGREREYChzn/OhTBh5BSYRERExlLlPUDdA1QXgtBhEREQEhjK3kSQJ0BkgePqSiIiIwFDmVpI+CjBmogPMSkJEREStxFDmTvpooLoKKClyd0uIiIjIzRjK3Eg6f7slcLA/ERFRp8dQ5k66C/fA5LgyIiKizo6hzI0kdQAQ2I1XYBIRERFDmdvpoyCM7CkjIiLq7BjK3EzSGYDcsxB2m7ubQkRERG7EUOZuEQbAYgEK8tzdEiIiInIjhjI3k3QX7oHJcWVERESdGUOZu+kiAfAemERERJ0dQ5mbSX5KoHsYwGkxiIiIOjWGMk+gi4Lg6UsiIqJOjaHMA0h6A5BvhKitdXdTiIiIyE0YyjyB3gDYbEBetrtbQkRERG7CUOYBJH3dFZi83RIREVHnxVDmCUJ1gFzOaTGIiIg6MYYyDyApfIBQPW+3RERE1IkxlHkISW9gTxkREVEnxlDmKfQGoDAPwmxyd0uIiIjIDRjKPISkj6p7wFOYREREnRJDmafQXbgCk6cwiYiIOiOGMk+hDQV8fdlTRkRE1EkpmrJRZWUl1q1bh4MHD0Kj0WD69OkYNWpUg+0OHz6MLVu24PTp01Cr1UhMTHRan5+fj3Xr1uHEiRPQarV46KGHMGjQIMf67du3Y9u2bbBYLIiNjUVCQgJ8fHxaeYjeQZLJgHDebomIiKizalJP2YYNG6BQKLB+/Xo89dRTWL9+PbKyshpsp1QqMW7cOMyaNeuy77NmzRpER0fjn//8J6ZNm4bVq1ejvLwcAHDgwAFs27YNCxYsQGJiIvLz87F58+ZWHJr3kfQG9pQRERF1Uo2GMrPZjNTUVEydOhVKpRJ9+/bFsGHDsHPnzgbb9u7dG3FxcQgJCWmwzmg04syZM5gyZQp8fX0xfPhwREVFYc+ePQCAlJQUjBs3DpGRkVCr1Zg8eTK+//771h+hN9EbgLISiIpyd7eEiIiI2lmjpy9zc3Mhl8uh0+kcywwGA44ePdqsHWVnZyM0NBT+/v5O75Odne1YHxMT47SurKwMFRUV0Gg0Tu+VlJSEpKQkAMDKlSuh1Wqb1ZaWUCgUbb6fmv4DUQogsKoUvj16tum+WqM9auFNWI96rIUz1sMZ61GPtXDGetRpNJSZzWanIAUAKpUKZrO5WTsym81QqVQN3qe4uPiy6y88NplMDUJZfHw84uPjHc8LCwub1ZaW0Gq1bb4foQ4CAJQePQhZWFSb7qs12qMW3oT1qMdaOGM9nLEe9VgLZ52pHhd3cl2q0dOXSqUSJpPzhKYmkwlKpbJZjVAqlaiurm7wPhcC36XrL+zz0kDYoQV2A1RqgDcmJyIi6nQaDWXh4eGw2WzIzc11LMvMzERkZGSzdhQREYH8/HyngJeZmYmIiAjH+szMTKd1gYGBDXrJOjJJkgB9FISRV2ASERF1Nk3qKYuNjcXHH38Ms9mM9PR07Nu3D3FxcQ22tdvtsFgssNlsEELAYrHAarUCqOuui46OxieffAKLxYK9e/ciMzMTw4cPBwCMGTMGycnJyM7ORlVVFbZs2YKxY8e69mi9gKSPBnLOQgjh7qYQERFRO2rSlBizZ8+GxWJBQkIC1qxZg4SEBERGRiItLc1p+ou0tDTMnDkTK1asQGFhIWbOnIkXX3zRsX7OnDk4ffo0HnzwQXzwwQeYO3cuAgICAACDBw/GxIkTsXjxYjz++OPo3r07pkyZ4uLD9QL6KMBUBZR0jnPrREREVEcSHaBLxmg0tvk+2msQojh+BPaXn4PsqYWQBg5t8/21RGcakNkUrEc91sIZ6+GM9ajHWjjrTPVo1UB/amfnb0zOcWVERESdC0OZh5G6aICgbgBvt0RERNSpMJR5Ip0BgtNiEBERdSoMZR5IijAAuVkQdpu7m0JERETthKHME+kMQK0FyD/n7pYQERFRO2Eo80DS+cH+4GB/IiKiToOhzBOFRwKSxHFlREREnQhDmQeS/JSANpRXYBIREXUiDGWeSm+AMLKnjIiIqLNgKPNQkt4A5OVA1Na6uylERETUDhjKPJXeANjtwLlsd7eEiIiI2gFDmYeSdAYAgOC4MiIiok6BocxThYYDcgWnxSAiIuokGMo8lKTwAcL0nBaDiIiok2Ao82CS3sBpMYiIiDoJhjJPposCivIhzNXubgkRERG1MYYyDybp6wb7g6cwiYiIOjyGMk92PpRxElkiIqKOj6HMkwWHAL5+HFdGRETUCTCUeTBJJgN0UewpIyIi6gQYyjwcr8AkIiLqHBjKPJ3eAJSXQlSUubslRERE1IYYyjycpI+qe8DeMiIiog6NoczTOe6ByXFlREREHRlDmacL7Ap00fAemERERB0cQ5mHkyQJ0BsgePqSiIioQ2Mo8wKSPgownoUQwt1NISIiojbCUOYNdAbAVA0UF7q7JURERNRGFE3ZqLKyEuvWrcPBgweh0Wgwffp0jBo1qsF2Qgh88MEHSE5OBgDcfPPNmDFjBiRJQlpaGpYvX+60fU1NDebOnYvhw4fj+++/x7p16+Dr6+tY/+yzz2LAgAGtOb4OQdIbIIC6cWXB3d3dHCIiImoDTQplGzZsgEKhwPr165GRkYEVK1bAYDAgMjLSabukpCTs27cPL7/8MiRJwtKlSxESEoLx48ejX79+2LRpk2PbI0eO4KWXXsLgwYMdy/r06YOlS5e66NA6EF3dtBgiJxPSwGFubgwRERG1hUZPX5rNZqSmpmLq1KlQKpXo27cvhg0bhp07dzbYNiUlBXfccQeCg4PRrVs33HHHHUhJSbns+6akpGD48OFQKpWtP4oOTuqiBoKCAU6LQURE1GE12lOWm5sLuVwOnU7nWGYwGHD06NEG22ZlZcFgMDhtl5WV1WA7s9mMPXv24JlnnnFanpGRgYcffhhqtRqjR4/GpEmTIJfLG7w+KSkJSUlJAICVK1dCq9U2dhitplAo2mU/V1LS4xrY83MQ7MY2XODuWnga1qMea+GM9XDGetRjLZyxHnUaDWVmsxn+/v5Oy1QqFcxm82W3ValUDbYTQtRN7XDe3r17odFo0L9/f8eyfv364dVXX4VWq0V2djZee+01yOVyTJo0qcF+4uPjER8f73heWNj2A+C1Wm277OdK7N3DIQ79jIK8PEiXCartyd218DSsRz3Wwhnr4Yz1qMdaOOtM9bi4k+tSjZ6+VCqVMJlMTstMJtNlTzteuu2F7S4OZEDdqcsxY8Y4LQ8NDUVISAhkMhmioqJwzz33YM+ePY01r/OI7g1YayG+3OzulhAREVEbaDSUhYeHw2azITc317EsMzOzwSB/AIiMjERGRobjeUZGRoPtCgsLceTIEcTFxV11v5cGuc5OGnoTpBE3Q3zxEezb/+Xu5hAREZGLNamnLDY2Fh9//DHMZjPS09Oxb9++y4aquLg4fPnllyguLkZxcTG2b9+OMWPGOG2zc+dOXHvttQgLC3Na/ssvv6C0tBQAkJOTgy1btmDYMF5peIEkk0N64I+QRoyD2PYhgxkREVEH06QpMWbPno21a9ciISEBarUaCQkJiIyMdMw9dmGqi1tuuQX5+fl4+umnAQC/+c1vcMsttzi9186dO3HnnXc22MehQ4ewdu1amM1mBAYGOgb6Uz1JJgceeAoA6oIZJMgmTHVzq4iIiMgVJNEB7t1jNBrbfB+eNAhR2G0QG9+A2PNfSHfNhOx3U9p1/55UC0/AetRjLZyxHs5Yj3qshbPOVI+rDfRvUk8ZeRZJJgcePN9j9tn7sAPtHsyIiIjItRjKvFR9MBMMZkRERB0AQ5kXqwtmcwCwx4yIiMjbMZR5OUcwE+d7zCQJstvvdXeziIiIqJkYyjoASSYHHvoTAEBs3VQXzG67x82tIiIiouZgKOsgHMFMAOLT9+pOZTKYEREReQ2Gsg7Eqcfs0/fq5jG7bbKbW0VERBfbmVGOHf814rc91bgpSsM72JADQ1kHI8kvBDMB8em7sEuA7LcMZkRE7mYXAh/8Woh/HymCyleOVcZyXKv1x4NDuqNfd5W7m0cegKGsA6oLZn+uG/y/5d26U5kMZkREblNda8PqH3OxL6cS43sH4tlb+2Prz2fw/q+FePbbs7gpSoP7BndHuMbX3U0lN2Io66AkuRx4eC4A1AUzSYLs1rvd3Coios4nt8KC5SnZyC634JFhobi9TxD8FHLE9wrCKEMAPksrxtajRdibXYHb+nTF1Ou00PjJ3d1scgOGsg7MKZj9+526HjMGMyKidnPwXBVW7cqBALDo5khcH9bFab1SIcO0gVqM7x2ED38twJfHSpB8ugxTr9Pi9j5B8JHL3NNwcguGsg6uYTCTILuVN3onImpLQgh8faIU6/+XB53GF/PHRlz11GQ3fwWeHB6OO/p2w7u/5OOf+/Px5fES3De4O0byYoBOg6GsEUII7MqswJ3dgt3dlBZzDmYb6wb/j2cwIyJqC7U2gfX/y8N/TpZimK4Lnh6lg8qnaacjDUF+WDAuEgdyq7Bxfz5e/sGIbcFKPDQkBP1CeDFAR8dQ1ogj+Sa8+qMRX5+qwFM3eu8gTEcwEwLik411pzIZzIiIXKrMbMWqXTk4nG/C3f27Yeb13SGXNb+Xa3B4F6y+LRr/PVOGD34txLM7zmJEpAb33+C9f4eocfJFixYtcncjWquioqLN3jtE7YPIQF8knyrFV8dLEKRUoGdXP6/sSpZkMuCG4UBuNkTS54DSH1Kvvs1+H5VKherq6jZooXdiPeqxFs5YD2cdvR4ZJWb87bssZJdbMGdEOO7qFwzZFf5WNKUWMklCz25K/PaaIPjKJfz3TBm2HytBRY0NvYP94afoOOPNOvpn42IajeaK6xjKmiAqyA933RCNQ9kl+OJYCTJKa3B9qMorvxB1wWxEfTDzVzU7mHWmL09TsB71WIs6Z8tqsGZ3Lj45eA6GQF8Eq3zc3SSP0JE/H3uyKrD0+2zIZBIW/yYSQ3Xqq27fnFooZBKuC1UhvlcQqix2fHOyFP85UQqZBPTqpmxRT5yn6cifjUsxlLlA964BuDFUAZWPDN+cKEPymXIYgvy8shtZksmAwcMhzmUBLQhmnenL0xSsR73OXovKGhveO1CAN/bkoqzGhlob8EV6MUy1dvTv7g9FB/jj2Rod8fMhhMAnh4uwdm8eenRVYulvIhER6Nfo61pSC38fGWIi1LgpUoPscgu+PlGKlDPlCFIqEBXo65VncC7oiJ+NK2EocwGVSgWTyYS+3VW4MUKNn42V+Dy9BJUWG64LUXndj60kk0FqYTDrTF+epmA96nXWWtjsAt+eLMWKnTk4kl+N8b2D8FycHveN6IX80kp8ebwEP5wtR48gJULUnbfXrKN9Pmqsdqz+KRdfHi/FmOgAPBunh8avaUO1W1OLQKUCY3oEol93fxwtqMZXx0vxs7EKEQG+6N7FOz9fHe2zcTUMZS5w8Qemq78Cv+kZCJPVju3HSpCaXYF+3f3R1d+7rptwBLPc88FMpYLUs/Fg1pm+PE3BetTrjLU4kleNFbtysONUGXoHK/FcXATG9w6CUiFDUIAaA7rJMCDEH3uzK7EtvQRlZiv6h/h3yvmnOtLno6CqFguTs3Aorxr3D+6OB4eEQNGM/6euqEWYxhe39ApCqNoHqdmV2H6sBJmlZvTsqvS6yWc70mejMQxlLnDpB0YhkzBUp8a1WiV2ZZRj+7FS+Ckk9NH6e1UXsiSTQbqhecGsM315moL1qNeZalFQVYvE1HN455cCKGQSnogNwwM3hKDbRePHLtQjVO2LW3oHocZmx1fH6045RQT6euXwh9boKJ+PtIJqLPguC6VmG56N0+M3vYKa/bvvqlpcuBjgtvMXAySfvxigrMaGa7opPX7ss10IFFRZ4eunhL22xt3NaRdXC2WSEEK0Y1vahNFobPN9aLVaFBYWXnZdudmKt1LPITW7EoNCVZhzUzi0XjawV1itsK9/Bdj/E6SpsyGLv/OK216tFp0R61GvM9SixmrH1qPF2HK0CAAwuX8wJvXvdtk/fperR3qBCW/uyUV2uQU39wzEw0NCoPayXo2W6gifj+9OlWLt3jxoVQq8MDYCUU0YP3Y5bVWLEpMVHx0sxI5TpfBXyHDPdcGYcG1X+Lq5Z7bSYkNOuQU55RYYyy3Iqah7nFthgcUmIJOA60JVGBUVgOGRagQqvevMU3PodLorrmMoa6LGvkBCCCSdKsOGn/OgkEl4/MYwjDQEtHm7XKkumL0M7N8NaVoCZL+547LbdYQfVldiPep15FoIIfDj2Qq8sz8fBdVWjIzS4MEhIVcdw3Olelhsdnx8qAifHi1CoJ8cf7gxDCMir/yv547Cmz8fNrvAO7/k4/P0EgwKU2HeKH2rThG2dS3Oltbg3V/y8T9jFUK6KDBrcAhGGTRXnKLDFWptAnmVFkf4yqk4H8DKLSirsTm2k0lAmNoHOo0v9AG+0AX4olr4Ykf6ORgraiGTgEGhKow0BGB4pAYBHewfLQxlLtDUL5Cx3ILVPxlxosiMcT0C8EhMaJNncvYEwmqF/R+rgF/2XDGYefMPa1tgPep11FqcLjZjw895OJJvQo+ufkgYGooBoY3Prt5YPU4Xm/HGnlycKanByCgNHokJRVAH7iHw1s9HpcWGV34w4pfcKvzu2q54aEhIqy/uaq9a/Hqu7s4AZ0pqcE2wEg/eENKkz+6VCCFQYrYhp7ymvtfrfADLq6yF/aJEEaiUQ6+pC136C/9pfBGq9oWP3Ll+Wq0WBQUFOFNSgx/PVuDHs+XIPR/Qrg/rgpFRGgyP1HjdWLnLYShzgeZ8gax2gY8PFeLfR4qgVflg7