Anomaly Detection- Anomaly MNIST nos.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "2cd0d948",
   "metadata": {},
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "import keras\n",
    "from keras.layers import Input, Dense, Conv2D, MaxPooling2D, UpSampling2D, Conv2DTranspose, Input, Flatten, Lambda, Reshape\n",
    "from keras.models import Model\n",
    "from keras.datasets import mnist\n",
    "import numpy as np\n",
    "from keras.callbacks import TensorBoard\n",
    "import matplotlib.pyplot as plt\n",
    "import random\n",
    "from keras import backend as K\n",
    "from tensorflow.keras.preprocessing.image import ImageDataGenerator\n",
    "from keras.models import Model\n",
    "import scipy.stats as stats"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "2351ee12",
   "metadata": {},
   "outputs": [],
   "source": [
    "(x_train, _), (x_test, _) = tf.keras.datasets.fashion_mnist.load_data()\n",
    "\n",
    "batch = 60\n",
    "\n",
    "train_samples = 10000\n",
    "test_samples = 1000\n",
    "#val_samples = int(train_samples * 0.2) \n",
    "\n",
    "x_train_list = list(x_train)\n",
    "x_train_small = random.sample(x_train_list, train_samples)\n",
    "x_train_small = np.array(x_train_small)\n",
    "\n",
    "#val_list_small = random.sample(x_train_list, val_samples)\n",
    "#val_list_small = np.array(val_list_small)\n",
    "\n",
    "x_test_list = list(x_test)\n",
    "x_test_small = random.sample(x_test_list, test_samples)\n",
    "x_test_small = np.array(x_test_small)\n",
    "\n",
    "x_train_small = x_train_small.astype('float32')\n",
    "x_test_small = x_test_small.astype('float32')\n",
    "#val_list_small = val_list_small.astype('float32')\n",
    "\n",
    "x_train_small = x_train_small / 255\n",
    "x_test_small = x_test_small / 255\n",
    "#val_list_small = val_list_small/255\n",
    "\n",
    "img_width  = x_train_small.shape[1]\n",
    "img_height = x_train_small.shape[2]\n",
    "num_channels = 1\n",
    "\n",
    "x_train_small = np.reshape(x_train_small, (len(x_train_small), img_width, img_height, num_channels))\n",
    "x_test_small = np.reshape(x_test_small, (len(x_test_small), img_width, img_height, num_channels))\n",
    "#val_list_small= np.reshape(val_list_small, (len(val_list_small), img_width, img_height, num_channels)) \n",
    "img_shape = (img_height, img_width, num_channels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "8d17f91e",
   "metadata": {},
   "outputs": [],
   "source": [
    "(x_anomaly_train, _),(x_anomaly_test, _) = mnist.load_data()\n",
    "anomaly_samples = 2000\n",
    "_\n",
    "anomaly_list = list(x_anomaly_train)\n",
    "anomaly_data = random.sample(anomaly_list, anomaly_samples)\n",
    "anomaly_data = np.array(anomaly_data)\n",
    "\n",
    "anomaly_data = anomaly_data.astype('float32')\n",
    "anomaly_data = anomaly_data / 255\n",
    "anomaly_data = np.reshape(anomaly_data, (len(anomaly_data), img_width, img_height, num_channels))\n",
    "\n",
    "\n",
    "anomaly_test_list = list(x_anomaly_test)\n",
    "anomaly_test_data = random.sample(anomaly_test_list, anomaly_samples)\n",
    "anomaly_test_data = np.array(anomaly_test_data)\n",
    "\n",
    "anomaly_test_data = anomaly_test_data.astype('float32')\n",
    "anomaly_test_data = anomaly_test_data / 255\n",
    "anomaly_test_data = np.reshape(anomaly_test_data, (len(anomaly_test_data), img_width, img_height, num_channels))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "069379f8",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'numpy.ndarray'>\n",
      "<class 'numpy.ndarray'>\n",
      "<class 'numpy.ndarray'>\n",
      "60000\n",
      "(60000, 28, 28)\n",
      "10000\n",
      "(10000, 28, 28)\n",
      "10000\n",
      "1000\n",
      "2000\n"
     ]
    }
   ],
   "source": [
    "print(type(x_train))\n",
    "print(type(x_train_small))\n",
    "print(type(anomaly_data))\n",
    "print(len(x_train))\n",
    "#print(len(val_list_small))\n",
    "print(x_train.shape)\n",
    "print(len(x_test))\n",
    "print(x_test.shape)\n",
    "print(len(x_train_small))\n",
    "print(len(x_test_small))\n",
    "print(len(anomaly_data))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "08b20d98",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[8576, 6386, 2145, 8880, 4652, 3434, 8794, 723, 4491, 9289]\n",
      "\n",
      "Few of the training data samples are:\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x288 with 10 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "n = []\n",
    "a = 0\n",
    "for j in range(0, 10):\n",
    "    n.append(random.randint(0, train_samples))\n",
    "print(n)\n",
    "\n",
    "print('\\nFew of the training data samples are:')\n",
    "plt.figure(figsize=(20, 4))\n",
    "for i in n:\n",
    "    a = a + 1\n",
    "    ax = plt.subplot(2, 10, a)\n",
    "    plt.imshow(x_train_small[i].reshape(28, 28))\n",
    "    plt.gray()\n",
    "    ax.get_xaxis().set_visible(False)\n",
    "    ax.get_yaxis().set_visible(False)\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "036f34a9",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[398, 132, 1541, 304, 200, 50, 1416, 956, 1655, 1008]\n",
      "\n",
      "Few of the Anomaly data samples are:\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x288 with 10 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "f = []\n",
    "b = 0\n",
    "for j in range(0, 10):\n",
    "    f.append(random.randint(0, len(anomaly_data)-1))\n",
    "print(f)\n",
    "\n",
    "print('\\nFew of the Anomaly data samples are:')\n",
    "plt.figure(figsize=(20, 4))\n",
    "             \n",
    "for i in f:\n",
    "    b = b + 1\n",
    "    ax = plt.subplot(2, 10, b)\n",
    "    plt.imshow(anomaly_data[i].reshape(28, 28))\n",
    "    plt.gray()\n",
    "    ax.get_xaxis().set_visible(False)\n",
    "    ax.get_yaxis().set_visible(False)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "id": "5209eb1b",
   "metadata": {},
   "outputs": [],
   "source": [
    "input_img = keras.Input(shape = img_shape)\n",
    "\n",
    "x = keras.layers.Conv2D(10, (3, 3), activation='relu', padding='same')(input_img)\n",
    "#x = keras.layers.MaxPooling2D((2, 2), padding='same')(x)\n",
    "#x = keras.layers.Conv2D(10, (3, 3), activation='relu', padding='same')(x)\n",
    "#x = keras.layers.MaxPooling2D((2, 2), padding='same')(x)\n",
    "#x = keras.layers.Conv2D(8, (3, 3), activation='relu', padding='same')(x)\n",
    "encoded = keras.layers.MaxPooling2D((2, 2), padding='same')(x)\n",
    "\n",
    "x = keras.layers.Conv2D(10, (3, 3), activation='relu', padding='same')(encoded)\n",
    "#x = keras.layers.UpSampling2D((2, 2))(x)\n",
    "#x = keras.layers.Conv2D(10, (3, 3), activation='relu', padding='same')(x)\n",
    "#x = keras.layers.UpSampling2D((2, 2))(x)\n",
    "#x = keras.layers.Conv2D(16, (3, 3), activation='relu')(x)\n",
    "x = keras.layers.UpSampling2D((2, 2))(x)\n",
    "decoded = keras.layers.Conv2D(1, (3, 3), activation='sigmoid', padding='same')(x)\n",
    "\n",
    "autoencoder = keras.Model(input_img, decoded)\n",
    "autoencoder.compile(optimizer='adam', loss='binary_crossentropy',metrics = ['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "id": "9598e4da",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model_7\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " input_8 (InputLayer)        [(None, 28, 28, 1)]       0         \n",
      "                                                                 \n",
      " conv2d_38 (Conv2D)          (None, 28, 28, 10)        100       \n",
      "                                                                 \n",
      " max_pooling2d_16 (MaxPoolin  (None, 14, 14, 10)       0         \n",
      " g2D)                                                            \n",
      "                                                                 \n",
      " conv2d_39 (Conv2D)          (None, 14, 14, 10)        910       \n",
      "                                                                 \n",
      " up_sampling2d_17 (UpSamplin  (None, 28, 28, 10)       0         \n",
      " g2D)                                                            \n",
      "                                                                 \n",
      " conv2d_40 (Conv2D)          (None, 28, 28, 1)         91        \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 1,101\n",
      "Trainable params: 1,101\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "autoencoder.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "id": "ceb98b27",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/25\n",
      "167/167 [==============================] - 3s 17ms/step - loss: 0.4197 - accuracy: 0.4987 - val_loss: 0.2892 - val_accuracy: 0.5197\n",
      "Epoch 2/25\n",
      "167/167 [==============================] - 3s 16ms/step - loss: 0.2872 - accuracy: 0.5090 - val_loss: 0.2743 - val_accuracy: 0.5200\n",
      "Epoch 3/25\n",
      "167/167 [==============================] - 3s 16ms/step - loss: 0.2768 - accuracy: 0.5094 - val_loss: 0.2665 - val_accuracy: 0.5205\n",
      "Epoch 4/25\n",
      "167/167 [==============================] - 3s 16ms/step - loss: 0.2706 - accuracy: 0.5097 - val_loss: 0.2619 - val_accuracy: 0.5206\n",
      "Epoch 5/25\n",
      "167/167 [==============================] - 3s 16ms/step - loss: 0.2671 - accuracy: 0.5098 - val_loss: 0.2593 - val_accuracy: 0.5207\n",
      "Epoch 6/25\n",
      "167/167 [==============================] - 3s 16ms/step - loss: 0.2651 - accuracy: 0.5098 - val_loss: 0.2580 - val_accuracy: 0.5208\n",
      "Epoch 7/25\n",
      "167/167 [==============================] - 3s 16ms/step - loss: 0.2636 - accuracy: 0.5099 - val_loss: 0.2564 - val_accuracy: 0.5208\n",
      "Epoch 8/25\n",
      "167/167 [==============================] - 3s 16ms/step - loss: 0.2625 - accuracy: 0.5099 - val_loss: 0.2554 - val_accuracy: 0.5209\n",
      "Epoch 9/25\n",
      "167/167 [==============================] - 3s 16ms/step - loss: 0.2616 - accuracy: 0.5099 - val_loss: 0.2547 - val_accuracy: 0.5209\n",
      "Epoch 10/25\n",
      "167/167 [==============================] - 3s 16ms/step - loss: 0.2608 - accuracy: 0.5099 - val_loss: 0.2537 - val_accuracy: 0.5209\n",
      "Epoch 11/25\n",
      "167/167 [==============================] - 3s 16ms/step - loss: 0.2602 - accuracy: 0.5099 - val_loss: 0.2531 - val_accuracy: 0.5209\n",
      "Epoch 12/25\n",
      "167/167 [==============================] - 3s 16ms/step - loss: 0.2596 - accuracy: 0.5100 - val_loss: 0.2527 - val_accuracy: 0.5209\n",
      "Epoch 13/25\n",
      "167/167 [==============================] - 3s 16ms/step - loss: 0.2591 - accuracy: 0.5100 - val_loss: 0.2521 - val_accuracy: 0.5209\n",
      "Epoch 14/25\n",
      "167/167 [==============================] - 3s 16ms/step - loss: 0.2586 - accuracy: 0.5100 - val_loss: 0.2519 - val_accuracy: 0.5209\n",
      "Epoch 15/25\n",
      "167/167 [==============================] - 3s 16ms/step - loss: 0.2583 - accuracy: 0.5100 - val_loss: 0.2513 - val_accuracy: 0.5210\n",
      "Epoch 16/25\n",
      "167/167 [==============================] - 3s 16ms/step - loss: 0.2579 - accuracy: 0.5100 - val_loss: 0.2510 - val_accuracy: 0.5210\n",
      "Epoch 17/25\n",
      "167/167 [==============================] - 3s 17ms/step - loss: 0.2575 - accuracy: 0.5100 - val_loss: 0.2507 - val_accuracy: 0.5210\n",
      "Epoch 18/25\n",
      "167/167 [==============================] - 3s 16ms/step - loss: 0.2573 - accuracy: 0.5100 - val_loss: 0.2503 - val_accuracy: 0.5210\n",
      "Epoch 19/25\n",
      "167/167 [==============================] - 3s 17ms/step - loss: 0.2569 - accuracy: 0.5100 - val_loss: 0.2500 - val_accuracy: 0.5210\n",
      "Epoch 20/25\n",
      "167/167 [==============================] - 3s 17ms/step - loss: 0.2567 - accuracy: 0.5100 - val_loss: 0.2499 - val_accuracy: 0.5210\n",
      "Epoch 21/25\n",
      "167/167 [==============================] - 3s 17ms/step - loss: 0.2564 - accuracy: 0.5100 - val_loss: 0.2495 - val_accuracy: 0.5210\n",
      "Epoch 22/25\n",
      "167/167 [==============================] - 3s 16ms/step - loss: 0.2562 - accuracy: 0.5100 - val_loss: 0.2493 - val_accuracy: 0.5210\n",
      "Epoch 23/25\n",
      "167/167 [==============================] - 3s 16ms/step - loss: 0.2559 - accuracy: 0.5100 - val_loss: 0.2490 - val_accuracy: 0.5210\n",
      "Epoch 24/25\n",
      "167/167 [==============================] - 3s 17ms/step - loss: 0.2556 - accuracy: 0.5100 - val_loss: 0.2487 - val_accuracy: 0.5210\n",
      "Epoch 25/25\n",
      "167/167 [==============================] - 3s 17ms/step - loss: 0.2554 - accuracy: 0.5100 - val_loss: 0.2485 - val_accuracy: 0.5210\n"
     ]
    }
   ],
   "source": [
    "a = autoencoder.fit(x_train_small, x_train_small,\n",
    "                epochs= 25,\n",
    "                batch_size=batch,\n",
    "                shuffle=True,\n",
    "                validation_data=(x_test_small, x_test_small))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "id": "01a3b62c",
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "\n",
    "decoded_imgs = autoencoder.predict(x_test_small)\n",
    "decoded_imgs_anom = autoencoder.predict(anomaly_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "id": "821c4395",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.rcParams[\"figure.figsize\"] = (10,6)\n",
    "loss = a.history['loss']\n",
    "val_loss = a.history['val_loss']\n",
    "epochs = range(1, len(loss) + 1)\n",
    "plt.plot(epochs, loss, '-o', label='Training loss')\n",
    "plt.plot(epochs, val_loss, '-o', label='Validation loss')\n",
    "plt.title('Training and train_samplesalidation loss')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('Loss')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "id": "77d09bed",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The orignal images are:\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x288 with 10 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "The reconstructed images are:\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x288 with 10 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "n = 10\n",
    "plt.figure(figsize=(20, 4))\n",
    "print(\"The orignal images are:\")\n",
    "for i in range(1, n + 1):\n",
    "    # Display original\n",
    "    ax = plt.subplot(2, n, i)\n",
    "    plt.imshow(x_test_small[i].reshape(28, 28))\n",
    "    plt.gray()\n",
    "    ax.get_xaxis().set_visible(False)\n",
    "    ax.get_yaxis().set_visible(False)\n",
    "plt.show()\n",
    "\n",
    "print(\"\\nThe reconstructed images are:\")\n",
    "plt.figure(figsize=(20, 4))\n",
    "for i in range(1, n + 1):\n",
    "    ax = plt.subplot(2, n, i + n)\n",
    "    plt.imshow(decoded_imgs[i].reshape(28, 28))\n",
    "    plt.gray()\n",
    "    ax.get_xaxis().set_visible(False)\n",
    "    ax.get_yaxis().set_visible(False)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "id": "cdb5d0d9",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:Model was constructed with shape (None, 28, 28, 1) for input KerasTensor(type_spec=TensorSpec(shape=(None, 28, 28, 1), dtype=tf.float32, name='input_8'), name='input_8', description=\"created by layer 'input_8'\"), but it was called on an input with incompatible shape (None, 28, 1, 1).\n",
      "1/1 [==============================] - 0s 89ms/step - loss: 0.2738 - accuracy: 0.3756\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.2618 - accuracy: 0.5587\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1839 - accuracy: 0.7022\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2482 - accuracy: 0.6518\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4385 - accuracy: 0.3546\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2197 - accuracy: 0.6798\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.2255 - accuracy: 0.6301\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2078 - accuracy: 0.4490\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2925 - accuracy: 0.4356\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3421 - accuracy: 0.4043\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4018 - accuracy: 0.3763\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.4424 - accuracy: 0.3712\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2008 - accuracy: 0.6199\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1916 - accuracy: 0.5778\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.1772 - accuracy: 0.6901\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.4427 - accuracy: 0.2915\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.1596 - accuracy: 0.6798\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1757 - accuracy: 0.6097\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2214 - accuracy: 0.5115\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2355 - accuracy: 0.5140\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2391 - accuracy: 0.5344\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4282 - accuracy: 0.3386\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.2868 - accuracy: 0.5249\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2529 - accuracy: 0.5778\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.2596 - accuracy: 0.5893\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.3976 - accuracy: 0.2557\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.3331 - accuracy: 0.3903\n",
      "1/1 [==============================] - 0s 17ms/step - loss: 0.3739 - accuracy: 0.2946\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.2250 - accuracy: 0.6658\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2776 - accuracy: 0.5242\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.2762 - accuracy: 0.4770\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.1683 - accuracy: 0.6888\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3255 - accuracy: 0.3214\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.1001 - accuracy: 0.8163\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2193 - accuracy: 0.5880\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2169 - accuracy: 0.6122\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1080 - accuracy: 0.7985\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2072 - accuracy: 0.6173\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.4022 - accuracy: 0.3635\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.3737 - accuracy: 0.4222\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.3069 - accuracy: 0.3763\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2337 - accuracy: 0.5242\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1404 - accuracy: 0.7296\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2390 - accuracy: 0.5682\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.2630 - accuracy: 0.5829\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3742 - accuracy: 0.3546\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.2915 - accuracy: 0.4349\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.3256 - accuracy: 0.4770\n",
      "1/1 [==============================] - 0s 20ms/step - loss: 0.3088 - accuracy: 0.4337\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.1819 - accuracy: 0.7149\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3596 - accuracy: 0.2679\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2980 - accuracy: 0.4860\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.3931 - accuracy: 0.3750\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1633 - accuracy: 0.7309\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.4236 - accuracy: 0.3214\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3928 - accuracy: 0.3068\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2483 - accuracy: 0.5631\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.3150 - accuracy: 0.4362\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.5052 - accuracy: 0.3202\n",
      "1/1 [==============================] - 0s 19ms/step - loss: 0.1648 - accuracy: 0.6760\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.3759 - accuracy: 0.3890\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.2337 - accuracy: 0.5651\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.2822 - accuracy: 0.4751\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2764 - accuracy: 0.5370\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.2805 - accuracy: 0.5172\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.3692 - accuracy: 0.3112\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.3493 - accuracy: 0.3788\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1922 - accuracy: 0.6741\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3316 - accuracy: 0.3814\n",
      "1/1 [==============================] - 0s 18ms/step - loss: 0.2079 - accuracy: 0.6875\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.3427 - accuracy: 0.4031\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2009 - accuracy: 0.5804\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1876 - accuracy: 0.6958\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.2824 - accuracy: 0.4745\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3109 - accuracy: 0.5147\n",