k3hXnV7DOdg9ghkv5ngtN5bf1jbCutRr6PVotxsxfu/1p0K6uIrx8zrtbilV1CT54VqSj2sdoFPjxbh40NF8PeRIWFoCOKiA7xqbGpTeePnI7u8Bsu+z0F+lQV/iAnD+N5BLnnf9qyFXQh8f6Yc7x8oQJHJitgINe6/IQT6gCuPaTRb7TCWW5BdboGxwuJ06tFktTu285VLCNfUB64LPV96jW+zTstfWg8hBM6U1OCHzHL8eLYC5yprIb8Q0AwaDI/QeO1pf4YyF2jJFyitoBqv/ZSLgqpa3DMgGFMHar1m6gxhrYX9Hy/XBbPfPwLZzfXBzBt/WNsS61Gvo9TCahf4+ngJPjpUCFOtHb/r0xXTBmqb/UegOfU4W1aDt/bk4lihGTH6LnjsxrAON+mst30+9hsr8coPRshlEp6N02OAC/9x7Y5a1Fjt2JZejC1HilFrs+O31wTh1mu6orCq1ulUY065BUUmq9Nru6sU9b1dAX6O4KXtonDJKdGr1UMIgVPFNfjxbF1Ay6ushUJWF9BGGQJwY4Qaal/vCWgMZS7Q0i9Qda0N6/+Xj+TTZbgmWIk/36S76r9OPImw1sL+95eBA87BzNt+WNsa61Enr9KC05UyhPha0aOrX5uOXWlLB3KrsOHnPGSVWTA4TIWHh4W222Bum11g+7ESvP9r3RWdDw4JwS29AjtMr5m3fFeEEPg8vQTv/JKPqEA/vDAmwuXzy7mzFqUmKz46VIhvT5Y6nW7s4iNrcKpRH1B3lXBbX8XZ1HoIIXCy2IwfM+tOceZXWaGQAYPDumCkIQCxEWp08fCAxlDmAq39Av14thxrU8+h1ibw8NBQjO/tHT+0dcFsFXAgFdL0P0A27nde88PaFoQQyK2oxeH8ahzJr0Z6gQkRXbtgTJQKsRFqj7/83NWsdoF9OZX4z4lSHMitwoUfkwA/OQaFqTA4rAsGh3fxigktcyss2Lg/H6nZlQhT++ChoSG4Ua9u1fe0pd+V3AoL3ko9h8N51RgUpsKTsWEIVXvHP+auxht+O2ptdqzdm4fk02UYEanGnBE6+Pu4/nvtCbXIKqvB8UITws6Hr0A/udv+LrWkHkIInCgy48ezFfghsxyF1VYoZBJuCO+CUQYNboxQe+SYboYyF3DFF6iouhav787FwXPVuDFCjSdjw7zist9Lg1nIvfe7/cekvQghkFVmwZH86vNBzISS8936gX5y9O3uj7Pltcgtr0EXXxniDAGI7xWEXt2886b1TZVXacG3J8vw3alSlJhtCFYpcEuvQNw2MAoHMs7hQG4VDpyrdtRKp/HF4HAVBod3wcBQlUf9UFbX2vDvw0XYll4ChQy49zotJvbt6pLJXVvzu2EXdXcJeGd/AexCYNbg7ri9T1evvs+hMfCIHgAAIABJREFUJwSRqykxWbFiZw6OFZowbWDdkJO26vH19Fq0t9bWQwiB40Vm/JhZjh/OVqCo2gofmYQhurqLBGI8KKC1OpRVVlZi3bp1OHjwIDQaDaZPn45Ro0Y12E4IgQ8++ADJyckAgJtvvhkzZsxw/HGaMmUK/PzqTwOMHDkSjz76aJNeezXeEsqAuh/aL9JL8N6BAqh9ZXhqeDiG6q9+41pPcHEwUz/4FKqH31x3D80OxmYXyCytwZHzPWFH8k0oP38pdzd/Ba4LUWFAqD8GhKgQEVB3r7luwcH4/shZJJ0qw+6sClhsAtFBfojvFYgxPQI7zOXcVrvA3uwK/OdkGQ7kVkEmAUN1atzaOwhDdF0gl0lO35MLgfaX3Cr8eq4Kh/OqUXN+PqJrtf4YHNYF14er0CfY3y1Bwy4EUs6U490DBSgxWTGuRwBmDe7u0nFcrvjdKKiqxbq95/CzsQp9tf744/CwJt1b0dOYrXaEhXRHaXGRu5tyWaeKzViWko3KGhvm3BSOkVFtO6URQ5kzV9bDLgSOF5rxw9ly/JRZgSJTfUAbZQjAMH0Xtwa0Voey119/HUIIPPbYY8jIyMCKFSvw4osvIjIy0mm7HTt2YPv27ViwYAEkScLSpUtx2223Yfz48QDqQtkbb7yBsLCwBvto7LVX402h7IKMEjNW/5iLzLIa3N4nCA/cEOLxp74uDmaI6gXZ5Psh9R/s7ma1is0ucKrY7AhhRwtMqLLUXVkU0sUH150PYANCVAhT+1z2HwkXfzYqLTbsyihH0qkynCw2QyH7//buPC7q+973+Os7C8wM+w4iEvc1riga1ygaY2P2xh7Nntia5La356ZLetpzmpvYNLXpNV2uNkftcoxtMPbcNs1eNGLjFpNqxCWmmoyiIIsKCDPDMvO9f/yGgREUjMKM8Hk+Hj74zW+Z+fJ1GN58f99Fkds3mryBcYxJj7omWzlKzzfwt6NVbP6smiqPl2SHhbmD4skbGNdmkuRL/Zw0en18UulmX6mLj0/XcfSMBw04rCauT3MwJj2KcRlRZMS0X89X06eVbtZ+VMaRSg+Dk2wszUljaLL9qr/O1frc0P7Rc2s/KqO+SfOV65O5fURiWA8cqvI0cajcxYFyN4fKXTjP1WM2KWN+Kn9/pdZf422hu3X2d2cNv9hVSlykmX+b2ZcBibYuf00JZcG6qj58WnOkwu2fZuM8Z91NRJgVE/pEMbVfLDmZ0V1ye/pSLhXKOrx35vF42L17Nz/72c+w2WwMGzaMnJwctm3bxpIlS4LOLSwsZOHChSQlJQGwcOFCNm/e3KlgdSXXXouuS7Dxws3Z/Ne+Cv76yTn2n3bx5NQ+3fJh8EUpixXTY08Rfegf1Kz/Nb6V/wHDxxjhLHtQqIvXKY1eH0fPeDjg/2XxSYUbj394d5+YCKb2iwmEsC/SDyo6wszNQxK4eUgCznMeCo5Vs9VpjBhKchhrps4ZEEd6mC+v0+htbhWr4uPTLkwKcjKNVrFxGV8sXFrNJq5Pi+L6tCjuI4Waei9Fp+vYd7qOfaUudp+sBSA1ysIYf1+00elRV7Wl8ay7ifX7KtjyWTUJNjP/c0oGs/rHhv2gBKUUNw6IY2xGFC/tKWP9xxVsP1HD1ydnhM1nRqWrkYNlRuvywXIXJ2saAGPKhGEpdhZdn4Ql0s6xsipKahrZW1JHY6te5naLqVVIsxoj/PzbXdWq4dOaP3xcyasHzzA8xc5TMzJ79DxxvZFJKYanOhie6uDhCal8UuHm/RPn2XG8hp3FtUSYFTmZ0UztF0NOZjS2EDeOdPjuKy0txWw2ByW77OxsDh061Obc4uJisrOzg84rLi4OOueHP/whPp+PoUOHcv/995Oamtrpa3uaCLOJRyekkdMnmhd3lvLtd5wsHp3C7cMTw7ZFRZnM2GfdTO3QMeitb6Hf3Ihv+f9CTZyOun0JKvXifwGEQn2TjyOVbg6VuzlQ7uJIpZsGr/GLIDsukhv7xzIqzcGIVAeJV3lB+esSbDyaY+OBcSl8cKqWzceq2XTwDBsPnGFUmoO5A+OYkhUTVi2kpecbePdoFZuPVVNd7yXFYWHx6GTyBsZd9ekZYiPNTM2OZWp2LFprTtc2+vui1bH9xHn+dqwaBQxItDE23eiPNjzliy3k3ej18ddPzpF/4AxNPh93jkjky6OSwqaPSWcl2C08NSOTHSdqeGlPGd9628ldI5O4Z1RSty5wrrWmrLYx0M/yULmL07WNgNHyOTzFzpwBcYxMczAgwRaYKLR1a4jXp6l0NVJyvjGw7E5JTQNHzrj5+/EaWt/CSbCZ6RMb4Q9pLaMC25uEtLNcjV5e3FHK7pO15A2MY9nEtF65SHxvYlKKEanG5/0j45sDWg07T5xnx4nzRJgVd41I4iujk0NWxk61lNntwc36DocDj8fT7rkOh6PNeVprlFI8/fTTDBkyhPr6el555RV+8pOfsGLFCsxmc4fXtlZQUEBBQQEAzz//PMnJXV+BFouly14nLzmZiYP68JPNR/mvfRUUVdTzg3lDSI8Nj7+AL2SxWEjJ6AP/8gi+Wxfh+ssfqHvtFfQ/dmCfextRix7GHJ8YkrK5GrwUldaw71Q1+07WcKjsPE0+jQIGp0Rx+/UZjOsbx+g+scTbr07I6Mx747a0VG4bD+Xn63nrcDlvHCpj5Y5S/jOinLwhKdwyMo3haVc20u+LavT62HbsDK8dOM2HxdWYFUwdkMito9KZ1O/yOpZfyc9JSgpc3x/uw+i/9knZeT44UcWeE1X8+ZNz/OnQWWwWE2Mz45jYL56J/eIZkOS4ZJ1prdnx+Tl+sc3JyWoPU/sn8vXp/clKuPq3KtvTVZ8btyYnM2tEFj/f9jkbD5TzQYmbf5s7mJHpXbNUk9aa4+fcgZ+rfSXVVNQaLWFxNgtjMuO4Z3ws4zLjGJh88ZbUC+sjDRjZznn1TT5OVbspPuemuMpD8Tk3J6rc7ClxUXWsOnCeSUFGrI2seDtZCXb6NX9NsJMSHXHRFtCSag/ff/sQzrMuvjlzAHePyej2n72u/J1yLQpFfaSlwsyRxh8I+0tq2PLPSgakRYf0/6XDUGaz2XC73UH73G43NlvbwHDhuc3nNb/ZR4wYYbyoxcJDDz3EAw88wKlTp+jXr1+H17aWl5dHXl5e4HF33Jfvjvv//5qbzOgUK2s+LOf+l//BsknpzLgu/NbPbFMX8+7ENGkW+vVXcL/7Z9zvvYmaeztq3u0oe9euZFDb4OWwvxXsYLmLY2c9+LTxYT0o0catwxIYmepgWIq91eSCmqa6airrrk4ZLue9YQK+NMDOzf2zOVTupuBYFW8dLuMvB06THRfJnIFxzOof2y2jcktqjFaxLZ8ZrWKpURaWjElmzoDmVjEf5y6zU/bV/DlJt8KtAx3cOtCBq9HLgTJXoCVt1/FzgNGCMiY9ijEZxu3O1q2dJ6vrWftROXtL6+gbG8EPb+zL+D7R4K2j8mr953egqz83HhufyMS0CFZ9cJplGz9m4dAEloxJueLWV582BrwcKGtpCWteuzDBbmFUqp27hhs/W33jWocfD+fOtv2Dvdnl1EcsMDIeRsbb4DobkABAbb2XkvMXzjLvYu/JKuq9Le1rEWYVaFkLrLEYE0Fdg5eVO0vxac0Pb8xibEYEZ850/+AD6VMWLNT1kWWDB66PA7o+U1xRn7KMjAy8Xi+lpaVkZGQAcPz48Tad/AGysrJwOp0MGmT0L3I6ne2e11rzOIMvcm1Po5Qib2A8I1MdrNxRws+2l/DhqVq+OjEt7GcrVvGJqHsfR+fdhv7zy+jXX0FvfRN1yyLUjPko6+W3SmmtqW3wcc7TRJW7iSqPlypPE+f825+f8+A8V48GLCbFkCQbd41IYlSag6HJ9m7vvHk5TEoxKs3BqDQHS3O8xu26o1X85h/l/H5vOZP6RpM38Iv337qYRq+PXcW1vHO0iqIyo6/YpL5GX7FwHojgsJqZ1DeGSX2NlqCKukY+Pl3HvtI6/lFax1ZnDWDckh6T4cCn4a1Pz2GzmHhkQioLhiSEdaf4K5GTGc2vbunP7/dW8JdPzrH7ZC1fn5zBqMtY37DJp/nsEgNexveJYlTapQe8dJfoSDNDIu0MuWBghtaas+4mI6Sdb1kSyHnOw67i80GTpPaNjeD7M/vS5xqZyFv0Hp1qKcvNzSU/P59ly5bhdDrZs2cPy5cvb3PujBkzeOONNxg/fjwAr7/+OvPnzweMPmNer5d+/frR0NDAK6+8QmJiIpmZmR1e29tkxETw47nZvHrgDPkHKjlU7uJfb+hzRYvIdheVnola9l305//E99+/R7+yBv23v6Buvxc1aQYohavRZwQsd5MRsjxNnHMbgas5fJ3zNFHtaaLVEmsBFpNxyyQzJoKvjE5mVKqDwUm2sOqbdTmiIszMGxTPvEHxnKiqZ/Nn1bz3WTU7i2tJtFuY7R8ccCW/QE61ahWrqfeSGmXl3jHJzBkYf9X70nWHlCgreQPjyRsYj8+/Rl5zK9pbn1bR5NPMHRTHkjEpvaLjtsNq5rFJ6UzLjuFXu07z/YITzB8czwPjUtrtN9fo9fHpGU9g2pdPKlx4mozUkhl75QNeQkEpRZLDSpLDyuj0qKBjTT6jD1xJTQPV9U1MyYoJ+1nfRe/U6XnKVq1aRVFREdHR0SxZsoRp06Zx+PBhnnvuOdavXw+0zDW2efNmAObMmROYa+zAgQOsWbOGs2fPEhkZyZAhQ7jvvvsCrW+XurYj1+KUGJ11pNLN/9leQlltIwuGxNM3LhKrSWExKaxmhdX/1WIyti3m4K9WsynofLPiiv/Kba8uPE0+fwtWE1Vub6B161xpGVUnijnnNVFtT6AqIoYG3fb1TcoIWgk2M/E2C/F2/7bdYjy2mUnwb0dHmMJqYtaueG80ejUfltSy+VgVH5XU4dMwIsXO3EHx3NAvplMjhBq9Pnb6W8UOlLkwK5jUN4abBsczJt3RJSMOQ30LAoz+SK5GHwlhEDZDUR+eJh8bPjZGdSc5LDw+KZ2RaQ6OVLo5UObiULmLI5WewMjH7PhIRqUaU7+MSHV0ab2Fw/sjXEhdBOtN9SEz+l8FoXzDuBq9rPuonIJWHVy/KAXth7hWIa/dwGdWWE0mLGZFREQkp6tqW1q3PE2Bv7IvfK1Ym5kEm5m4+vPElxwl/nwl8fFRJEyYSEJWX3/QMhMTaQ77aQkupqvfG2dcjWz93Jj7rOR8AzaLienZMeQNjGdoctt+lyer641Wsc9rOF/vJS3ayryB8cwZGNflQaU3fbB2Rijr40ilm1/sLOVkTQNmBV5/X8sBCTZGptoZmeZgRIqDmG6c3FjeHy2kLoL1pvqQUHYVhMMbxtXopaFJ0+jTNHqNr02BbR9NPqN1pPl