      "1/1 [==============================] - 0s 17ms/step - loss: 0.1651 - accuracy: 0.6811\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.2181 - accuracy: 0.5714\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.1484 - accuracy: 0.7219\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3642 - accuracy: 0.4005\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2379 - accuracy: 0.5708\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3908 - accuracy: 0.3571\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.4331 - accuracy: 0.3457\n",
      "1/1 [==============================] - 0s 17ms/step - loss: 0.3012 - accuracy: 0.4362\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1743 - accuracy: 0.7041\n",
      "1/1 [==============================] - 0s 17ms/step - loss: 0.3868 - accuracy: 0.3820\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2141 - accuracy: 0.6575\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3434 - accuracy: 0.3763\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1567 - accuracy: 0.7908\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1883 - accuracy: 0.7105\n",
      "1/1 [==============================] - 0s 18ms/step - loss: 0.2692 - accuracy: 0.5855\n",
      "1/1 [==============================] - 0s 17ms/step - loss: 0.3420 - accuracy: 0.3929\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2925 - accuracy: 0.5051\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1858 - accuracy: 0.6869\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2591 - accuracy: 0.4796\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.1450 - accuracy: 0.7436\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1379 - accuracy: 0.7704\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2257 - accuracy: 0.6148\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.3969 - accuracy: 0.3827\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2005 - accuracy: 0.6805\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.0659 - accuracy: 0.8724\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1010 - accuracy: 0.8036\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1370 - accuracy: 0.7283\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.1033 - accuracy: 0.7691\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1061 - accuracy: 0.7997\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1066 - accuracy: 0.7819\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1232 - accuracy: 0.7564\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.0895 - accuracy: 0.8457\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.0643 - accuracy: 0.8737\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.0876 - accuracy: 0.8227\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.0881 - accuracy: 0.8393\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1073 - accuracy: 0.7908\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.0932 - accuracy: 0.8431\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.0649 - accuracy: 0.8788\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.1018 - accuracy: 0.8099\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.0777 - accuracy: 0.8380\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.0888 - accuracy: 0.8112\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.0939 - accuracy: 0.8048\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.1001 - accuracy: 0.8036\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1113 - accuracy: 0.7360\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.0705 - accuracy: 0.8635\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1387 - accuracy: 0.6977\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.0978 - accuracy: 0.8138\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.0695 - accuracy: 0.8559\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1084 - accuracy: 0.7691\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.0883 - accuracy: 0.8163\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1180 - accuracy: 0.7372\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.0836 - accuracy: 0.8852\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.1206 - accuracy: 0.7385\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1193 - accuracy: 0.7551\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.0906 - accuracy: 0.8367\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1042 - accuracy: 0.7832\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1060 - accuracy: 0.9209\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1034 - accuracy: 0.8074\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1144 - accuracy: 0.7666\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1052 - accuracy: 0.7857\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1001 - accuracy: 0.7844\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.0979 - accuracy: 0.9177\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1097 - accuracy: 0.7844\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1101 - accuracy: 0.7844\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.0984 - accuracy: 0.8151\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1168 - accuracy: 0.7768\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.0730 - accuracy: 0.8520\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.0606 - accuracy: 0.8827\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.0883 - accuracy: 0.8304\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.0548 - accuracy: 0.9196\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.0850 - accuracy: 0.8304\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.0996 - accuracy: 0.8138\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.0678 - accuracy: 0.9503\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.1014 - accuracy: 0.8023\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.0968 - accuracy: 0.8189\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.0820 - accuracy: 0.8265\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1016 - accuracy: 0.7946\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1357 - accuracy: 0.7768\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.0765 - accuracy: 0.8291\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.0835 - accuracy: 0.8406\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.0825 - accuracy: 0.8801\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.1159 - accuracy: 0.7857\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.0892 - accuracy: 0.8457\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.0849 - accuracy: 0.8329\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.1083 - accuracy: 0.7589\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.0775 - accuracy: 0.8648\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.1139 - accuracy: 0.7730\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.0850 - accuracy: 0.8622\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.0984 - accuracy: 0.8061\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1081 - accuracy: 0.7704\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1103 - accuracy: 0.7704\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.0928 - accuracy: 0.8029\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.0909 - accuracy: 0.8265\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.1006 - accuracy: 0.8010\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1181 - accuracy: 0.7959\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.0657 - accuracy: 0.8763\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.0928 - accuracy: 0.8533\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.0955 - accuracy: 0.8087\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.0985 - accuracy: 0.8457\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1010 - accuracy: 0.8431\n",
      "1/1 [==============================] - 0s 17ms/step - loss: 0.0720 - accuracy: 0.8597\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.0760 - accuracy: 0.8457\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.0927 - accuracy: 0.8240\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.0765 - accuracy: 0.8495\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.0665 - accuracy: 0.8941\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1142 - accuracy: 0.7640\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.1237 - accuracy: 0.7309\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.0854 - accuracy: 0.8380\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.0600 - accuracy: 0.8839\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.0809 - accuracy: 0.8622\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.0973 - accuracy: 0.8265\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.0897 - accuracy: 0.8520\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.0998 - accuracy: 0.8265\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.0793 - accuracy: 0.8559\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.1215 - accuracy: 0.7538\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1216 - accuracy: 0.7500\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1122 - accuracy: 0.7793\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.0931 - accuracy: 0.9356\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.0924 - accuracy: 0.9311\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1025 - accuracy: 0.7781\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.1229 - accuracy: 0.7436\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.0921 - accuracy: 0.8380\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.1077 - accuracy: 0.7997\n"
     ]
    }
   ],
   "source": [
    "recon_error_list=[]\n",
    "\n",
    "for i in range(1, 100):\n",
    "    #print(x_test_small[i].shape,decoded_imgs[i].shape)\n",
    "    reconstruction_error = autoencoder.evaluate([decoded_imgs[i]],[x_test_small[i]])\n",
    "    recon_error_list.append(reconstruction_error)\n",
    "recon_error_list = np.array(recon_error_list)  \n",
    "#print(recon_error_list)\n",
    "\n",
    "ano_recon_error_list=[]\n",
    "\n",
    "for i in range(1, 100):\n",
    "    #print(x_test_small[i].shape,decoded_imgs[i].shape)\n",
    "    ano_recon_error = autoencoder.evaluate([decoded_imgs_anom[i]],[anomaly_data[i]])\n",
    "    ano_recon_error_list.append(ano_recon_error)\n",
    "ano_recon_error_list = np.array(ano_recon_error_list)\n",
    "\n",
    "def Extract(lst):\n",
    "    return [item[0] for item in lst]\n",
    "\n",
    "recon_error_list = Extract(recon_error_list)\n",