4U+uvPk2TVwcfa2e79XM2+Xw0eaHR5ws6ZjaZiI00kWDzt2LZzca2P2DF2ywk2C3ERpqD+ijppkb0tnfQr+fD+WqYcAOm2+9DpWeGsFavXHe9N7TWHK5wU3Csmu0navA0afrGRpA3MI6p/WI5VOHi3aNVHCx3Y1aQmxXDTYPiGd1FrWLtCYefk3AS6vpo9Pr465Fz1DX4GJlqZ1iKPaTTgIS6PsKJ1EWw3lQfEsqugt70hunIFa9R5nGh3/0z+t0/Q2MDato81MKvoEI0jcaVCsV7w9XoZceJ8/ztaDWfVLaMWk6PtjJvUDxzBsQRH4Lbd/JzEkzqI5jURwupi2C9qT6uaPSlEFebsjlQty5Gz7oZ/fpGo/Vs1xZU3m2om+5EOaI6fpJezmE1Bzq6n6yuZ/fJWgYm2rq1VUwIIcTVJaFMhIyKTUAt/ho671b0Xzag33wVXfg2asGXUTcuQFlluHpn9I2LvCYXqBZCCBHs2pxDQPQoKjUD09JvYfrBSsgehH71N/h+8Bi+HZvRPm+oiyeEEEJ0CwllImyo7IGY//V/Y/pfz0JMHPq3P8f3zDfR+/fQA7o+CiGEEJckoUyEHTV8DKZ/ewH11e9AYwO+Xz6L76ffQx/7JNRFE0IIIbqM9CkTYUmZTKiJ09DjJqPffxf911fwPf8dGDsZ0533oTJ61xJcQgghej4JZSKsKYsFNWsBevKN6ILX0O/8N74ffoCaOsdYVzMpNdRFFEIIIa4KCWXimqBsdtQti9Azb0a/uRG99U309gIYPBKVOxM1YSoqKjrUxRRCCCG+MAll4pqiYmJRix5F592G3rEZvbsQvf7/ov/4Elyfgyl3FozOkek0hBBCXHMklIlrkkpKQS38CvqWRXD8qBHO9vwd395dYI9CjZ+Cyp0JQ0ehTKFbVkYIIYToLAll4pqmlILrBqOuG4y++yE4sh+9qxD90Xbj9mZ8ImrSDCOgZQ1os3i3EEIIES4klIkeQ5nNMGIcasQ4dMNj6I/3oHdvRW/+q7HOZkaW0f9s0gxUSnqoiyuEEEIEkVAmeiQVEYmaOA0mTkPX1qA/3G7c4vzzy+g/vwwDh6FyZ6FypqFiYkNdXCGEEEJCmej5VHQsatbNMOtmdGUZ+oNtRkD7w6/R+WuM1rXcmaixk1GRsoakEEKI0JBQJnoVlZyGWvBl9M13w0mncXtz9zZ00YfoSBtq3GRU7iwYPsa4HSqEEEJ0EwlloldSSkFWf1RWf/SdD8A/DxqtZx9uR+/aCjFxLQMErhssAwSEEEJ0OQllotdTJhMMvR419Hr0v3wNij7Et7sQXfg2evNfITXDuL2ZOwuV1ifUxRVCCNFDSSgTohVltcL4KZjHT0G7atH/2InetRX9ej76r68YrWa5M1GTpqNiE0JdXCGEED2IhDIhLkI5olHT5sK0uehzZ/wDBLai89eiN/7G6HeWOxPf3FtCXVQhhBA9gIQyITpBJSShbroDbroDXXLC6H+2uxD92xep+MOvUeOmoG6YbdwGNZlCXVwhhBDXIAllQlwm1acf6o770LffC0cPE7l3B+73C9C73oPEFNTkG1E3zJb+Z0IIIS6LhDIhviClFAweQeyUGdTffh963270zi3otzah39xoTFB7w2xjglpHdKiLK4QQIsxJKBPiKlARkahJM2DSDHTVGWNwwI4t6PWr0H9cY8x/NmU2jBgr858JIYRol4QyIa4yFZ+Emn8X+qY74fhR9I7N6A/+jt7zd4hLRE2eiZoyB5XZL9RFFUIIEUYklAnRRZRSxhQa1w1Gf/kRKNqDb8cWdMFr6Hf+H2QPMm5vTpqBipb1N4UQoreTUCZENzDmP7sB8/gb0DVV6A8Kjdubf/xPY3qN0TmYbpgNo3JQFvmxFEKI3qhTn/61tbWsXr2a/fv3ExMTw+LFi5k2bVqb87TWbNiwgS1btgAwe/ZslixZglKKkpISXn75ZY4cOYLP52PQoEE89NBD9OljjFDbunUrq1evJiIiIvB8Tz31FCNHjrwa36cQYUPFxqPyboO829AnPzfC2a6t+PbuguhYY3LaG2ZD1gBZ3kkIIXqRToWytWvXYrFYWLNmDU6nkx//+MdkZ2eTlZUVdF5BQQF79uzhpz/9KUopnn32WVJTU5k3bx4ul4ucnBwef/xxbDYbmzZtYsWKFbz44ouB64cMGcKzzz57db9DIcKY6tsfdc8jxvqbB/fi27kZXfiWsbxTZrZxezN3FipOVg8QQoiersNZLj0eD7t372bRokXYbDaGDRtGTk4O27Zta3NuYWEhCxcuJCkpicTERBYuXEhhYSEAgwYNYvbs2URHR2OxWLjlllsoKSnh/PnzV/+7EuIaoywW1JiJmJc9hemF36MWL4OISPSrv8X3nYfw/uIZ9IfvoxsbQl1UIYQQXaTDlrLS0lLMZnPgNiNAdnY2hw4danNucXEx2dnZQecVFxe3+7yHDh0iPj6emJiYwD6n08kjjzxCdHQ006dP54477sAs0weIXkZFxaBuXAA3LkCXFhtzn+3ciu+lFeCIMgYGTJkN/YfI7U0hhOhBOgxlHo8Hu90etM/hcODxeNo91+FwtDlPax30y+PMmTOsW7eO+++/P7Bv+PDh/OxnPyM5OZmTJ0+ycuVKzGYzd9xxR5vXKSgooKCgAIDnn3+e5OTkTnyrV8ZisXTL61wLpC6CdWl9JCfD9ePQj3yThqKP8Lz3Jp4dW9Bb38Kc2Q/7jQuwzZyPOTm1a17/Msl7I5jURzCpjxZSF8GkPgwdhjKbzYbb7Q7a53a7sdlsHZ7bfF7rQFZTU8Py5cu56aabggYLpKWlBbb79evH3XffzWuvvdZuKMvLyyMvLy/wuLKysqNv44olJyd3y+tcC6QugnVbffQdAPf9D0x3P4z+8H28O7dQ+/Kvqd3wEgwbjRo4HNL6GMs7pfUJySoC8t4IJvURTOqjhdRFsN5UH63vPF6ow1CWkZGB1+ultLSUjIwMAI4fP96mkz9AVlYWTqeTQYMGAcbtyNbn1dbWsnz5cnJycrjzzjsv+bpyW0aI9im7AzV9Hkyfhy4vRe98D/3h39Fv5IPW6OYTY+IgNQOVlhkU1kjpg4qMDOW3IIQQoh2dainLzc0lPz+fZcuW4XQ62bNnD8uXL29z7owZM3jjjTcYP348AK+//jrz588HwOVy8aMf/YihQ4eyZMmSNtfu3buX/v37Ex8fz6lTp/jTn/7E5MmTr/T7E6JHU6kZqNsWw22L0Y2NUHkayk6hy0qgrARdVoI+uBd2bG4JawAJyS1BLbVPILiRnCbzpAkhRIh06tP30UcfZdWqVSxdupTo6GiWLl1KVlYWhw8f5rnnnmP9+vUAzJ07l/Lycp588kkA5syZw9y5cwH44IMPOHbsGCdPnmTr1q2B5165ciXJyckUFRWxatUqPB4PcXFxgY7+QojOUVYrZGRBRhYXtjNrjwvKS9FlpVB2yghs5SXoD7dD3fmWwGYyQXKaP6j1gbRMVFoGpGVCQjLK1OGAbSGEEF+Q0lrrjk8LbyUlJV3+Gr3pfndHpC6CXev1oWtrAq1qlDe3sJ2C8lKobzWgxxoBKen+FrbMlluj6X0gJh6l1DVfF1eb1EcwqY8WUhfBelN9XFGfMiFEz6aiY42VBAYOC9qvtYbqsy2BrTmsnT6F3v8heJtaWthsdkjLpCrrOnwx8cZt0ORUo9UtMQVlsXb79yWEENcaCWVCiHYppSA+CeKTUEOvDzqmfV44UxHUwqbLTtH0z0PoyjLwelsCW/PzJKWiktPAH9ZUkj+0JSSjZD5CIYSQUCaEuHzKZDZuZaako0aND+xPTk6morwMzp2FM2VGQKssh8oy9Jly9KcHYPcZ0L7gfmwJycFBrXWAi080Xk8IIXo4CWVCiKtKmcyQlAJJKagho9oc101NcK7SCGqVZXCmVWg7tBeqzhrnNV9gtkCiP7T5A1vQdlyCTKEjhOgRJJQJIbqVslhaWtnaOa4bG4xbo2fK/aHNaG3TlWXofbvhfLVxXvMF1gh/CPS3riWlGaEt1f8aIZhEVwghvggJZUKIsKKsEZCeCemZ7Ye2+nojqDWHtsryQIubdh6FuvPGec0XRMUY4SwlHVIzICXDv50OcYnSyiaECBsSyoQQ1xQVGQl9+kGffu2HNrcLKsugohRdcRrKT6MrStGffwofbg/uzxYRAcn+FrWUDEhtFd4SU2UiXSFEt5JPHCFEj6LsDsjqD1n9206i29Ro3Bq9ILBRcRp9eB80NAQPQEhMaRvYUjKMfTZ7N39nQoieTkKZEKLXUBarsZxUWp+2ga15Xrby0/7AVhoIb/qjC1Y+gJa1Rf0hzQht/u2YOLktKoS4bBLKhBCCC+ZlGzKyzXHtqjVa1MpPQ3PrWnkp+kgR7N4avBh8pB1S0jmXloHPbAWHA2wOY5JdRxTYHCi7HezGNq22lVUm2hWit5JQJoQQnaAc0ZA9CJU9qM0x3dhg9GNrfTu04jS+sxXo2vPgdhn/vE0t11zshSwWf1Dz//Nvq1bbndov/eGEuObIT60QQlwhZY1odzH4pAvW89ONDUY487hagprHhXa13YfbjXbXGdtnKvzbbnDXgc/X8pwXK5Q1wmiZs0e1BDa7AxV4HBVooVOBxxd8jYiQ27BCdCMJZUII0U2UNcIIS7Hxwfsv4zm01tDQ0E6IcxkjTy+2310HNVUt2x53y3Ne7MXM5paA1irgBQU7hyOw3dJq5z/ucECkXYKdEJ0koUwIIa4hSimIjDT+xSUEH7uM59E+r7/lrVWIc9e1hLbWX10utMf/+Ex5q3PcoDtotVOmQItcpSMKr8UKEZEQEYmKtBnfR4QNWm9HREKkzZj+5CLHiIyUhe5FjyOhTAgheiFlMoMj2vjXev9lPIfWGurd4GoJdXhcwcHO1bJt0T68tTVQXw/VZ42JgOs90OAx9rXqcweXaMFrZjb7Q1ukP+j5tyNtEOEPdZG2tseaB1rY7C398ZoHYkTaUCbTZdSCEFePhDIhhBBfiFLKH2Ycwfsvcn78BX3sLqSbmqChviWkNQe2hnqor0e3DnCBYw3Gdr0H3eDf73ZB9bmWx/X+59QtMe+SgS8orPnDm92Bat4XaTda/5oHVQTOv+A6i1Vu3YrLIqFMCCFEWFAWizH61BHV/vEreG6tNTQ2GAHN42q5dVvvbtUXz23sb+6L17yv3m2EvMA5rs4FPLOlJajZgkNcTXwCPp82+hharWCNNFaYsEYYAyyskf79EUZLn38/zfv928psvoJaEeFGQpkQQogeTykV6MtGTGzwsct8LmOwRX3QSNnmoBc02KLeHThm7Hcbgy3KSqg/Wo/2eKCxHpqa2r5GZwtjNvuDXXN4uyDIBUJeRLv7jT56dqO1L9LmD482ozWwuZVQWvy6jYQyIYQQ4jIYgy38AxBIDD7WyedIbnUrV/u80Nho3IptbP5X3/LY/1U3NhhhsDF4f8u2cY1ubLXfVWdc2/q6xoY2QfCSIdBsbglszWHN/1hdZL9xzH5ByPMfiwhtvz3t8xnhVd8uAAALV0lEQVQtndr/1ddq22xGRUSGrGwSyoQQQogQUiYzRPqDz6XOu4qv2RIE640WvPrmW7ce45aux3PBfrfRb6/1vtoa/2P/uY0Nwa9xqQI0h1p/gDsTacPb0BAclrQ25uQLbHtbtrXPH6Zan+9rOd93wf7W+y5BzVqAWrLsyiv4C5JQJoQQQvQyQUEwJq7t8S/wnMZADU8gwLWEObcR3vyBjwsCn673YDIpiPSCUmAyGV+VAmUyWiZNJmN6FVPLfi61P7Cv1deg65qPBe9XWf2vvHKvgIQyIYQQQlwxY6BG22lWoOOQl9DByNzeQiZjEUIIIYQIAxLKhBBCCCHCgIQyIYQQQogwIKFMCCGEECIMSCgTQgghhAgDEsqEEEIIIcJAp6bEqK2tZfXq1ezfv5+YmBgWL17MtGnT2pyntWbDhg1s2bIFgNmzZ7NkyZLA8gxOp5PVq1dz6tQpMjMzeeyxx7juuus6da0QQgghRE/WqZaytWvXYrFYWLNmDd/4xjdYs2YNxcXFbc4rKChgz549/PSnP+WFF17go48+4m9/+xsATU1NrFixgunTp/Pb3/6WmTNnsmLFCpr8Sz1c6lohhBBCiJ6uw1Dm8XjYvXs3ixYtwmazMWzYMHJycti2bVubcwsLC1m4cCFJSUkkJiaycOFCCgsLATh48CBer5cvfelLWK1WFixYgNaaAwcOdHitEEIIIURP12EoKy0txWw206dPn8C+7OzsdlvKiouLyc7Obve85mOtb0e2d7yj1xBCCCGE6Ik67FPm8Xiw2+1B+xwOBx6Pp91zHQ5Hm/O01m2ONR93u90dXnthv7KCggIKCgoAeP7550lOTu7o27hiFoulW17nWiB1EUzqo4XURTCpj2BSHy2kLoJJfRg6DGU2my0QnJq53W5strar2V94bvN5Sql2n8flcgUC36WuvVBeXh55eXmBx92xXlayrMsVIHURTOqjhdRFMKmPYFIfLaQugvWm+mh95/FCHd6+zMjIwOv1UlpaGth3/PhxsrKy2pyblZWF0+kMPHY6nYHzsrKyOH78OFrrwPETJ04EHb/YtUIIIYQQPV2nWspyc3PJz89n2bJlOJ1O9uzZw/Lly9ucO2PGDN544w3Gjx8PwOuvv878+fMBGDlyJCaTibfeeou5c+eyefNmAEaNGtXhtR25VOq8mrrrda4FUhfBpD5aSF0Ek/oIJvXRQuoimNRHJ6fEePTRR2loaGDp0qX8/Oc/Z+nSpWRlZXH48GHuu+++wHlz585lwoQJPPnkkzz55JOMHz+euXPnAsb94m9/+9sUFhby4IMP8t577/Htb38bi8XS4bXh4Kmnngp1EcKG1EUwqY8WUhfBpD6CSX20kLoIJvVh6NTksdHR0XznO99ps3/48OGsX78+8Fgpxb333su9997b7vP079+fn/zkJ+0e6+haIYQQQoieTJZZEkIIIYQIA+ann3766VAX4loxYMCAUBchbEhdBJP6aCF1EUzqI5jURwupi2BSH6B06+GQQgghhBAiJOT2pRBCCCFEGJBQJoQQQggRBjo1+rI3q62tZfXq1ezfv5+YmBgWL17MtGnTQl2sbtfY2MjatWspKiqitraWtLQ0Fi9ezLhx40JdtJArLS3lW9/6Frm5uXzjG98IdXFCavv27WzatInKykri4+N5/PHHGT58eKiL1e3Ky8tZt24dn376KRaLhcmTJ/Pggw9iNptDXbRu8fbbb7N161ZOnDjB1KlTeeKJJwLHioqKWLduHZWVlQwePJjHH3+clJSUEJa2a12sLj799FPy8/P57LPPMJlMjBw5koceeoiEhIQQl7hrXeq90WzTpk1s3LiRH/zgB4wePToEpQwdCWUdWLt2LRaLhTVr1uB0Ovnxj39MdnZ2r1ttwOv1kpSUxNNPP01ycjJ79+5l5cqVvPDCC6Smpoa6eCG1bt06Bg4cGOpihNz+/fvZsGED3/zmNxk0aBBVVVWhLlLIrFu3jtjYWF566SVcLhfPPvss77zzDgsWLAh10bpFQkICd955Jx9//DENDQ2B/TU1NbzwwgssW7aMCRMmkJ+fz4svvsiPfvSjEJa2a12sLurq6sjLy2PMmDGYzWbWrVvHqlWr+P73vx/C0na9i9VHs9OnT7Nz584eH04vRm5fXoLH42H37t0sWrQIm83GsGHDyMnJYdu2baEuWrez2Wzcc889pKamYjKZmDBhAqmpqXz22WehLlpIbd++HYfDEViZojfbuHEjd999N0OGDMFkMpGYmEhiYmKoixUS5eXlTJkyhYiICOLj4xk7diwnT54MdbG6TW5uLpMmTSImJiZo/wcffEBWVlagbr785S/jdDo5depUiEra9S5WF+PGjWPKlCk4HA4iIyOZP38+R44cCVEpu8/F6qPZunXrWLJkSWBi+d5GQtkllJaWYjabg5Z+yM7Opri4OISlCg9VVVWUlpb2uhbD1lwuFxs3buSBBx4IdVFCzufzcezYMWpqavj617/OsmXLWLduXbt/CfcGCxYsYMeOHdTX13P27Fn27dvH2LFjQ12skCsuLiY7Ozvw2GazkZ6eLp+pwOHDh3v15ynAzp07sVqtgeUWeyMJZZfg8Xiw2+1B+xwOBx6PJ0QlCg9NTU388pe/ZObMmWRmZoa6OCGTn5/PjTfeSFJSUqiLEnJVVVV4vV527drFM888w4oVK3A6nfzpT38KddFCYvjw4RQXF/PAAw+wbNkyBgwYwMSJE0NdrJDzeDw4HI6gffKZCsePH2fTpk29ekUbt9vNH//4Rx588MFQFyWkJJRdgs1mw+12B+1zu93YbLYQlSj0fD4fv/rVr7BYLDz88MOhLk7IOJ1OioqKuOWWW0JdlLAQEREBwPz580lISCA2NpYvfelL7N27N8Ql634+n4/nnnuO3Nxc1q9fz7p166irq2PDhg2hLlrItfeZ6nK5evVn6unTp3nuued46KGHeuWgmGavvvoqM2bM6PV9lHvnTdtOysjIwOv1UlpaSkZGBmD8RdNbm5i11vz617+murqa733ve732nj/AwYMHqaio4LHHHgOMFgCfz8d3v/vdi67v2pNFR0eTlJSEUiqwr/V2b1JbW0tlZSXz58/HarVitVqZNWsW+fn5vbolBCArK4vCwsLAY4/HQ1lZWa/9TK2oqODZZ5/lrrvuYsaMGaEuTkgVFRVx9uxZ3nnnHcAYFLJy5Upuu+02br/99hCXrvv03t+qnWCz2cjNzSU/P59ly5bhdDrZs2cPy5cvD3XRQmLNmjWcOnWKf//3fw+0jPRWeXl5TJ06NfD4tddeo6KigqVLl4awVKE1a9Ys3n77bcaOHYvZbOaNN97olX1DYmNjSU1N5d1332XhwoV4PB4KCwvp169fqIvWbbxeL16vF5/Ph8/no6GhAbPZzKRJk1i/fj27du1i/PjxbNq0iezs7B7dDeJidVFdXc0zzzzDTTfdxLx580JdzG5zsfr4j//4D7xeb+C8733ve9x///29btolWWapA7W1taxatYqioiKio6NZsmRJr5ynrKKigieeeAKr1YrJ1HLX+6tf/SrTp08PYcnCw8aNGzl9+nSvnqesqamJ3/3ud7z//vtYrVamTJnCvffe2ysDvNPp5He/+x3Hjx8PzEH18MMPEx8fH+qidYuNGzeyadOmoH13330399xzD/v37+c3v/kNFRUVgXnKevItq4vVhVKKV199lcjIyKBj69ev787idbtLvTdae+KJJ/ja177W6+Ypk1AmhBBCCBEGpKO/EEIIIUQYkFAmhBBCCBEGJJQJIYQQQoQBCWVCCCGEEGFAQpkQQgghRBiQUCaEEEIIEQYklAkhhBBChAEJZUIIIYQQYUBCmRBCCCFEGPj/8jwqKCIWAewAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 720x432 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 393
        },
        "id": "BD6EMMy3ZTrU",
        "outputId": "fbc4d949-98fe-4eb0-b393-ce0d77548d8b"
      },
      "source": [
        "pd.DataFrame(history.history)[['accuracy','val_accuracy']].plot(figsize=(10,6))\n",
        "plt.title(\"accuracy change of Non-Bayesian NN\")\n",
        "plt.show()"
      ],
      "execution_count": 48,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 720x432 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "d2-cOGBZP4i1"
      },
      "source": [
        "### 2.3 The prediction of non-Bayesian model\n",
        "we can see the prediction(take X_test[0] as an example) is the same all the time as the Non-Bayesian model provides point esitmation"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "mQMAlMcEuimt",
        "outputId": "6f415b5c-612f-4367-b1de-be44db5954d8"
      },
      "source": [
        "for i in range(5):\n",
        "  print(f\"iter {i}\")\n",
        "  print(f\"predicted label: {np.argmax(mymodel.predict(X_test)[0])}\")\n",
        "  print(f\"predicted mat: {mymodel.predict(X_test)[0]}\")\n",
        "  "
      ],
      "execution_count": 49,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "iter 0\n",
            "predicted label: 7\n",
            "predicted mat: [2.9812066e-13 2.4266643e-13 1.5029285e-11 3.6832851e-10 3.3333031e-16\n",
            " 8.5711522e-17 9.2583576e-19 1.0000000e+00 8.7349545e-13 5.5102101e-10]\n",
            "iter 1\n",
            "predicted label: 7\n",
            "predicted mat: [2.9812066e-13 2.4266643e-13 1.5029285e-11 3.6832851e-10 3.3333031e-16\n",
            " 8.5711522e-17 9.2583576e-19 1.0000000e+00 8.7349545e-13 5.5102101e-10]\n",
            "iter 2\n",
            "predicted label: 7\n",
            "predicted mat: [2.9812066e-13 2.4266643e-13 1.5029285e-11 3.6832851e-10 3.3333031e-16\n",
            " 8.5711522e-17 9.2583576e-19 1.0000000e+00 8.7349545e-13 5.5102101e-10]\n",
            "iter 3\n",
            "predicted label: 7\n",
            "predicted mat: [2.9812066e-13 2.4266643e-13 1.5029285e-11 3.6832851e-10 3.3333031e-16\n",
            " 8.5711522e-17 9.2583576e-19 1.0000000e+00 8.7349545e-13 5.5102101e-10]\n",
            "iter 4\n",
            "predicted label: 7\n",
            "predicted mat: [2.9812066e-13 2.4266643e-13 1.5029285e-11 3.6832851e-10 3.3333031e-16\n",
            " 8.5711522e-17 9.2583576e-19 1.0000000e+00 8.7349545e-13 5.5102101e-10]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 269
        },
        "id": "VbXFIIELPuoK",
        "outputId": "c75ce7fa-1478-4522-b03c-350f77900f44"
      },
      "source": [
        "plt.imshow(X_test[0].reshape(28,28),cmap='binary')\n",
        "plt.show()"
      ],
      "execution_count": 50,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "iMBinL0ITvLm"
      },
      "source": [
        "## 3.0 Bayesian Neural Netwrok"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "EMfBHJUPXXv1"
      },
      "source": [
        "### 3.1 Define prior and posterior for Densevairantion layers (can be skipped for now)\n",
        "not used in BNN classification here"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BwSMYHe9W3sY"
      },
      "source": [
        "# prior is non-trainable, so we use tfp.layers.DistributionLambda\n",
        "# multivariate normal distribution is often chosen\n",
        "def prior(kernel_size, bias_size, dtype = None):\n",
        "  n = kernel_size+bias_size\n",
        "  prior_model = keras.models.Sequential([\n",
        "                                         tfp.layers.DistributionLambda(lambda t: tfd.MultivariateNormalDiag(loc = tf.zeros(n),scale_diag=tf.ones(n)))\n",
        "  ])\n",
        "  print(f'PRIOR: kerel size is {kernel_size}, bias size is {bias_size}\\n ')\n",
        "  return prior_model\n",
        "\n",
        "# using tfp.layers.VariableLayer to create a trainable layer and followed by the layer\n",
        "def posterior(kernel_size, bias_size, dtype = None):\n",
        "  n = kernel_size+bias_size\n",