    "ano_recon_error_list = Extract(ano_recon_error_list)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "id": "5e8eeaf8",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAmEAAAGDCAYAAABjkcdfAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjUuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/YYfK9AAAACXBIWXMAAAsTAAALEwEAmpwYAABH/0lEQVR4nO3debgU5Zn38e8tnCCbcQETBRVxi6BsooIxLMEQE3CJSwwaRzRqNOPomKDRkVcNkk2ZxDDZRqMcjLtmc4xJQEXQRGYAY9yDRkVRI0cSRVxY7/ePKo5N031O9emurqqu3+e6+jqnq6urnq71rue56ylzd0RERESkvrZKugAiIiIieaQgTERERCQBCsJEREREEqAgTERERCQBCsJEREREEqAgTERERCQBCsJEJPXM7CQzm5N0OerFzNzM9gz//6mZ/b8aTXdXM1ttZp3C9w+Y2em1mHY4vd+Z2Sm1mp5IozP1EyaSDmb2IvARYAOwGvg9cI67r06yXKWYmQN7uftzMUy7H/AC0OTu62s9/SzoyPINt5/T3f3eCr7zAHCju/+sA2W8HNjT3b9Y6XdFJKCaMJF0OcLdewBDgKHAxckWp2PMrHPSZai1Ur+p0t+Z5HJpxHUiknUKwkRSyN3/DvyBIBgDwMxGmNmfzOxNM/uLmY0p+Gx7M5tlZq+a2T/N7NcFn51hZs+Z2T/M7C4z27ngMzezs8zs2fB7PzIzCz/b08zmm9lbZvaGmd0WDl8Qfv0vYdPWCWY2xsyWm9nXzezvwCwzm2xmDxX+rqJmtq5m9p9mtiycx0Nm1hXYNP03w+mPLJ6WmR1iZovC7y0ys0MKPnvAzK4wsz+a2dtmNsfMepVb1mY20cweDZfrn8xsUMFnL4a/6THgnXCZuJl9ycxeAu43s63MbGr4O1aY2Q1m9uHw+/2Kxy9ThgvM7LVw/Z1W9FmzmU0P/+9lZneHZf2HmT0Yzv/nwK7A/4TL7MJS8y4YVhiQ7WFm/xcuy9+Y2fbhvMaY2fKisrxoZoeZ2eHAfwAnhPP7S8GyPz38P8pyOcXMXgq3r0vKrSORRqUgTCSFzKwv8BngufB9H+C3wHRge2AK8Asz6x1+5edAN2AgsCPw/fB7nwS+DXwe2AlYBtxaNLuJwIHA4HC8T4fDrwDmANsBfYH/AnD3UeHng929h7vfFr7/aFi23YAzI/zMGcABwCHh9y4ENgKbpr9tOP2Hi5bN9uGymAnsAHwP+K2Z7VAw2onAqeGy+BDB8tqCmQ0Drge+HE7rv4G7zKxLwWiTgAnAtsCm5tHRwL4Ey2py+BoL9Ad6AD8smlXh+MVlODws36eAvYDDSpU19DVgOdCboOn6PwB395OBlwhrUt39yijzDv0LcBqwc/j7ZrYxfwhm+HvgW8Bt4fwGlxhtMu0vl0OBfYBxwKVmtm978xZpJArCRNLl12b2NvAysAK4LBz+ReAed7/H3Te6+1xgMfBZM9uJIGA7y93/6e7r3H1++L2TgOvd/RF3X0PQvDnSgryrTb7j7m+6+0vAPD6ofVtHEFDt7O7vu/tmtVolbAQuc/c17v5eWyOa2VYEJ/7z3P0Vd9/g7n8Ky9ieCcCz7v5zd1/v7rcAzwBHFIwzy92XhuW4veA3FTsD+G93/9+wDLOBNcCIgnFmuvvLRb/pcnd/Jxx2EvA9d38+zN+7GPhCUW1T4fjFPh+W9wl3fwe4vI3fvo4gmN4tXM8PevuJvW3NG+DnBfP+f8DnLUzcr1KU5fINd3/P3f8C/IXgQkAkNxSEiaTL0e7eExgDfAzY1Iy2G3B82Az1ppm9SVCLsBOwC/APd/9nientTFD7BUB4MlwJ9CkY5+8F/79LUGMBQc2UAf9nZk8WN5OV0OLu77f/E4Hgd20N/C3i+IU2+02hZUT7TcV2A75WtFx3Ceexycslvlc4rLg8y4DOBDVVbU2j8PuFnxf/tkJXEdSOzjGz583sojbGjTLv4s+XAU18sN1VI8pyibqeRBqSgjCRFAprspoJmuwgOFH+3N23LXh1d/fvhJ9tb2bblpjUqwSBBgBm1p2g2e2VCGX4u7uf4e47EzTX/XhTPle5rxS9f4egiXTTvD9a8NkbwPvAHhGmU2yz3xTalQi/qYSXgW8WLdduYe1aW+UpHFZcnl0JmvVeb2cam7xGEPgVfr8kd3/b3b/m7v0Jav6+ambj2plHe8uzeN7rCNZP8frrRNAMGnW6UZaLSK4pCBNJr6uBT5nZEOBG4Agz+7SZdTKzrcPE6b7u/hrwO4IgaTszazKzTXlVNwOnmtmQMM/pW8D/uvuL7c3czI4Pc9MA/klw0t0Qvn+dIM+nLX8BBobz3pqCZjZ330iQi/U9M9s5/E0jwzK2EDRtlpv+PcDeZnaimXU2sxOAAcDd7f2mEq4FzjKzgy3Q3cwmmFnPCqZxC3C+me1uZj34IFcqavcatwOTzWyAmXXjgyboLVhwE8GeZmbAKoL1Uck6KeWLBfOeBtzp7huApcDW4fJoAqYChblyrwP9wqblUqpdLiINT0GYSEq5ewtwA/D/3P1l4CiCROwWghqcC/hgHz6ZoAbjGYJcsn8Pp3EfQZ7PLwhqXPYAvhCxCAcC/2tmq4G7CPK3Xgg/uxyYHTbhfb5M+ZcSnNTvBZ4FinPKpgCPA4uAfwDfBbZy93eBbwJ/DKdfmJ+Fu68kuJngawRNqxcCE939jYi/q3Baiwnywn5IEGg+R5BMXonrCW6MWEDQv9n7wL9VUIbfEQTc94fzL3kHZWgvguW5GngY+LG7PxB+9m1garjMSt6IUMbPCWpd/07QRHxuWK63gK8APyOoZXyH4KaATe4I/640s0dKTLeq5SKSB+qsVURERCQBqgkTERERSYCCMBEREZEEKAgTERERSYCCMBEREZEEKAgTERERSUDn9kdJl169enm/fv2SLoaIiIhIu5YsWfKGu/cu9VnmgrB+/fqxePHipIshIiIi0i4zK/soMjVHioiIiCRAQZiIiIhIAhSEiYiIiCQgczlhIiLSGNatW8fy5ct5//33ky6KSNW23npr+vbtS1NTU+TvKAgTEZFELF++nJ49e9KvXz/MLOniiHSYu7Ny5UqWL1/O7rvvHvl7ao4UEZFEvP/+++ywww4KwCTzzIwddtih4lpdBWEiIpIYBWDSKDqyLSsIExERSdC3vvWtmk3r0Ucf5Z577ml9f9ddd/Gd73ynZtOX2lIQJiIiuefubNy4MZF5lwvCOlKm4iDsyCOP5KKLLqqqfJXYsGFDm+9LSXLZJ01BmIhIhq1bB+edF/zNg1r+3hdffJF9992Xr3zlKwwbNoyXX36Zq666igMPPJBBgwZx2WWXtY57ww03MGjQIAYPHszJJ58MwLJlyxg3bhyDBg1i3LhxvPTSSwBMnjyZc889l0MOOYT+/ftz5513AvDaa68xatQohgwZwn777ceDDz7IRRddxHvvvceQIUM46aSTSpbpyiuvZP/992fw4MGtAdWYMWNanx7zxhtv0K9fP9auXcull17KbbfdxpAhQ7jttttobm7mnHPO6VB5i914440cdNBBDBkyhC9/+cutAVaPHj249NJLOfjgg3n44Ye3eP+9732P/fbbj/3224+rr7667LLPJXfP1OuAAw5wEREJzJnjDu5z5yZdkso99dRTFX+nlr/3hRdecDPzhx9+2N3d//CHP/gZZ5zhGzdu9A0bNviECRN8/vz5/sQTT/jee+/tLS0t7u6+cuVKd3efOHGiNzc3u7v7dddd50cddZS7u59yyil+3HHH+YYNG/zJJ5/0PfbYw93dZ8yY4dOnT3d39/Xr1/uqVavc3b179+5ly3TPPff4yJEj/Z133tls3qNHj/ZFixa5u3tLS4vvtttu7u4+a9Ys/9d//dfW6RW+r7S8hZ566imfOHGir1271t3dzz77bJ89e7a7uwN+2223tY5b+H7x4sW+3377+erVq/3tt9/2AQMG+COPPLLF72wUpbZpYLGXiWlUEyYikkHNzdC3Lxx7LJjBMccE75ubky5ZPOL6vbvtthsjRowAYM6cOcyZM4ehQ4cybNgwnnnmGZ599lnuv/9+jjvuOHr16gXA9ttvD8DDDz/MiSeeCMDJJ5/MQw891Drdo48+mq222ooBAwbw+uuvA3DggQcya9YsLr/8ch5//HF69uzZbpnuvfdeTj31VLp167bZvDui0vIWuu+++1iyZAkHHnggQ4YM4b777uP5558HoFOnThx77LGt4xa+f+ihh/jc5z5H9+7d6dGjB8cccwwPPvjgFr8zr2INwszscDP7q5k9Z2ZbNEqb2Rgze8vMHg1fl8ZZHhGRRjFhAuy5J6xdC+7B3z33DIY3orh+b/fu3Vv/d3cuvvhiHn30UR599FGee+45vvSlL+Huke58KxynS5cum00XYNSoUSxYsIA+ffpw8sknc8MNN0QqU6l5d+7cuTWPqqOd3bZX3kLuzimnnNK6bP76179y+eWXA0EnpZ06dWodt/B9qWltUvg78yq2IMzMOgE/Aj4DDAAmmdmAEqM+6O5Dwte0uMojItJIeveGadOCYKR79yBHatq0YHgjqsfv/fSnP83111/P6tWrAXjllVdYsWIF48aN4/bbb2flypUA/OMf/wDgkEMO4dZbbwXgpptu4tBDD21z+suWLWPHHXfkjDPO4Etf+hKPPPIIAE1NTawrk+Q2fvx4rr/+et59993N5t2vXz+WLFkCsFkOV8+ePXn77bdLTqvS8hYaN24cd955JytWrGgtx7Jly9r93qhRo/j1r3/Nu+++yzvvvMOvfvUrPvGJT0Seb6OLsybsIOA5d3/e3dcCtwJHxTg/EZFcmTMH+vSBa6+FnXeGuXOTLlG84v6948eP58QTT2TkyJHsv//+HHfccbz99tsMHDiQSy65hNGjRzN48GC++tWvAjBz5kxmzZrFoEGD+PnPf84PfvCDNqf/wAMPMGTIEIYOHcovfvELzjvvPADOPPNMBg0axEknnbTFdw4//HCOPPJIhg8fzpAhQ5gxYwYAU6ZM4Sc/+QmHHHIIb7zxRuv4Y8eO5amnnmpNzC9UaXkLDRgwgOnTpzN+/HgGDRrEpz71KV577bV2vzds2DAmT57MQQcdxMEHH8zpp5/O0KFDI8+30VlbVYVVTdjsOOBwdz89fH8ycLC7n1MwzhjgF8By4FVgirs/2dZ0hw8f7pvuCBERybNVq6CpCbp2hffeC2qHttkm6VJF9/TTT7PvvvtGHj/rv1caX6lt2syWuPvwUuPH+ezIUg3oxRHfI8Bu7r7azD4L/BrYa4sJmZ0JnAmw66671riYIiLZVBiAdO0avBpZ3n6vNL44myOXA7sUvO9LUNvVyt1Xufvq8P97gCYz61U8IXe/xt2Hu/vw3o2a8CAiIiK5EmcQtgjYy8x2N7MPAV8A7iocwcw+auHtGWZ2UFielTGWSURERCQVYmuOdPf1ZnYO8AegE3C9uz9pZmeFn/8UOA4428zWA+8BX/C4ktREREREUiTOnLBNTYz3FA37acH/PwR+GGcZRERERNJIPeaLiIiIJEBBmIiISMIKH8idFVdffXVrJ7LVevHFF7n55ptb3y9evJhzzz23JtNOMwVhIiIiGbZhw4ZE5ttWEFZpmYqDsOHDhzNz5syqyleJ4vJGLX+1y15BmIiIZMe6dXDeecHfGjj66KM54IADGDhwINdcc03r8B49enDJJZcwePBgRowY0fpQ62XLljFu3DgGDRrEuHHjeOmllwCYPHkyZ599NmPHjqV///7Mnz+f0047jX333ZfJkye3Tvfss89m+PDhDBw4kMsuu2yL8lx33XWcf/75re+vvfba1h76C/Xo0YNLL72Ugw8+mIcffpgbb7yRgw46iCFDhvDlL3+5NTj4/e9/z7Bhwxg8eDDjxo0DgkcOHX300QwaNIgRI0bw2GOPAXD55Zdz2mmnMWbMGPr3798aBL3zzjtMmDCBwYMHs99++3Hbbbcxc+ZMXn31VcaOHcvYsWNLlumGG25g0KBBDB48mJNPPrl1ORU+ZqlHjx4AXHTRRTz44IMMGTKE73//+zzwwANMnDixQ+UtNmfOHEaOHMmwYcM4/vjjWx9L1a9fP6ZNm8ahhx7KHXfcscX7W265hf3335/99tuPr3/962WXfVXcPVOvAw44wEVEJPueeuqpyr80Z447uM+dW5MyrFy50t3d3333XR84cKC/8cYb7u4O+F133eXu7hdccIFfccUV7u4+ceJEb25udnf36667zo866ih3dz/llFP8hBNO8I0bN/qvf/1r79mzpz/22GO+YcMGHzZsmP/5z3/ebH7r16/30aNH+1/+8hd3dx89erQvWrTIV69e7f379/e1a9e6u/vIkSP9scce26LcgN92223uHizHiRMntn7n7LPP9tmzZ/uKFSu8b9++/vzzz28273POOccvv/xyd3e/7777fPDgwe7uftlll/nIkSP9/fff95aWFt9+++197dq1fuedd/rpp5/eOu8333zT3d132203b2lpKVmmJ554wvfee+/WzzfN+5RTTvE77rij9Tvdu3d3d/d58+b5hAkTWocXvq+0vIVaWlr8E5/4hK9evdrd3b/zne/4N77xjdbyf/e7320dt/D9K6+84rvssouvWLHC161b52PHjvVf/epXW/zOYqW2aWCxl4lpVBMmIiLp19wMffvCsceCGRxzTPC+ubmqyc6cObO1tuvll1/m2WefBeBDH/pQa03MAQccwIsvvgjAww8/zIknngjAySefzEMPPdQ6rSOOOAIzY//99+cjH/kI+++/P1tttRUDBw5s/f7tt9/OsGHDGDp0KE8++SRPPfXUZuXp3r07n/zkJ7n77rt55plnWLduHfvvv/8W5e7UqRPHHnssAPfddx9LlizhwAMPZMiQIdx33308//zzLFy4kFGjRrH77rsDsP322wPw0EMPtdZMffKTn2TlypW89dZbAEyYMIEuXbrQq1cvdtxxR15//XX2339/7r33Xr7+9a/z4IMP8uEPf7jksiws0/33389xxx1Hr169Npt3R1Ra3kILFy7kqaee4uMf/zhDhgxh9uzZmz14/IQTTths/E3vFy1axJgxY+jduzedO3fmpJNOYsGCBVv8zmrF2kWFiIhITUyYEARcCxeCO6xdC3vuGQzvoAceeIB7772Xhx9+mG7dujFmzBjef/99AJqamgj7EqdTp06sX7++5DQ2jQPQpUsXALbaaqvW/ze9X79+PS+88AIzZsxg0aJFbLfddkyePLl1foVOP/10vvWtb/Gxj32MU089teR8t956azp16gQELVqnnHIK3/72tzcb56677tqsfJt4ie44N41XWO5Nv3vvvfdmyZIl3HPPPVx88cWMHz+eSy+9tN0ylZp3586d2bhxY+s4a9euLfn7qilv8Xc/9alPccstt5Scdvfu3Uu+LzXPTQp/Z7VUEyYiIunXuzdMmxYEX927Bzlh06YFwzvorbfeYrvttqNbt24888wzLFy4sN3vHHLIIdx6660A3HTTTRx66KGR57dq1Sq6d+/Ohz/8YV5//XV+97vflRzv4IMP5uWXX+bmm29m0qRJ7U533Lhx3HnnnaxYsQIIcqiWLVvGyJEjmT9/Pi+88ELrcIBRo0Zx0003AUEg2qtXL7Zp40nor776Kt26deOLX/wiU6ZM4ZFHHgGgZ8+evP3222XLdPvtt7Ny5crN5t2vXz+WLFkCwG9+8xvWhbl9bU2r0vIWGjFiBH/84x957rnnAHj33XdZunRpu987+OCDmT9/Pm+88QYbNmzglltuYfTo0ZHmWQnVhImISDbMmQN9+sCVV8KFF8LcuTBqVIcnd/jhh/PTn/6UQYMGsc8++zBixIh2vzNz5kxOO+00rrrqKnr37s2sWbMiz2/w4MEMHTqUgQMH0r9/fz7+8Y+XHffzn/88jz76KNttt1270x0wYADTp09n/PjxbNy4kaamJn70ox8xYsQIrrnmGo455hg2btzIjjvuyNy5c7n88ss59dRTGTRoEN26dWP27NltTv/xxx/nggsuYKuttqKpqYmf/OQnAJx55pl85jOfYaeddmLevHmbfWfgwIFccskljB49mk6dOjF06FCam5s544wzOOqoozjooIMYN25ca83ToEGD6Ny5M4MHD2by5MkMHTq0dVqVlrdQ7969aW5uZtKkSaxZswaA6dOns/fee7f5vZ122olvf/vbjB07Fnfns5/9LEcddVTk+UZlbVW5pdHw4cM9a32piIjIlp5++mn23Xff6F9YtQqamqBrV3jvvaA2LGKNSNZMnDiR888/v/WORsmGUtu0mS1x9+GlxldzpIiIZMM22wQBGAR/GzAAe/PNN9l7773p2rWrArAcUHOkiIhISmy77baRcpakMagmTERERCQBCsJERCQxWctLFimnI9uygjAREUnE1ltvzcqVKxWISea5OytXrmTrrbeu6HvKCRMRkUT07duX5cuX09LSknRRRKq29dZb07dv34q+oyBMREQS0dTU1PpIHZE8UnOkiIiISAIUhImIiIgkQEGYiIiISAIUhImIiIgkQEGYiIiISAIUhImIiIgkQEGYiIiISAIUhImIiIgkQEGYiIiISAIUhImIiIgkQEGYiIiISAIUhImIiIgkQEGYiIiISAIUhOXUunVw3nnB3zROT0REpNEpCMupBx6AmTNh/vx0Tk9ERKTRKQirgzTVEjU3Q9++cOyxYAbHHBO8b25Ox/RERETyQkFYHaSplmjCBNhzT1i7FtyDv3vuGQxPw/RERETyQkFYjNJYS9S7N0ybFgRL3bsHtXPTpgXD0zA9ERGRvFAQFqO01hLNmQN9+sC118LOO8PcuemanoiISB6YuyddhooMHz7cFy9enHQxIluwAMaMgW7d4L33YN48GDUq2TKtWgVNTdC1a1Cmdetgm23SMz0REQmOpVOmwIwZwTFWssnMlrj78FKfqSYsZmmsJdpmmyBgguBvtQFTracnIiLpyieWeKgmLGaqJRIRkUo0N8PUqcH5Y/Vq6NEjOG9Mnw6TJyddOqmUasISpFqi+KWpCxARkWqlNZ9Yak9BmGSequxFpJHorvP8UBAmmZXGLkBERGohjfnEUnsKwiSzVGUvWaOmc4nqwgth6VKYNCn4e8EFSZdI4qAgTDJLVfaSNWo6l6iUT5wPCsIk01RlL1mgpnMRKUVdVEimqQsQyYKWFjj+eFi4ENasgS5dYMQIuOMO1dyKNDp1USENS1X2kgVqOheRUhSEiYjUgZrORaSYmiNFROpATeci+dRWc2TnehdGRCSPCgOurl0/aEYXkfxSc6SIiIhIAhSEiYiIiCRAQZiIiIhIAhSEiYiIiCRAQZiIiIhIAhSEiYiIiCRAQZiItGndOjjvvOCviIjUTqxBmJkdbmZ/NbPnzOyiNsY70Mw2mNlxcZYnj3QClWo98ADMnAnz5yddEhGRxhJbEGZmnYAfAZ8BBgCTzGxAmfG+C/whrrLkmU6g0lHNzdC3Lxx7LJjBMccE75ubky6ZiEhjiLMm7CDgOXd/3t3XArcCR5UY79+AXwArYixL7ugEWh3VIMKECbDnnsFDp92Dv3vuGQwXEZHqxRmE9QFeLni/PBzWysz6AJ8DfhpjOXJJJ9DqqAYReveGadOCbad79yAgnTYtGC4ikiVpvbCOMwizEsOKnxZ+NfB1d9/Q5oTMzjSzxWa2uKWlpVbla2g6gXaMahA3N2cO9OkD114LO+8Mc+cmXSIRkcql9cI6ziBsObBLwfu+wKtF4wwHbjWzF4HjgB+b2dHFE3L3a9x9uLsP760oIjKdQCunGsTNXXghLF0KkyYFfy+4IOkSpVM1V9lpvUIXaQRpv7A29+LKqRpN2KwzsBQYB7wCLAJOdPcny4zfDNzt7ne2Nd3hw4f74sWLa1zaxrRqFTQ1Qdeu8N57wUF+m22SLlX6LVgAY8ZAt27Bcps3D0aNSrpUkmZz58L48cHfww6r33dFpG0tLXD88bBwIaxZA126wIgRcMcd9WsZMrMl7j681Gex1YS5+3rgHIK7Hp8Gbnf3J83sLDM7K675yge22SYIwCD4qwAsGtUgSlTVXGWn/Qpd6ks1ovFIe2pOrP2Eufs97r63u+/h7t8Mh/3U3bdIxHf3ye3VgonUg5rgJKpqmq/V9C2F0pqz1AjSfGEdW3NkXNQcKSJpUk3ztZq+pbkZpk4N0kdWr4YePYJWi+nTYfLkjk933TqYMgVmzAjSUvIs6dScRJojRUTyoJqr7DRfoUt9xFUjqpq1D6Q5NUc1YSIiVajmKjvpK3RJh1rWiMZVsyYdp5owEZGYVHOVneYrdKmfWtaIKtcwW1QTJiIikqBa14gq1zBdVBMmIiKSUrWuEVWuYXaoJkxERKSBKNcwXdqqCetc78KIiIhIfAoDrq5dP6hlk/RRc6SISA6oR3aR9FEQJlKFvJ7YkvzdeV3m1VK/USLpoyBMpAp5PbEl+bvzusw7Ss+oFEkvBWEiHZDXE1uSvzuvy7xa6jdKJL0UhIl0QF5PbEn+7rwu82r17g3TpgXLq3v3oBl32rRgeBzUXCwSnYIwkQ6o94ktLZL83Xld5rVQz36j1FwsEp2CMJEOymuHiEn+7rwu82pdeCEsXQqTJgV/L7ig9vNQc7FI5dRZq0gH5bVDxCR/d16XeRa0tMDxx8PChbBmDXTpAiNGwB135Le2ct06mDIFZswItlvJJz22SCQGeX34cpK/O6/LPAvUXLwlNc1KexSEiYhITai5OKCmWYlKQZiISAm6y69y9cg9ywLdyStRKQgTESlBTUmVU3NxQE2zEpWCMBGRAmpKklpQ06xEobsjRUQK6C4/qQXdySub