        "  posterior_model = keras.models.Sequential([\n",
        "                                             tfp.layers.VariableLayer(\n",
        "                                                 tfp.layers.MultivariateNormalTriL.params_size(n),dtype=dtype\n",
        "                                             ),\n",
        "                                             tfp.layers.MultivariateNormalTriL(n)\n",
        "  ])\n",
        "  print(f'POSTERIOR: kerel size is {kernel_size}, bias size is {bias_size}\\n ')\n",
        "  return posterior_model"
      ],
      "execution_count": 51,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "kV5aqSYwXdxe"
      },
      "source": [
        "### 3.2 Define KL divergence for Denseflipout layers\n",
        "prior and posterior are default in the layer"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "iPHNqYrGjhZw"
      },
      "source": [
        "kernel_div_fn = lambda q,p,_: tfd.kl_divergence(q,p)/X_train.shape[0]"
      ],
      "execution_count": 52,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "nfek9brFAjmQ"
      },
      "source": [
        "### 3.3 Build Bayesian model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "NyU05iVhCQmI"
      },
      "source": [
        "# Sequential model, similar structure with the Non-Bayesian model, but no dropout, since VI(minimizing KL is used)\n",
        "myBNN = keras.models.Sequential([\n",
        "                                 tfp.layers.Convolution2DFlipout(filters=32,kernel_size=(3,3),padding='valid',activation='relu',kernel_divergence_fn=kernel_div_fn),\n",
        "                                 layers.MaxPooling2D(pool_size=(2,2),padding='valid',strides=None),\n",
        "                                 layers.Flatten(),\n",
        "                                 tfp.layers.DenseFlipout(100,activation='relu',kernel_divergence_fn=kernel_div_fn),\n",
        "                                 tfp.layers.DenseFlipout(10,activation='softmax',kernel_divergence_fn=kernel_div_fn) # 10 classes\n",
        "\n",
        "])"
      ],
      "execution_count": 53,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "NAP8HTluQmuz"
      },
      "source": [
        "### 3.4 Run Bayesian model\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "NGIfYOGj0ZzG"
      },
      "source": [
        "# same as Non-Bayesian model\n",
        "loss='sparse_categorical_crossentropy'\n",
        "myBNN.compile(\n",
        "      optimizer=keras.optimizers.Adam(learning_rate=1e-3),\n",
        "      loss=loss,\n",
        "      metrics=[tf.keras.metrics.SparseCategoricalCrossentropy(name='Cat_Etp'),'accuracy'],\n",
        "  )"
      ],
      "execution_count": 54,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "TaZlcsH00gtl",
        "outputId": "78aea5d7-476f-4691-da83-bdb991e1e56f"
      },
      "source": [
        "early_stop = tf.keras.callbacks.EarlyStopping(monitor='val_accuracy',mode = 'max', patience=5)\n",
        "print(\"training the model...\")\n",
        "BNNhistory = myBNN.fit(X_train,y_train,validation_split=0.1,epochs=200,verbose=1,callbacks=[early_stop])"
      ],
      "execution_count": 55,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "training the model...\n",
            "Epoch 1/200\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/tensorflow/python/keras/engine/base_layer.py:2281: UserWarning: `layer.add_variable` is deprecated and will be removed in a future version. Please use `layer.add_weight` method instead.\n",
            "  warnings.warn('`layer.add_variable` is deprecated and '\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "1688/1688 [==============================] - 10s 5ms/step - loss: 21.2851 - Cat_Etp: 0.5423 - accuracy: 0.8233 - val_loss: 14.5478 - val_Cat_Etp: 0.0894 - val_accuracy: 0.9738\n",
            "Epoch 2/200\n",
            "1688/1688 [==============================] - 8s 5ms/step - loss: 12.7546 - Cat_Etp: 0.0971 - accuracy: 0.9702 - val_loss: 8.2657 - val_Cat_Etp: 0.0715 - val_accuracy: 0.9783\n",
            "Epoch 3/200\n",
            "1688/1688 [==============================] - 8s 5ms/step - loss: 7.3176 - Cat_Etp: 0.0729 - accuracy: 0.9769 - val_loss: 5.1667 - val_Cat_Etp: 0.0763 - val_accuracy: 0.9773\n",
            "Epoch 4/200\n",
            "1688/1688 [==============================] - 8s 5ms/step - loss: 4.6989 - Cat_Etp: 0.0663 - accuracy: 0.9800 - val_loss: 3.5855 - val_Cat_Etp: 0.0790 - val_accuracy: 0.9773\n",
            "Epoch 5/200\n",
            "1688/1688 [==============================] - 8s 5ms/step - loss: 3.3023 - Cat_Etp: 0.0625 - accuracy: 0.9783 - val_loss: 2.6467 - val_Cat_Etp: 0.0786 - val_accuracy: 0.9788\n",
            "Epoch 6/200\n",
            "1688/1688 [==============================] - 8s 5ms/step - loss: 2.4643 - Cat_Etp: 0.0625 - accuracy: 0.9797 - val_loss: 2.0414 - val_Cat_Etp: 0.0778 - val_accuracy: 0.9775\n",
            "Epoch 7/200\n",
            "1688/1688 [==============================] - 8s 5ms/step - loss: 1.9063 - Cat_Etp: 0.0615 - accuracy: 0.9799 - val_loss: 1.6016 - val_Cat_Etp: 0.0766 - val_accuracy: 0.9778\n",
            "Epoch 8/200\n",
            "1688/1688 [==============================] - 8s 5ms/step - loss: 1.4999 - Cat_Etp: 0.0626 - accuracy: 0.9792 - val_loss: 1.2862 - val_Cat_Etp: 0.0750 - val_accuracy: 0.9793\n",
            "Epoch 9/200\n",
            "1688/1688 [==============================] - 8s 5ms/step - loss: 1.2099 - Cat_Etp: 0.0608 - accuracy: 0.9816 - val_loss: 1.0576 - val_Cat_Etp: 0.0734 - val_accuracy: 0.9792\n",
            "Epoch 10/200\n",
            "1688/1688 [==============================] - 8s 5ms/step - loss: 1.0022 - Cat_Etp: 0.0622 - accuracy: 0.9796 - val_loss: 0.8987 - val_Cat_Etp: 0.0773 - val_accuracy: 0.9777\n",
            "Epoch 11/200\n",
            "1688/1688 [==============================] - 8s 5ms/step - loss: 0.8533 - Cat_Etp: 0.0649 - accuracy: 0.9797 - val_loss: 0.7850 - val_Cat_Etp: 0.0857 - val_accuracy: 0.9775\n",
            "Epoch 12/200\n",
            "1688/1688 [==============================] - 8s 5ms/step - loss: 0.7353 - Cat_Etp: 0.0613 - accuracy: 0.9807 - val_loss: 0.6848 - val_Cat_Etp: 0.0807 - val_accuracy: 0.9783\n",
            "Epoch 13/200\n",
            "1688/1688 [==============================] - 8s 5ms/step - loss: 0.6472 - Cat_Etp: 0.0633 - accuracy: 0.9793 - val_loss: 0.6068 - val_Cat_Etp: 0.0782 - val_accuracy: 0.9777\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "BO5JVzASzg80",
        "outputId": "a27e363b-f41d-4a77-de9b-5986d9971dd7"
      },
      "source": [
        "print(\"evaluating model on test data\")\n",
        "loss_tt, Cat_Etp, acc_tt = myBNN.evaluate(X_test, y_test)\n",
        "print(f\"Testing loss: {loss_tt:.3f}, testing acc: {acc_tt:.3f}\")"
      ],
      "execution_count": 56,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "evaluating model on test data\n",
            "313/313 [==============================] - 1s 3ms/step - loss: 0.6111 - Cat_Etp: 0.0825 - accuracy: 0.9763\n",
            "Testing loss: 0.611, testing acc: 0.976\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 393
        },
        "id": "1MOuJkm-rB--",
        "outputId": "a72ebe26-772d-467d-bd2c-50005b851c16"
      },
      "source": [
        "pd.DataFrame(BNNhistory.history)[['loss','val_loss']].plot(figsize=(10,6))\n",
        "plt.title(\"loss change of BNN\")\n",
        "plt.legend()\n",
        "plt.show()"
      ],
      "execution_count": 57,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 720x432 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 393
        },
        "id": "sDh326GlrYLx",
        "outputId": "8a24061a-f651-4a1b-aa7c-83a4e8d59306"
      },
      "source": [
        "pd.DataFrame(BNNhistory.history)[['accuracy','val_accuracy']].plot(figsize=(10,6))\n",
        "plt.title(\"accuracy change of BNN\")\n",
        "plt.ylim(0.95,1)\n",
        "plt.legend()\n",
        "plt.show()"
      ],
      "execution_count": 58,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 720x432 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 393
        },
        "id": "WFOf9KfE5Yxv",
        "outputId": "f2d41e30-261c-4153-d99e-2c800964cef2"
      },
      "source": [
        "pd.DataFrame(BNNhistory.history)[['Cat_Etp','val_Cat_Etp']].plot(figsize=(10,6))\n",
        "plt.title(\"Cross Categorical Entropy change of BNN\")\n",
        "plt.legend()\n",
        "plt.show()"
      ],
      "execution_count": 59,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 720x432 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "wOp5kbBt4wQN"
      },
      "source": [
        "Note that every weight in the network is now sampled from a normal distribution. The normal distribution has two paramters and therefore we have almost doubled our paramters (we don't use a distribution for bias terms) in the network. "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "tUkpVHxJltl4",
        "outputId": "3db21730-c30e-4961-8788-3d31d5bc755d"
      },
      "source": [
        "myBNN.summary()"
      ],
      "execution_count": 60,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential_2\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "conv2d_flipout (Conv2DFlipou (None, 26, 26, 32)        608       \n",
            "_________________________________________________________________\n",
            "max_pooling2d_2 (MaxPooling2 (None, 13, 13, 32)        0         \n",
            "_________________________________________________________________\n",
            "flatten_2 (Flatten)          (None, 5408)              0         \n",
            "_________________________________________________________________\n",
            "dense_flipout (DenseFlipout) (None, 100)               1081700   \n",
            "_________________________________________________________________\n",
            "dense_flipout_1 (DenseFlipou (None, 10)                2010      \n",
            "=================================================================\n",
            "Total params: 1,084,318\n",
            "Trainable params: 1,084,318\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "dPnahRdttzp0"
      },
      "source": [
        "myBNN.save('Bayesian_CNN.h5')"
      ],
      "execution_count": 61,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "e1zEJAaODqnn"
      },
      "source": [
        "### 3.5 The prediction of Bayesian model\n",
        "\n",
        "we can see the distribution variates every time, so we run model.predict 50 time and calculate the mean and std"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ZDg_tmeEuFEj",
        "outputId": "afc3061e-e999-4e1b-9b6a-8bf75afae903"
      },
      "source": [
        "for i in range(5):\n",
        "  print(f\"iter {i}\")\n",
        "  print(f\"predicted label: {np.argmax(myBNN.predict(X_test)[0])}\")\n",
        "  print(f\"predicted mat: {myBNN.predict(X_test)[0]}\")"
      ],
      "execution_count": 62,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "iter 0\n",
            "predicted label: 7\n",
            "predicted mat: [5.2457055e-10 1.5984302e-08 4.6855840e-07 8.0754754e-09 1.9600828e-11\n",
            " 1.6784928e-11 2.1923887e-15 9.9999952e-01 7.4856704e-11 1.6420231e-08]\n",
            "iter 1\n",
            "predicted label: 7\n",
            "predicted mat: [2.85207108e-07 5.41888312e-06 3.69237358e-04 3.22879059e-04\n",
            " 4.42147474e-09 1.94733687e-08 1.04162567e-10 9.99289751e-01\n",
            " 1.12793805e-05 1.09590576e-06]\n",
            "iter 2\n",
            "predicted label: 7\n",
            "predicted mat: [3.9584656e-09 1.0939628e-07 3.8393546e-06 2.3624561e-06 6.6433181e-10\n",
            " 1.7833178e-10 4.5917068e-13 9.9999321e-01 5.5033091e-08 4.0211361e-07]\n",
            "iter 3\n",
            "predicted label: 7\n",
            "predicted mat: [3.4216588e-10 1.6334364e-07 2.0281982e-06 1.4054432e-07 1.7117481e-11\n",
            " 6.2784992e-11 5.9684625e-14 9.9999750e-01 8.8515799e-09 5.4865737e-08]\n",
            "iter 4\n",
            "predicted label: 7\n",
            "predicted mat: [1.5234047e-11 3.4100902e-09 3.1271443e-07 3.5005872e-08 3.4040735e-12\n",
            " 1.8767481e-12 3.3218990e-18 9.9999952e-01 1.5048332e-10 1.0674973e-07]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "hksKzLXA6pVk"
      },
      "source": [
        "#### 3.5.1 Run model.evaluate 50 times directly\n",
        "then calculate the mean and std fo accuracy, loss"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Hv2FNd3JnUX9"
      },
      "source": [
        "# myBNN= keras.models.load_model('/content/Bayesian_CNN.h5') # not working in Google Colab, reason unknown"
      ],
      "execution_count": 63,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "EvulBOuGQ_jd",
        "outputId": "0324c724-3cdb-4328-b3ac-518dff2ef5fc"
      },
      "source": [
        "acc_tt_esb=[]\n",
        "loss_tt_esb = []\n",
        "for i in tqdm(range(50)):\n",
        "  acc_tt_esb.append(myBNN.evaluate(X_test, y_test,verbose=0)[2])\n",
        "  loss_tt_esb.append(myBNN.evaluate(X_test, y_test,verbose=0)[0])\n",
        "\n",
        "acc_tt_esb = np.array(acc_tt_esb)\n",
        "acc_mean = np.mean(acc_tt_esb)\n",
        "acc_std = np.std(acc_tt_esb,ddof=1)\n",
        "\n",
        "loss_tt_esb = np.array(loss_tt_esb)\n",
        "loss_mean = np.mean(loss_tt_esb)\n",
        "loss_std = np.std(loss_tt_esb,ddof=1)\n",
        "\n",
        "print(f'the mean and std of predcion accuracy are: {acc_mean :.3f}, {acc_std :.6f}')\n",