6O5IEZGI1JQktaCmWYlCQZiISBE1JYlIPag5UkSkiJqSRKRWOtwcaWZbmdnn4ymWiEg6qSmpttTdh0hpbQZh7r4ROKdOZRERkQak7j5ESouSEzbXzKaY2S5mtv2mV+wlExGRTFN3HyJtixKEnQb8K7AAWBK+lJQlIiJtUs/xIm1rNwhz991LvPrXo3AiIpJdcXb3oTyz2tGyTE67QZiZNZnZuWZ2Z/g6x8ya6lE4qYx2JBFJm7i6+1CeWe3UelnqXBRdlObInwAHAD8OXweEwyRldFASkbSp9UO9lWdWO3EtS52LoosShB3o7qe4+/3h61TgwLgLJtHpoCQiaVXr7j6UZ1Y7tV6WOhdVLkoQtsHM9tj0xsz6AxviK5JUSgclEckLPVaqdmq9LHUuqlyUIGwKMM/MHjCz+cD9wNfiLZZUQgclEckTPVaqdmq5LHUuqlx7PeZ3AgYDewHnhq993H1eHcomFdBBSUTyotZ5ZnlW62Wpc1Fl2n12pJnNc/exdSpPu/TsyNL0rLv4rVsHU6bAjBnBshYRkc3pXLSlDj87MvQnM/uhmX3CzIZtetW4jFIlPesufrrjRyR+6t4g23QuqkyUIOwQYCAwDfjP8DUjzkKJpInu+BGpH13sSJ5EyQm7y93HFr0+WafyiSROd/yIxE8XO5JHbQZh7r4BOLJOZRFJJd3xIxI/XexIHiknTCSCpO/4UZ5Mx2i5ZYcudiSPlBMmEkHSt8QrT6ZjtNyyJemLHZF6a7eLirRRFxWSJ83NMHVqcNv36tXQo0dwt9H06TB5ctKlSy8tt2xS9wbSiKrqosLMPmJm15nZ78L3A8zsS7UupIhsSXkyHaPllk3q3kDyJkpzZDPwB2Dn8P1S4N9jKo+IFFCeTMdouYlIFkQJwnq5++3ARgB3X48e4C1SN8qT6RgtNxFJuyiPLXoAOBaY6+7DzGwE8F13H12H8m1BOWGSN8qT6RgtNxFJg2ofW/RV4C5gDzP7I3AD8G81LJ+ItEF5Mh3TiMtNXW6INJZ2gzB3fwQYTdBVxZeBge7+WNwFExGRzanLDZHGEqUmDHdf7+5PuvsT7q5rMBGROtIjfepDNY1Sb5GCMBERSY663KgP1TRKvSkIE5FcyWJth7rciJdqGiUpkYIwM+tjZoeY2ahNr7gLJiLZk4UAJ6u1HepyIz6qaZSkROmi4rvACcBTfNA/mLv7kTGXrSR1USGSXnPnwvjxwd/DDku6NJvL+qOM1OVGvBYsgDFjoFu3YPnOmwejVN0gNVBtFxVHA/u4+2fd/YjwlUgAJiLplIXmnKzXdjRilxtpoppGSUKUmrDfAce7++r6FKltqgkTSZ+WFjj+eFi4ENasgS5dYMQIuOOOdOUtqbZDylFNo8Sl2pqwd4FHzey/zWzmplfEGR9uZn81s+fM7KISnx9lZo+Z2aNmttjMDo0yXRFJl6wkjqu2Q8rJSk1jFvIuJbooQdhdwBXAn4AlBa82mVkn4EfAZ4ABwCQzG1A02n3AYHcfApwG/CxyyUUkVbIQ4Fx4ISxdCpMmBX8vuCDpEolUJqs3lkhpUXrMnw3cwgfB183hsPYcBDzn7s+7+1rgVuCoommv9g/aQ7sDbbeNikhqZSHAqUVth2oiJAlZyLuUyrUbhJnZGOBZglqtHwNLI3ZR0Qd4ueD98nBY8fQ/Z2bPAL8lqA0TkQzKSnNOtVQTIUnI+o0lUlqU5sj/BMa7+2h3HwV8Gvh+hO9ZiWFb1HS5+6/c/WMEd2FeUXJCZmeGOWOLW1paIsxaRKS2VBMhScpK3qVUJkoQ1uTuf930xt2XAk0Rvrcc2KXgfV/g1XIju/sCYA8z61Xis2vcfbi7D++tLU5EEqCaCElaFvIupTJRgrDFZnadmY0JX9cSITEfWATsZWa7m9mHgC8QJPm3MrM9zczC/4cBHwJWVvYT8kl5KSKbi3ufUE2EJC0LeZdt0XlrS1GCsLOBJ4FzgfMIes4/q70vuft64BzgD8DTwO3u/qSZnWVmm75/LPCEmT1KkHN2grfXcZkAyksRKVaPfUI1EZKkrOdd6ry1pXY7a02bvHfWmvVHr4jUWj33CXXoKVK5vJ+3OtRZq5ndHv59POxQdbNXXIWVtikvJRpVe+dHPfeJrNdEiCRB563yytaEmdlO7v6ame1W6nN3XxZrycrIe00Y6NErUaT5QdJSe9onRNItz/toh2rC3P218N+vuPuywhfwlTgKKtEoL6U8dSOQT9onRNJN+2hpUR7g/Yi7Dysa9pi7D4q1ZGWoJkx5KW3JyoOkpba0T9TOunUwZQrMmBEs00aUh9+YNnneRzuaE3a2mT0OfKwoH+wF4PG4CivtU15KeepGIJ+0T9ROJXewZTX3Unfp1Z/20dLa6qLiZuAI4Dfh302vA9z9pDqUTaRDVO0tUrmONOVnLZhRuoKkTVs5YW+5+4vAD4B/FOSDrTOzg+tVQJFKpbVDw6zWGkg+VHIHW1aDGd2lJ2kTpbPWnwCrC96/Ew4TSaW0VnuXqjVQYCZp0V5TfuG2mtVgRukKkjZRgjAr7MXe3TcCneMrkkhjaavWIGvNOdLY2mrKL9xWsxzMKF1B0iTK3ZG/BB7gg9qvrwBj3f3oWEtWRj3ujtSdM1JLpe7Y3G234G6hd97JZw/Skk6l7mD75S9L93Y+bBj8+c9w5ZVBCsDkyXDFFUn/gvbl+S49SUaH7o4scBZwCPAKsBw4GDizdsVLH9VOSC2VqjW46irYZ5/sNedIYyvVlF+u6fHqq9OZe9metKYrSD61G4S5+wp3/4K77+juH3H3E919RT0KV29ZTTaV9CtuAlm0KLvNOZIv5Zoe+/dXMCNSrXaDMDObZWbXF7/qUbh6y2qyqaRfqTs2lZsiWaFtVcrRzUXViZITdmzB262BzwGvuvu5cRasnLhzwvL8fCupL+WmSFZoW5Vy9Jze9lWVE+buvyh43QR8Htiv1oVMC13xSb0oN0WyQtuqFFP6Tm1EScwvthewa60LkhZp7ehTREQkLZS+UxtRcsLeNrNVm17A/wBfj79oydAVn0h6KN9EJJ2y3FdcmrQZhJmZAQPdfZuC197u/os6lS81dDIQqT91FyOSXkrfqV6bQVjYU/6v6lSWVNPJoHEpwE4f5ZuIpF9c6Tt5OiZHyQlbaGYHxl6SlNLJoPEpwE4f5ZuIpF9c6Tt5OiZHCcLGAg+b2d/M7DEze9zMHou7YGmhk0HjUoCdXso3EcmfPB6TowRhnwH2AD4JHAFMDP/mgk4GjUsBdrop30QkX/J4TI4ShE1392WFL2B63AVLE50MGpMC7HRTdzEi+ZLHY3KUIGxg4Rsz6wQcEE9x0kkng8ZRnPCpADu91F2MSP7k7Zhc9rFFZnYx8B9AV+DdTYOBtcA17n5xXUpYJO7HFkljK37Ehh7HIiKSHo14TO7QY4vc/dvu3hO4qqCPsJ7uvkNSAZhIR5VL+PzlL1XbIiKSFnmrAY/SHHm3mXUHMLMvmtn3zGy3mMuVWXnq3yRL8pjwKdml44hIPkQJwn4CvGtmg4ELgWXADbGWKsPy1L9JluQx4VOyS8cRkXyIEoStD3vOPwr4gbv/AOgZb7GyJ4/9m2RN3hI+JXt0HBHJlyhB2Nthkv4Xgd+Gd0c2xVus7FFzV/rpLldJOx1HRPIlShB2ArAG+JK7/x3oA1wVa6kySM1d6Ze3hE/JHh1HRPKl3SDM3f/u7t9z9wfD9y+5u3LCSlBzl4hUS8cRkfwo209Y6whmxwDfBXYk6CfMAHf3ROoR0txPWCP2b5I269bBlCkwY0awrKVxaN0GdBxpbNrO86dD/YQVuBI40t0/XNBXmA4JJai5K366a6xxad0GdBxpbNrOpVCUIOx1d3869pKIFCnsK0l3jTUurVvJA23nUkqUIGyxmd1mZpPM7JhNr9hLJrlXeMWou8ayq72OR7Vuk6NOYetH27mUEiUI24bg2ZHjgSPC18Q4CyX5VuqKcehQGDlSd41lUXvNL7ojMDlqGqsfbedSSpS7I08t8TqtHoWTfCp3xbhmje4ay5JKml90R2B9qWksGdrOpViUuyP7Av8FfBxw4CHgPHdfHn/xtpTmuyOldhYsgDFjoFu34A6xefNgyBDdNZYlLS1w/PGwcGEQQHfpAiNGwB13bHn1rzsC66uSdSO1o+08n6q9O3IWcBewM0FHrf8TDhOJTakrRt01li2VNL9o3daXmsaSoe1cikUJwnq7+yx3Xx++mgHtqhIrPWKoMaj5Jb20bqQRZe1mkyjNkfcCzcAt4aBJwKnuPi7eopXW6M2R6sgv+7QOP6Dml/TSupFGNHcujB8f/D3ssKRLE6i2OfI04PPA34HXgOPCYRID3a2UfVqHH1DzS3pp3UgjyerNJlHujnzJ3Y90997uvqO7H+3uy+pRuDzJ6gZUK1mrQi4l7+tQRCQpWe2Hrd0gzMxmm9m2Be+3M7PrYy1VDhQHHVndgGqlEWqP8r4ORUSSktWbTaI0Rw5y9zc3vXH3fwJDYytRThQHHVndgKrVSLVHaVmHjVCrKJJm2sfSKYs3m0QJwrYys+02vTGz7YHO8RWpsbUVdGRxA6pWo9UepWEdlqpV1ElDpHYaoea+EWXxrvood0f+C3AxcCdBZ62fB77p7j+Pv3hbyvrdkW11ktilSz7vVirVMeuoUUmXqmOSvOOsuRmmTg3KsHo19OgRzHv69CAwTNsdQyJZ09Y+Nnly0qWTtKrq7kh3vwE4FngdaAGOSSoAawRtNVnl9W6lNNQe1UqS67BUrWL37nDxxY3R3CuStEaouVeteLpEaY4E2B54x93/C2gxs91jLFPDa6SgoxayWIWcRqUC/Kuugn32yfZJQyQt0pL3WQ01paZLlLsjLwO+TtAkCdAE3BhnoRqdgo7N5bUGMA7FAf6iRdk/aYikSVYvohvpJqhGEiUn7FGCuyEfcfeh4bDH3H1Q/MXbUtZzwkTiVCon7corYfbs4O+FFwa5K1dckXRJRbIpq08a0EPbk1Ntj/lrPYjUPJxY91oWTtqnNnyJqlStompeJevSdAzMas19IzSlNqIoQdjtZvbfwLZmdgZwL3BtvMWSQmrDl2pUctJI08lOZBMdA2sjq02pjazN5kgzM6Av8DFgPGDAH9w9sVWXp+ZI3Q4t9ZbGh99KfukYWFtZbUrNug43R4bNkL9297nufoG7T0kyAMubRrgdWrJBSbuSRjoG1lZWm1IbWZTmyIVmdmDsJZEtqA1f6kUnO0kjHQOl0UUJwsYCD5vZ38zsMTN73MweizJxMzvczP5qZs+Z2UUlPj8pnOZjZvYnMxtc6Q9odPVsw1c+UH7pZCdpVckxUMcwqUQatpcoXVTsVmq4uy9r53udgKXAp4DlwCJgkrs/VTDOIcDT7v5PM/sMcLm7H9zWdPOUEwb1bcNXPlC+TZ2qriwkfSo5BuoYJpWo1/bSVk5Yu0FYFTMdSRBUfTp8fzGAu3+7zPjbAU+4e5+2ppu3IKwelPwqoKRdyS4dw6QS9d5equ0nrKP6AC8XvF8eDivnS8DvSn1gZmea2WIzW9zS0lLDIgooH0gCStqVrNIxTCqRpu0lziDMSgwrWe1mZmMJgrCvl/rc3a9x9+HuPry3klRqTvlAIpJlOoZJJdK0vcQZhC0Hdil43xd4tXgkMxsE/Aw4yt1XxlgeaYM68RORLNMxTCqRlu0lzpywzgSJ+eOAVwgS80909ycLxtkVuB/4F3f/U5TpKicsHuXygdatgylTYMaM4HMRkVqo9bFFOY1SiXpuL4nkhLn7euAc4A/A08Dt7v6kmZ1lZmeFo10K7AD82MweNTNFVwkplw+kx4WISBxqfWxRTqNUIi3bS5zNkbj7Pe6+t7vv4e7fDIf91N1/Gv5/urtv5+5DwlfJSFHqTz2oi0gcdGzJrjT0q9VoYg3CJLvSdPeIiDSOeh9bFDjUjlpGak9BmJSUprtHRKRx1PvYosChenHWXuY9SFYQVoVG33jScveIiDSWehxb1OxZO3HWXuY9SI7t7si4pOnuyEZ/RIbuNpK80F3A9VWPY0tLCxx/PCxcCGvWQJcuMGIE3HGHavQ7YsECGDMGunUL1tm8eTBqVMenl6enHCTVY37DyssVVlruHhGJW96vxuutHscWpVTUVq1rL5V3HFAQ1gHaeEQaQ14uqPJKKRW1c+GFsHQpTJoU/L3gguqmpyA5oCCsA/K+8TR6Lpzkhy6oGlutA4c8i6P2UkGygrAOy/PGU23TjYI4SYu8X1A1OqVUpJuCZAVhHZbHjadWTTfKv5E0yfMFlTSGrF7YKkhWENZhedx4qm26Uf6NpFEeL6iksejCNrsUhElk1TbdKP9G0iiPF1TSGHRhm30KwqQi1TTdKP9GRKR2dGGbfQrCpCLVNt0o/6Z9Wc3vEJH60oVt9ikIk4pU23Sj/Jv2Kb9DRKLShW22KQiTulL+TXnK7xBJn7TXTOvCNtsUhImkhPI7pFbSHjhkSdprpnVhm20KwkRSQvkdjSXJQCjtgUMWqGZa6kFBmEiKKL+jcSQRCClwqB3VTEs9KAiT2KhJpHLK78i+JAMhBQ61o5ppqQcFYRIbNYlUTvkd2ZdkIKTAobZUMy1xUxAmNacmEcmzpAMhBQ61k+eaabVk1IeCMKk5NYlI3iUZCOU5cKi1PNdMqyWjPszdky5DRYYPH+6LFy9OuhjSjgULYMwY6NYN3nsP5s2DUaOSLpVIfaxaBU1NwYn7vfeC2oQ8ncAlu5qbYerUYBtevRp69Ai23enTYfLkpEuXTWa2xN2Hl/pMNWESCzWJSJ7luQZFsk0tGfWlmjCJhWoCRESySS0ZtaWaMKk71QSISJ40UiK7WjLqRzVhIiIiVZo7F8aPD/4edljSpamOWjJqSzVhIiIiMWjELnnUklE/CsJEREQ6SIns9dFIzb2FFISJiIh0UNKd8+ZFo/ZbpiBMRESkCkpkj08jNvcWUhAmInXXqE0Lkk96SkF8Gr25V0GYZJ/O6JnTqE0Lkk9KZI9PW829jXDoVxBWjUbYAhqBzuiZ0ehNCyJSe+Waexvh0K8grBqNsAVkmc7omdPoTQub0UWaSE0UN/fuvHPjHPoVhHWETv7pkKszemPI1Z1kukgTqYni5t7jjmucQ7+CsI7QyT8dcnVGbxwNfyeZLtJEYtVIh34FYR3RSFtA1jX8Gb3xNPydZLpIE4ldoxz69ezIjpo6FWbPhiuvDM4qkyfDFVckXar80UPOJI0WLIAxY6Bbt2C7nDcPRo1KulQiDSNLh349OzIODX85nxG6N1zSqFEu00VSqlEO/aoJExGptSxdpotIrNqqCetc78KIiDS8woCra9cPLtlFRAqoOTJt1LeQiIiUo3NE2zK2fBSEpY36FhIRkXKycI4oFwjVI0DKwvIpoCAsLdS3UOPI2JWYiGRAls4R5QKhOAOkLC2fAgrC0kJ9C6Vf1OCq3IFGwZlI7eRtf8rCOaJcIHT66fEHSFlYPiUoCEsLdQCbfu1dxbV3JZaxavKaydvJUuojb/tTFs4R5QKhr30t/gApC8unBAVhaaK+hdIpajV3uQPQ229nspq8ZvJ2soyTAtrMNjvVRNrPEeUCoX33rU+AlPblU4q7Z+p1wAEHeMN66y33d98N/n/33eC9JG/FCvfRo927dHGH4O/o0cHwYvPnu5u5d+/uvtVWwftKvl8La9e6n3tu8DdJs2a59+nj3rNnsEx69gzez5qVbLmybM6cYBuaOze+eaRl+ymn3vtTkorXRRbOEZdc4t63r/vNNwd/p05te3gtpXT5AIu9TEyTeFBV6auhgzBJr1LBVSnlDjRRv18L9ThRR5Gnk2Xc6hnQpmX7aUs996ckZWFdFCsXCJUbnvagvwbaCsLUHCnplaaml6jV3OUeZ1WPavK0NdNkNEcjleqRdBzn9lPrfTmLzU6VSNu+XIlyzxMqNzzv6QrlorO0vlQTliNpugqstpq7HtXkcdU8lbpSjXr1Wo8miLyIu/YnzprLWu/LKW12qpk81CLnKF0BNUfKFtJcBVzvnTPNy6JSFZyoI//sUifQqCfVRj9Z1lO1AW2UFV6LQK9wPjk60dZcoze5piHQrNOxX0FY0qqpSYhLmmqZitV750zzstgkhpqndn92qRPottsGL51U66/agDbKdl6LmsvC+aThRJtVeahFTjrQrNOxX0FY0qqpSai1rFyZ1mPnTOuyKBVwVVvzVDDNyD+71Al05MjgpZNqdlSynVcT6JWbz0UXNXaNTlzyUIucVKBZ52O/grB6aq8qPumahKxcmdZj50zrsigMuNo6WFRSm1owzYp+dqlgOOmrV6lMvbbzcvM5//zGr9GRjkkq0KzzsV9BWD21VxWfhpqEak+i9WhKrdfOmaaAolTA9dGPuu+zT+ntJUrtWJkg7pmLZkX72aWC4UZsJql2m046vaA99drOS80nDzU6koxq9rs6HvvbCsJi7aLCzA43s7+a2XNmdlGJzz9mZg+b2RozmxJnWarW3i3WpW4pHjoURo7c/Bb973wneCV52361t3fHcUtx8fItdztzraXpVvdS3RDssw9ceeXm28vIkTB0KH7MsTiGf66N29fLdG3wi/cnsNvO61hy6HnsutO68j+7VJcb5brhSKNqn/cZVdpvs6/Xdl5qPrXYl5PsriZNXeUkLW3Lopr9Li3H/nLRWbUvoBPwN6A/8CHgL8CAonF2BA4EvglMiTLdxGrC2qt1qKQqPumahI5emcbZjp5UjlzartJLXZ0Vby/nn+8+erSvbwq2tfVN7dSmlpjmW2+5v/8/wTJ//+65tfvZaasRam+7qnabTmteYbG4tvN69eher+NDNfmYeZCWZVGL/a6Ox36SaI4ERgJ/KHh/MXBxmXEvT20QVsnKjloVn7YTf1RxtKPHdRJLWzAQVakAvWh7ufHHb/kxveb7BszfprtvYCs/ptf88ouseJpHHFGbPLNSsnaQbmubjrIs0ppXWC9xr+96B7lR8zEbTXvbetqWRcb2u6SCsOOAnxW8Pxn4YZlx2wzCgDOBxcDiXXfdNbYFVVIlKzvpGq56qHU7elw7U1qCgUpFCNBXrHCfvesl/hJ9/Qvc7C/R15t3nVp+kRVP829/K7/MO7rcsnyQLrdNR10WacorrJd6re96dUBcaT5mo+loS0+SyyJD+11SQdjxJYKw/yozbnprwtyjr+ys1nBVohaBZvEBsJY7U9qCgZg8dM9b3pV3vXt39272rj90T4XbWvEyv+ii6pZblg/SldQURvl+I158Favn+o7jZFscdJT7Pb/5TWZO9B1SbUtPkjK037UVhMWZmL8c2KXgfV/g1RjnF5+oCXz1SiaPS5Sky1okZRcnU9YyQbIez9hLgd/9cRt26NuVa6+F7ft05fd/qnBbK17ma9ZUt9zS+JzIjj7v8+qrK1sWWbpRoVZqsb6jJnlXe3wonE+5ZzL+9relf8///V/t5p1GlRwv65nIXmq5FQ+rdL9L67ooF51V+wI6A88Du/NBYv7AMuNeTpprwvJQw+WeXH7Hj39c2+Wbtiu2tnQwB6vqTbLUBKpdbmm7Mq1mIWVpG0pKteu73PGm1sn+UXvwj5CPGctTCpKWxpaeODo4T3BdkFQ/YcBngaUEd0leEg47Czgr/P+jBDVmq4A3w/+3aWuaqe8nLA5xJ5lnPb+jWNqCgbak6SBd7XJrpIuVLG1DSd2EEtdd1rXaJyrtwb+W22+W0iLKbetJbFdxdHCegnWRWBAWxyuXQVjUg1JHn1FZr+RX9/rUMNTiYNoogW8l0hhEZS3ASEKaAvkoyh1vZs6s7T6RZA/+acyRLKfctp7EdhVHB+cpWBcKwrKq0hN1NVW49Uh+dc9ODUPcB6AUHBgyIWsBRj2lMZCPqtTxJo59Iske/LPapJ30dhXHo9ISXhdtBWGx9pgvVYqaNFkq2XS77YJXcQJqqV7VobZJl+WSX5ub05/E3FbZaymNiexp0pH1kNbE27i0dXyIY1nUcpqljjdx7BNx9eDf0XnHpZbrJumbm0ott2qXZVp6xy+lXHSW1leuasLco0XwtajCreXVYZZreSotezXNZVmpFUxCR7ahrNaaxfH8uziWRS2nWe54U+t9Islm5aQT2auRZM1RHB2cJ5xegJojMyzqQSmOKtxqZLUq3r2ysldz8MtS3lESoq6Hejef1DpPrdQ2FHUe1fZxFkU9l6/2icrEtW50gVhTCsKyLOpBqdROk+SOlOWdOErZk86byIOo21C9a17jvntv1qzo86jkaQgdleWa7UbXkXUTJcBXMFxTCsLyIG3PqMzyThyl7Doxxa+SbageNa+1DrxLbUN77x08LqeaecSxLLJcs93oKl03tW66zOpzeuuorSBMifmNolSyaT178C9ODM3y0wOilF2J9fGrZBuqR+JtrROWS21DV10F++xT3TwqWRb16rVe4hN13cR101HxE1CkMuWis7S+VBOWUllNiq5GlptcG01Wuh0orjUotQ1VO49KlkVHmz2zVLNdL1FrhGpdcxR13dS69l4pGZGh5kiJTZ53RJ2Y8qfWj+optQ3VI7jP834bl6gBbZIXrLVsVlZKRmQKwiQ+2hGlI7KaRxLXo3pqMY9KaL+tnajrNg2Bb60DfOUKRtJWEKacMKmOcqOkI7KaR9LRXMdK8snqkU+p/bY6hbl0Uddt0p2gQu07y1auYNUUhEn1stQzdN56VU+bej2RIG3SGPToBNpxhRcRUddtGraBWgf4aX8CSgYoCJPq1XNHrLYGJas1MI0iDbUBSQXiaQt6dAKtXLmLiBkzoq3btG0D1cryXfApYUFzZXYMHz7cFy9enHQxpN6am2HqVFi1Clavhh49gh1++nSYPDn+70vtLFgAY8ZAt27w3nswbx6MGlW/+c+dC+PHB38PO6x+8121CpqagpPVe+8FQaBOWtnS0gLHHw8LF8KaNdClC4wYAddfDzvt1P66bWsbWLcOpkwJArqmpvr+LomVmS1x9+GlPlNNWClqsvpAWpZFtTUoaaiBkUBStQFJN4Wq1iB+cR+vyjUp9u8fbd22tQ2olj6XFISVkoWdoV7BUVqWRbX5FGnIx5BAUs1gCsQbXz2OV7W+iEj64kASpSCsUJZ2hrgPNmlcFtUe/BotHyOrkqoRUiDeuOp5vKr1RYQuDnJNQVihLOwM9TrYpHFZVHvwUyJyutWjdleBeGOq5/Gq1hcRujjINQVhhbKwM9TrYJPGZVHtwU85OelWj6YkBeKNKY3Hq0rEcXGQlnxeaZOCsGJpv1Ku58Em7ctCGkM9m5IUiDeuLB+v4rg4SEs+r7RJXVQUy8Jt5FOnwuzZcOWVwc47eTJccUXt55OFZSHZV+62/zvuyE5NhiRPx6uAuuNJnba6qFAQlkU62EijqaTvMPWnJFKeLmpSR/2ENRo1qUijqaQpSc0sIuVlPT8uZxSEiUjyouTEpLHbFJE0ynJ+XM6oOVJEskHNLCLRKGUlVdQcKSLZp2YWkWiUspIZCsJEJDvUzCIiDUTNkSKSHWpmEZGMaas5snO9CyMi0mGFAVfXrh80uYiIZJCaI0VEREQSoCBMREREJAEKwkREREQSoCBMREREJAEKwkREREQSoCBMREREJAEKwkREREQSoCBMREREJAEKwkREREQSoCBMREREJAGZe3akmbUAy2o4yV7AGzWcntSW1k96ad2km9ZPemndpFut189u7t671AeZC8JqzcwWl3uwpiRP6ye9tG7STesnvbRu0q2e60fNkSIiIiIJUBAmIiIikgAFYXBN0gWQNmn9pJfWTbpp/aSX1k261W395D4nTERERCQJqgkTERERSUCugzAzO9zM/mpmz5nZRUmXJ8/MbBczm2dmT5vZk2Z2Xjh8ezOba2bPhn+3S7qseWVmnczsz2Z2d/he6yYlzGxbM7vTzJ4J96GRWj/pYWbnh8e1J8zsFjPbWusnOWZ2vZmtMLMnCoaVXR9mdnEYJ/zVzD5dy7LkNggzs07Aj4DPAAOASWY2INlS5dp64Gvuvi8wAvjXcH1cBNzn7nsB94XvJRnnAU8XvNe6SY8fAL93948BgwnWk9ZPCphZH+BcYLi77wd0Ar6A1k+SmoHDi4aVXB/heegLwMDwOz8O44eayG0QBhwEPOfuz7v7WuBW4KiEy5Rb7v6auz8S/v82wUmkD8E6mR2ONhs4OpEC5pyZ9QUmAD8rGKx1kwJmtg0wCrgOwN3XuvubaP2kSWegq5l1BroBr6L1kxh3XwD8o2hwufVxFHCru69x9xeA5wjih5rIcxDWB3i54P3ycJgkzMz6AUOB/wU+4u6vQRCoATsmWLQ8uxq4ENhYMEzrJh36Ay3ArLC5+Gdm1h2tn1Rw91eAGcBLwGvAW+4+B62ftCm3PmKNFfIchFmJYbpVNGFm1gP4BfDv7r4q6fIImNlEYIW7L0m6LFJSZ2AY8BN3Hwq8g5q2UiPMLToK2B3YGehuZl9MtlRSgVhjhTwHYcuBXQre9yWoIpaEmFkTQQB2k7v/Mhz8upntFH6+E7AiqfLl2MeBI83sRYJm+0+a2Y1o3aTFcmC5u/9v+P5OgqBM6ycdDgNecPcWd18H/BI4BK2ftCm3PmKNFfIchC0C9jKz3c3sQwSJd3clXKbcMjMjyGl52t2/V/DRXcAp4f+nAL+pd9nyzt0vdve+7t6PYD+5392/iNZNKrj734GXzWyfcNA44Cm0ftLiJWCEmXULj3PjCHJetX7Spdz6uAv4gpl1MbPdgb2A/6vVTHPdWauZfZYg16UTcL27fzPZEuWXmR0KPAg8zgd5R/9BkBd2O7ArwcHseHcvTqiUOjGzMcAUd59oZjugdZMKZjaE4KaJDwHPA6cSXGRr/aSAmX0DOIHgLvA/A6cDPdD6SYSZ3QKMAXoBrwOXAb+mzPows0uA0wjW37+7++9qVpY8B2EiIiIiSclzc6SIiIhIYhSEiYiIiCRAQZiIiIhIAhSEiYiIiCRAQZiIiIhIAhSEiUgqmdkGM3vUzJ40s7+Y2VfNrM1jlpn1M7MT61C2n4UP9m1rnKPbG0dE8k1BmIik1XvuPsTdBwKfAj5L0J9PW/oBsQdh7n66uz/VzmhHAwrCRKQsBWEiknruvgI4EzjHAv3M7EEzeyR8HRKO+h3gE2EN2vltjNcqHOcZM5ttZo+Z2Z1m1i38bFz4UOzHzex6M+sSDn/AzIaH/682s2+GtXULzewj4XyOBK4Ky7KHmZ1rZk+F87i1HstNRNJNnbWKSCqZ2Wp371E07J/Ax4C3gY3u/r6Z7QXc4u7DC3v0D8fvVmq8omn2A14ADnX3P5rZ9QSP/fkh8Cwwzt2XmtkNwCPufrWZPRDOZ7GZOXCku/+PmV0JrHL36WbWDNzt7neG83kV2N3d15jZtu7+Zs0XmohkimrCRCRLLPzbBFxrZo8Dd1C+2S/qeC+7+x/D/28EDgX2IXjw8tJw+GxgVInvrgXuDv9fQtAkWspjwE1m9kWCx5+ISM4pCBORTDCz/sAGYAVwPsEz3wYDwwmemVhK1PGKmwScDwK+9qzzD5oUNgCdy4w3AfgRcACwxMzKjSciOaEgTERSz8x6Az8FfhgGPB8GXnP3jcDJQKdw1LeBngVfLTdesV3NbGT4/yTgIeAZoJ+Z7RkOPxmYX0GxW8sS3tW5i7vPAy4EtiV4gLOI5JiCMBFJq66buqgA7gXmAN8IP/sxcIqZLQT2Bt4Jhz8GrA+T5M9vY7xiT4fjPQZsD/zE3d8HTgXuCJszNxIEglHdClxgZn8G9gJuDKfzZ+D7ygkTESXmi0iuhYn5d7v7fkmXRUTyRTVhIiIiIglQTZiIiIhIAlQTJiIiIpIABWEiIiIiCVAQJiIiIpIABWEiIiIiCVAQJiIiIpIABWEiIiIiCfj/Sy+leoQLK3AAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "x = list(range(1,len(recon_error_list)+1,1))\n",
    "y = list(range(1,len(ano_recon_error_list)+1,1))\n",
    "\n",
    "plt.scatter(x, recon_error_list, label= \"reconstrcution error\", color= \"blue\", marker= \"*\", s=30)\n",
    "plt.scatter(y, ano_recon_error_list, label= \"anomaly reconstrcution error\", color= \"red\", marker= \"*\", s=30)\n",
    "plt.xlabel('Data points')\n",
    "plt.ylabel('reconstruction error')\n",
    "plt.title('Reconstruction error distribution')\n",
    "plt.legend()\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "id": "54de716c",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The distribution of the reconstruction error is:\n",
      "\n",
      "The mean of the reconstruction error is 0.2742 \n",
      "The standard deviation of the reconstruction error is 0.0890\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "print(\"The distribution of the reconstruction error is:\")\n",
    "\n",
    "re = recon_error_list\n",
    "re.sort()\n",
    "re_mean = np.mean(re)\n",
    "re_std = np.std(re)\n",
    "\n",
    "print(\"\\nThe mean of the reconstruction error is\",format(re_mean,\".4f\"),\n",
    "      \"\\nThe standard deviation of the reconstruction error is\",format(re_std,\".4f\"))\n",
    "\n",
    "pdf = stats.norm.pdf(re, re_mean, re_std)\n",
    "plt.plot(re, pdf)\n",
    "plt.xlabel('reconstruction error')\n",
    "plt.title('Reconstruction error distribution')\n",
    "#plt.xlim(0.3, 0.7)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "id": "4fb4dbdf",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The threshold values are as follows: 0.3632 , 0.1852\n",
      "\n",
      "The thresholds can be visualised below:\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "threshold = [re_mean + (1*re_std) ,re_mean - (1*re_std)]\n",
    "print(\"The threshold values are as follows:\",format(threshold[0],\".4f\"),\",\",format(threshold[1],\".4f\"))\n",
    "\n",
    "print(\"\\nThe thresholds can be visualised below:\")\n",
    "\n",
    "re = recon_error_list\n",
    "re.sort()\n",
    "re_mean = np.mean(re)\n",
    "re_std = np.std(re)\n",
    "pdf = stats.norm.pdf(re, re_mean, re_std)\n",
    "plt.plot(re, pdf)\n",
    "#plt.plot(threshold[0], 0, marker=\"o\", markersize=10, markerfacecolor=\"red\")\n",
    "#plt.plot(threshold[1], 0, marker=\"o\", markersize=10, markerfacecolor=\"red\")\n",
    "plt.axvline(x = threshold[0], color = 'red', linestyle='dotted')\n",
    "plt.axvline(x = threshold[1], color = 'red', linestyle='dotted')\n",
    "plt.xlabel('reconstruction error')\n",
    "plt.title('Threshold')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "id": "bb0851d2",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 16ms/step - loss: 0.4105 - accuracy: 0.8138\n",
      "1/1 [==============================] - 0s 17ms/step - loss: 0.6949 - accuracy: 0.7321\n",
      "1/1 [==============================] - 0s 18ms/step - loss: 0.9456 - accuracy: 0.6677\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7089 - accuracy: 0.6849\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3678 - accuracy: 0.7309\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.6115 - accuracy: 0.7315\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3910 - accuracy: 0.7385\n",
      "1/1 [==============================] - 0s 19ms/step - loss: 0.5209 - accuracy: 0.7468\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2993 - accuracy: 0.8163\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.6010 - accuracy: 0.7717\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4579 - accuracy: 0.7730\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5783 - accuracy: 0.7589\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4461 - accuracy: 0.7474\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.3446 - accuracy: 0.8202\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.6053 - accuracy: 0.7041\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.5247 - accuracy: 0.7800\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4162 - accuracy: 0.7462\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3002 - accuracy: 0.7755\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5341 - accuracy: 0.7372\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7956 - accuracy: 0.6958\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.4815 - accuracy: 0.7755\n",
      "1/1 [==============================] - 0s 17ms/step - loss: 1.1379 - accuracy: 0.6901\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.5946 - accuracy: 0.7404\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.4977 - accuracy: 0.8316\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.6263 - accuracy: 0.7315\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5279 - accuracy: 0.7423\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.6538 - accuracy: 0.7092\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3360 - accuracy: 0.8425\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.8339 - accuracy: 0.6773\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.6699 - accuracy: 0.7417\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.6040 - accuracy: 0.7895\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.6116 - accuracy: 0.6894\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7773 - accuracy: 0.7851\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.6016 - accuracy: 0.7736\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.7553 - accuracy: 0.6996\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.6473 - accuracy: 0.7226\n",
      "1/1 [==============================] - 0s 17ms/step - loss: 0.5391 - accuracy: 0.7398\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.6758 - accuracy: 0.7832\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.4960 - accuracy: 0.7258\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.6458 - accuracy: 0.7526\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4931 - accuracy: 0.7915\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5191 - accuracy: 0.7226\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.3635 - accuracy: 0.8017\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.3046 - accuracy: 0.8157\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.4888 - accuracy: 0.7009\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.3643 - accuracy: 0.7921\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5539 - accuracy: 0.7851\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4301 - accuracy: 0.8042\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4830 - accuracy: 0.8291\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.5307 - accuracy: 0.7685\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.5012 - accuracy: 0.7876\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2711 - accuracy: 0.8023\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.3630 - accuracy: 0.7742\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5046 - accuracy: 0.7615\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4600 - accuracy: 0.7972\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2644 - accuracy: 0.8093\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3618 - accuracy: 0.7908\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.2706 - accuracy: 0.7366\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5289 - accuracy: 0.7659\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3661 - accuracy: 0.7966\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.5199 - accuracy: 0.7226\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4324 - accuracy: 0.7908\n",
      "1/1 [==============================] - 0s 19ms/step - loss: 0.6445 - accuracy: 0.7264\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.5095 - accuracy: 0.7679\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5553 - accuracy: 0.7857\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.8106 - accuracy: 0.6849\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4726 - accuracy: 0.7181\n",
      "1/1 [==============================] - 0s 17ms/step - loss: 0.6036 - accuracy: 0.6562\n",
      "1/1 [==============================] - 0s 17ms/step - loss: 0.3757 - accuracy: 0.8048\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4275 - accuracy: 0.7710\n",
      "1/1 [==============================] - 0s 18ms/step - loss: 0.5362 - accuracy: 0.7423\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.4196 - accuracy: 0.8546\n",
      "1/1 [==============================] - 0s 17ms/step - loss: 0.4160 - accuracy: 0.7659\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.4055 - accuracy: 0.8048\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.4424 - accuracy: 0.7506\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3986 - accuracy: 0.8004\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.4617 - accuracy: 0.7876\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4326 - accuracy: 0.7749\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7388 - accuracy: 0.7366\n",
      "1/1 [==============================] - 0s 17ms/step - loss: 0.5398 - accuracy: 0.7934\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2727 - accuracy: 0.8253\n",
      "1/1 [==============================] - 0s 17ms/step - loss: 0.6004 - accuracy: 0.6741\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 1.0730 - accuracy: 0.6926\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3920 - accuracy: 0.7793\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4439 - accuracy: 0.7806\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.4272 - accuracy: 0.7934\n",
      "1/1 [==============================] - 0s 17ms/step - loss: 0.3563 - accuracy: 0.7953\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.4124 - accuracy: 0.8259\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4057 - accuracy: 0.7290\n",
      "1/1 [==============================] - 0s 20ms/step - loss: 0.3392 - accuracy: 0.8310\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5831 - accuracy: 0.6779\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4387 - accuracy: 0.6575\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.7657 - accuracy: 0.7449\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.8257 - accuracy: 0.7787\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5362 - accuracy: 0.7812\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.6378 - accuracy: 0.7111\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.7009 - accuracy: 0.6550\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5893 - accuracy: 0.7073\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.3394 - accuracy: 0.7698\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.5211 - accuracy: 0.7844\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.9417 - accuracy: 0.6620\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.1945 - accuracy: 0.8438\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.5716 - accuracy: 0.7468\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.4330 - accuracy: 0.7915\n",
      "1/1 [==============================] - 0s 17ms/step - loss: 0.4162 - accuracy: 0.7838\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.4538 - accuracy: 0.6652\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4341 - accuracy: 0.8399\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.8921 - accuracy: 0.7302\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.5665 - accuracy: 0.7551\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4672 - accuracy: 0.7870\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.6747 - accuracy: 0.7717\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3238 - accuracy: 0.8374\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.4862 - accuracy: 0.7679\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4754 - accuracy: 0.7022\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.5863 - accuracy: 0.7270\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5596 - accuracy: 0.7449\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.7796 - accuracy: 0.7162\n",
      "1/1 [==============================] - 0s 19ms/step - loss: 0.4389 - accuracy: 0.7647\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4835 - accuracy: 0.7819\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.2803 - accuracy: 0.8438\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.6726 - accuracy: 0.7538\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5798 - accuracy: 0.6996\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3537 - accuracy: 0.8214\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3770 - accuracy: 0.7436\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.6436 - accuracy: 0.7366\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4646 - accuracy: 0.7774\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3645 - accuracy: 0.7519\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.4376 - accuracy: 0.7710\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.6180 - accuracy: 0.7659\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.6921 - accuracy: 0.6862\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4535 - accuracy: 0.8246\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5178 - accuracy: 0.7870\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.9946 - accuracy: 0.7085\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3831 - accuracy: 0.7577\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.6454 - accuracy: 0.6830\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.9500 - accuracy: 0.6875\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.6922 - accuracy: 0.7334\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5688 - accuracy: 0.7487\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2520 - accuracy: 0.8074\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.8480 - accuracy: 0.7009\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.5258 - accuracy: 0.7347\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4871 - accuracy: 0.6460\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5645 - accuracy: 0.8023\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 1.0947 - accuracy: 0.7117\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.5456 - accuracy: 0.7704\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.6152 - accuracy: 0.6907\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.9498 - accuracy: 0.6747\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.7414 - accuracy: 0.7417\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7364 - accuracy: 0.6339\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.6516 - accuracy: 0.7054\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.7986 - accuracy: 0.7117\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.2767 - accuracy: 0.7972\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.8250 - accuracy: 0.6582\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.6707 - accuracy: 0.7557\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2452 - accuracy: 0.8941\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.8037 - accuracy: 0.6805\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.5595 - accuracy: 0.7372\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.7840 - accuracy: 0.7640\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1178 - accuracy: 0.8501\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5188 - accuracy: 0.7302\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5512 - accuracy: 0.7117\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.8063 - accuracy: 0.7436\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5630 - accuracy: 0.7462\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.3150 - accuracy: 0.8010\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5713 - accuracy: 0.7296\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3920 - accuracy: 0.7781\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3879 - accuracy: 0.7691\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5172 - accuracy: 0.7111\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.4526 - accuracy: 0.7851\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.3045 - accuracy: 0.7768\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.3711 - accuracy: 0.7494\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4708 - accuracy: 0.7895\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.7120 - accuracy: 0.7207\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.7792 - accuracy: 0.6983\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.4324 - accuracy: 0.7136\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4146 - accuracy: 0.7895\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3406 - accuracy: 0.7946\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.3982 - accuracy: 0.7073\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.3421 - accuracy: 0.7921\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3256 - accuracy: 0.7315\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.6072 - accuracy: 0.7870\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.7014 - accuracy: 0.7181\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4149 - accuracy: 0.7953\n",
      "1/1 [==============================] - 0s 18ms/step - loss: 0.4583 - accuracy: 0.7341\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.8349 - accuracy: 0.7041\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.4934 - accuracy: 0.7085\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3716 - accuracy: 0.8055\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.9073 - accuracy: 0.7519\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.3382 - accuracy: 0.8304\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.8259 - accuracy: 0.7194\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3201 - accuracy: 0.8042\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5566 - accuracy: 0.6996\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.6712 - accuracy: 0.7481\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.3690 - accuracy: 0.7972\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5383 - accuracy: 0.7328\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.5905 - accuracy: 0.7551\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5898 - accuracy: 0.7353\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5409 - accuracy: 0.7704\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4623 - accuracy: 0.6830\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3517 - accuracy: 0.8265\n",