        "print(f'the mean and std of predcion loss are: {loss_mean :.3f}, {loss_std :.6f}')\n"
      ],
      "execution_count": 64,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "100%|██████████| 50/50 [01:20<00:00,  1.62s/it]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "the mean and std of predcion accuracy are: 0.975, 0.001013\n",
            "the mean and std of predcion loss are: 0.612, 0.003421\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "dsbMG2EK62L0"
      },
      "source": [
        "### 3.6 Plot the result"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "S7GM5sJ16_A4"
      },
      "source": [
        "#### 3.6.1 Prepare data for each digit number"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 284
        },
        "id": "CXTvqOIXE5O8",
        "outputId": "c4646e4b-a7ac-4ccf-fe32-deaa27da1f06"
      },
      "source": [
        "plt.hist(y_train,bins=[0,1,2,3,4,5,6,7,8,9])\n",
        "plt.xticks([0,1,2,3,4,5,6,7,8,9])\n",
        "plt.title(\"Full training data class distribution\")\n",
        "plt.show()"
      ],
      "execution_count": 65,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "HTm6aWnSE0te"
      },
      "source": [
        ""
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "KHspGTe_dOfs"
      },
      "source": [
        "keys = ['idx_0','idx_1','idx_2','idx_3','idx_4','idx_5','idx_6','idx_7','idx_8','idx_9']\n",
        "cat_dic = {}\n",
        "for i in range(10):\n",
        "  cat_dic[keys[i]]=np.array(np.where(y_test == i)).reshape(-1)"
      ],
      "execution_count": 66,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "JrkrgMiQWpmt",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "5a501685-8671-4d26-d861-bffa649c5e6e"
      },
      "source": [
        "pred_loss=np.zeros((10,10))\n",
        "pred_cat_etp = np.zeros((10,10))\n",
        "cat_acc =np.zeros((10,10))\n",
        "\n",
        "for i in tqdm(range(10)):\n",
        "  for k in range(10):\n",
        "    pred_loss1,pred_cat_etp1, cat_acc1 = myBNN.evaluate(X_test[cat_dic[keys[k]]],y_test[np.where(y_test==k)],verbose=0)\n",
        "    pred_loss[i,k] = pred_loss1\n",
        "    pred_cat_etp[i,k]=pred_cat_etp1\n",
        "    cat_acc[i,k]=cat_acc1\n",
        "\n"
      ],
      "execution_count": 67,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "100%|██████████| 10/10 [00:11<00:00,  1.15s/it]\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "aq-_ifuVWqdO"
      },
      "source": [
        "cat_acc_mean = cat_acc.mean(axis=1)\n",
        "cat_acc__std = cat_acc.std(axis=1)"
      ],
      "execution_count": 68,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "40sbpvmt7IUp"
      },
      "source": [
        "#### 3.6.2 Visualize the mean and std for each digit"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 666
        },
        "id": "HzrRZPYye0R6",
        "outputId": "f4319ffe-5641-48f3-bda7-cadac1d9a8e5"
      },
      "source": [
        "plt.figure(figsize=(20,10))\n",
        "plt.suptitle(\"Prediction accuracy mean and std for each digit class\")\n",
        "x = np.linspace(0,2,10000)\n",
        "for i in range(10):\n",
        "  plt.plot(x, ss.norm.pdf(x,cat_acc_mean[i],cat_acc__std[i]),label=f'digit {i}')\n",
        "  plt.legend()\n",
        "  plt.xlim(0.9,1.05)\n",
        "  # plt.legend()\n",
        "plt.show()"
      ],
      "execution_count": 69,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x720 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "YDRIaB7n7cw_"
      },
      "source": [
        "#### 3.6.3 Visulize the histogram of prediction for each digit\n",
        "easy to see the wrong prediction label, then we can tell wether the wrong prediction makes sense"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3UIQ2PzzvVTX"
      },
      "source": [
        "pred_dic={}\n",
        "for k in cat_dic.keys():\n",
        "  pred_dic[k] = np.argmax(myBNN(X_test[cat_dic[k]]).numpy(),axis=1)"
      ],
      "execution_count": 70,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 649
        },
        "id": "Tu2Hd8SUoxew",
        "outputId": "7235b4e3-f23e-448b-f8fb-8bfb096e1160"
      },
      "source": [
        "plt.figure(figsize=(16, 10))\n",
        "plt.suptitle('Prediction histogram for every digit\\n',y=0.98)\n",
        "for i in range(10):\n",
        "  plt.subplot(2,5, i+1)\n",
        "  plt.hist(pred_dic[keys[i]],density=False,bins=[0,1,2,3,4,5,6,7,8,9])\n",
        "  # sns.distplot(pred_dic[keys[i]],bins=[0,1,2,3,4,5,6,7,8,9])\n",
        "  plt.title(f'\\n \\n digit {i}')\n",
        "  plt.xticks([0,1,2,3,4,5,6,7,8,9])\n",
        "  plt.tight_layout()\n",
        "\n",
        "plt.show()"
      ],
      "execution_count": 71,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1152x720 with 10 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "GI8u9ZyznGpl",
        "outputId": "5ebf7c3a-9e14-45a1-c9e1-e821a17e54a8"
      },
      "source": [
        "X_test[cat_dic[keys[0]]].shape"
      ],
      "execution_count": 72,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(980, 28, 28, 1)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 72
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "iT_cpEB6oiWM"
      },
      "source": [
        "## 4.0 Predction on unseen data\n",
        "we can see big standard deviation in prediction in Bayesian NN, while we only get one point estimation for non-Bayesian NN"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "iRFyJRfqorY4"
      },
      "source": [
        "# generate unseen data\n",
        "unseen_digit = np.zeros((1,28,28,1))\n",
        "loc = np.random.choice(28,(40,2))\n",
        "col = np.random.choice((150,255),40)\n",
        "for i in range(40):\n",
        "  unseen_digit[0,loc[i][0],loc[i][1],0]= col[i]"
      ],
      "execution_count": 78,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 269
        },
        "id": "RWZxFxfdpyEI",
        "outputId": "6ddacade-f319-48b5-e6c3-406dd74441d8"
      },
      "source": [
        "plt.imshow(unseen_digit.reshape(28,28),cmap='binary')\n",
        "plt.show()"
      ],
      "execution_count": 79,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "8q6LK89zsL6-"
      },
      "source": [
        "#### 4.1 Bayesian model\n",
        "the prediction shows large variation, meaning low confidence and high uncertainty"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "yrLGIuitqYot",
        "outputId": "828abf1c-9f01-4b69-f558-e4f873f878ee"
      },
      "source": [
        "unseen_pred = np.zeros((1000,1))\n",
        "for i in tqdm(range(1000)):\n",
        "  unseen_pred[i,0] = np.argmax(myBNN.predict(unseen_digit),axis=1)"
      ],
      "execution_count": 89,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "100%|██████████| 1000/1000 [00:31<00:00, 31.91it/s]\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 521
        },
        "id": "L_c2HdI0s_t9",
        "outputId": "1908a9b8-7be0-4555-d5c7-17f35e2de90e"
      },
      "source": [
        "plt.figure(figsize=(10,8))\n",
        "plt.hist(unseen_pred)\n",
        "plt.title('Bayesian NN prediction distribution of noise data')\n",
        "plt.ylabel(\"frequency\")\n",
        "plt.xlabel(\"digits\")\n",
        "plt.xticks([0,1,2,3,4,5,6,7,8,9])\n",
        "plt.show()"
      ],
      "execution_count": 96,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 720x576 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "8OYlP2hush0K"
      },
      "source": [
        "#### 4.2 Non-Bayesian model\n",
        "point estimate of 6, and even the softmax matric shows 100% for digit 6"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "c45NeQCnouFE"
      },
      "source": [
        "unseen_pred_nonBayesian = mymodel.predict(unseen_digit)[0]"
      ],
      "execution_count": 103,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "DdxV6lQGrTds",
        "outputId": "d945f77a-60fc-477b-c0cf-1e6607a9ccbd"
      },
      "source": [
        "unseen_pred_nonBayesian"
      ],
      "execution_count": 104,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([0., 0., 0., 0., 0., 0., 1., 0., 0., 0.], dtype=float32)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 104
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "3j4qSZQoDSXE"
      },
      "source": [
        "## 4.0 Probabilisitic Bayesian neural network\n",
        "Try to use tfp.layers.OneHotCategorical() to output distribution of prediction directly, but the model is not learning. <br/>\n",
        "Refecence at [Keras blog](https://keras.io/examples/keras_recipes/bayesian_neural_networks/), experiment 3"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "aVMmvKHzDYQk"
      },
      "source": [
        "# Sequential model, similar structure with the Non-Bayesian model, but no dropout, since VI(minimizing KL is used)\n",
        "myProbBNN = keras.models.Sequential([\n",
        "                                 tfp.layers.Convolution2DFlipout(filters=32,kernel_size=(3,3),padding='valid',activation='relu',kernel_divergence_fn=kernel_div_fn),\n",
        "                                 layers.MaxPooling2D(pool_size=(2,2),padding='valid',strides=None),\n",
        "                                 layers.Flatten(),\n",
        "                                 tfp.layers.DenseFlipout(100,activation='relu',kernel_divergence_fn=kernel_div_fn),\n",
        "                                 tfp.layers.DenseFlipout(tfp.layers.OneHotCategorical.params_size(10),kernel_divergence_fn=kernel_div_fn,activation='relu'),\n",
        "                                 tfp.layers.OneHotCategorical(10)\n",
        "])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "S_oDnu-1O6cN"
      },
      "source": [
        "def neg_log_lik(y_true, y_pred): \n",
        "  return - y_pred.log_prob(y_true)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6Ry1SrOeQQ_n"
      },
      "source": [
        "myProbBNN.compile(optimizer='Adam',loss=neg_log_lik,metrics=['accuracy'])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "PxqRuS93Qv7L",
        "outputId": "8e352baf-6831-4ba5-a51e-8cab9629a03d"
      },
      "source": [
        "probBNNHistory = myProbBNN.fit(X_train,y_train,validation_split=0.1,epochs=200,callbacks=[early_stop])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/200\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/tensorflow/python/keras/engine/base_layer.py:2281: UserWarning: `layer.add_variable` is deprecated and will be removed in a future version. Please use `layer.add_weight` method instead.\n",
            "  warnings.warn('`layer.add_variable` is deprecated and '\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "1688/1688 [==============================] - 11s 5ms/step - loss: 7688676.6963 - accuracy: 0.1006 - val_loss: 73151312.0000 - val_accuracy: 0.0975\n",
            "Epoch 2/200\n",
            "1688/1688 [==============================] - 9s 5ms/step - loss: 110135516.8028 - accuracy: 0.1002 - val_loss: 241581440.0000 - val_accuracy: 0.0993\n",
            "Epoch 3/200\n",
            "1688/1688 [==============================] - 9s 5ms/step - loss: 294189202.9272 - accuracy: 0.1041 - val_loss: 469680288.0000 - val_accuracy: 0.0972\n",
            "Epoch 4/200\n",
            "1688/1688 [==============================] - 9s 5ms/step - loss: 533735070.0675 - accuracy: 0.0975 - val_loss: 748003264.0000 - val_accuracy: 0.1037\n",
            "Epoch 5/200\n",
            "1688/1688 [==============================] - 9s 5ms/step - loss: 822682305.0989 - accuracy: 0.0984 - val_loss: 1075418880.0000 - val_accuracy: 0.1008\n",
            "Epoch 6/200\n",
            "1688/1688 [==============================] - 9s 5ms/step - loss: 1159162235.5287 - accuracy: 0.1027 - val_loss: 1448762496.0000 - val_accuracy: 0.0975\n",
            "Epoch 7/200\n",
            "1688/1688 [==============================] - 9s 5ms/step - loss: 1533400229.2860 - accuracy: 0.1014 - val_loss: 1867341696.0000 - val_accuracy: 0.1005\n",
            "Epoch 8/200\n",
            "1688/1688 [==============================] - 9s 5ms/step - loss: 1968256618.8561 - accuracy: 0.0979 - val_loss: 2327996160.0000 - val_accuracy: 0.1038\n",
            "Epoch 9/200\n",
            "1688/1688 [==============================] - 9s 5ms/step - loss: 2452655066.4109 - accuracy: 0.0971 - val_loss: 2838974720.0000 - val_accuracy: 0.1075\n",
            "Epoch 10/200\n",
            "1688/1688 [==============================] - 9s 5ms/step - loss: 2966537553.3925 - accuracy: 0.1011 - val_loss: 3386609920.0000 - val_accuracy: 0.1020\n",
            "Epoch 11/200\n",
            "1688/1688 [==============================] - 9s 5ms/step - loss: 3509317578.1930 - accuracy: 0.0976 - val_loss: 3976645888.0000 - val_accuracy: 0.0958\n",
            "Epoch 12/200\n",
            "1688/1688 [==============================] - 8s 5ms/step - loss: 4117622878.4275 - accuracy: 0.0979 - val_loss: 4605295616.0000 - val_accuracy: 0.1053\n",
            "Epoch 13/200\n",
            "1688/1688 [==============================] - 9s 5ms/step - loss: 4766083978.0793 - accuracy: 0.1027 - val_loss: 5278502400.0000 - val_accuracy: 0.1022\n",
            "Epoch 14/200\n",
            "1688/1688 [==============================] - 8s 5ms/step - loss: 5451888399.6116 - accuracy: 0.1014 - val_loss: 5990082560.0000 - val_accuracy: 0.0972\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "KQJUvL3LQ7eN"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}