      "1/1 [==============================] - 0s 17ms/step - loss: 0.6524 - accuracy: 0.7679\n",
      "1/1 [==============================] - 0s 18ms/step - loss: 0.4835 - accuracy: 0.7640\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.6268 - accuracy: 0.7513\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4027 - accuracy: 0.7851\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.6016 - accuracy: 0.8157\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.2944 - accuracy: 0.7806\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3504 - accuracy: 0.8074\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4997 - accuracy: 0.6703\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.5206 - accuracy: 0.7596\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.5216 - accuracy: 0.7672\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3856 - accuracy: 0.8036\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2531 - accuracy: 0.8195\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.9620 - accuracy: 0.6869\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.7084 - accuracy: 0.7812\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5264 - accuracy: 0.7864\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5931 - accuracy: 0.7085\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.3630 - accuracy: 0.7602\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.7939 - accuracy: 0.7870\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5828 - accuracy: 0.7136\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5823 - accuracy: 0.7774\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2756 - accuracy: 0.8380\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.9103 - accuracy: 0.7054\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4370 - accuracy: 0.7621\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.6131 - accuracy: 0.7870\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5421 - accuracy: 0.7666\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.5887 - accuracy: 0.7634\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.4313 - accuracy: 0.7972\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5860 - accuracy: 0.7653\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.4930 - accuracy: 0.7730\n",
      "1/1 [==============================] - 0s 20ms/step - loss: 0.5872 - accuracy: 0.7570\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.5113 - accuracy: 0.7334\n",
      "1/1 [==============================] - 0s 17ms/step - loss: 0.5792 - accuracy: 0.7372\n",
      "1/1 [==============================] - 0s 17ms/step - loss: 0.4691 - accuracy: 0.7347\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5293 - accuracy: 0.7761\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.9098 - accuracy: 0.7372\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3889 - accuracy: 0.8029\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.6859 - accuracy: 0.7009\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5561 - accuracy: 0.7474\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5579 - accuracy: 0.7730\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.4923 - accuracy: 0.7985\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.6526 - accuracy: 0.6894\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3037 - accuracy: 0.7500\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4481 - accuracy: 0.7870\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.3184 - accuracy: 0.7819\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.3632 - accuracy: 0.7353\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3655 - accuracy: 0.8087\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4101 - accuracy: 0.7698\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4525 - accuracy: 0.7717\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4193 - accuracy: 0.8329\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.5228 - accuracy: 0.7921\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.2932 - accuracy: 0.7742\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.6314 - accuracy: 0.7226\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.7274 - accuracy: 0.6971\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.6731 - accuracy: 0.7621\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5192 - accuracy: 0.7341\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.8962 - accuracy: 0.6971\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.5244 - accuracy: 0.7615\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.3730 - accuracy: 0.7806\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.1738 - accuracy: 0.7628\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4105 - accuracy: 0.7353\n",
      "1/1 [==============================] - 0s 17ms/step - loss: 0.3351 - accuracy: 0.8010\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.6993 - accuracy: 0.7481\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4392 - accuracy: 0.7207\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3084 - accuracy: 0.7640\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5497 - accuracy: 0.7577\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.6010 - accuracy: 0.6607\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 1.0597 - accuracy: 0.7034\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5217 - accuracy: 0.7092\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.7925 - accuracy: 0.7066\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5645 - accuracy: 0.7200\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3870 - accuracy: 0.8048\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7240 - accuracy: 0.6926\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3988 - accuracy: 0.7736\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4491 - accuracy: 0.8055\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5481 - accuracy: 0.7188\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2191 - accuracy: 0.8495\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.5240 - accuracy: 0.7219\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5846 - accuracy: 0.7334\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2009 - accuracy: 0.8782\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.2568 - accuracy: 0.8221\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5545 - accuracy: 0.7245\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3165 - accuracy: 0.8189\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5462 - accuracy: 0.7749\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.5232 - accuracy: 0.7972\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5518 - accuracy: 0.7264\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.2549 - accuracy: 0.8202\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4824 - accuracy: 0.7232\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.8610 - accuracy: 0.7321\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5496 - accuracy: 0.7761\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.6897 - accuracy: 0.7730\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.6219 - accuracy: 0.8106\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4776 - accuracy: 0.7800\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4888 - accuracy: 0.7462\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.7041 - accuracy: 0.7347\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.8883 - accuracy: 0.6843\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.8096 - accuracy: 0.6569\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4017 - accuracy: 0.7251\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.6171 - accuracy: 0.7105\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3757 - accuracy: 0.7972\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.3779 - accuracy: 0.7736\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4228 - accuracy: 0.7538\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.6528 - accuracy: 0.7640\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.4288 - accuracy: 0.7921\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.3978 - accuracy: 0.7832\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5743 - accuracy: 0.7436\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.8395 - accuracy: 0.6677\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.8914 - accuracy: 0.6492\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2961 - accuracy: 0.8214\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.3592 - accuracy: 0.8304\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.3187 - accuracy: 0.8017\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.7544 - accuracy: 0.7130\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5667 - accuracy: 0.7232\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.4390 - accuracy: 0.8042\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 1.0487 - accuracy: 0.6614\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.4082 - accuracy: 0.7793\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4474 - accuracy: 0.8520\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.5808 - accuracy: 0.7474\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5257 - accuracy: 0.7608\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4767 - accuracy: 0.7034\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.7472 - accuracy: 0.7423\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.2582 - accuracy: 0.8284\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5793 - accuracy: 0.7175\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.4387 - accuracy: 0.8157\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.6245 - accuracy: 0.7838\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.8590 - accuracy: 0.6665\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.6982 - accuracy: 0.6760\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 1.0199 - accuracy: 0.6626\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5567 - accuracy: 0.7557\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4011 - accuracy: 0.7946\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.4703 - accuracy: 0.7213\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.5137 - accuracy: 0.7398\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4761 - accuracy: 0.7500\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.8377 - accuracy: 0.7360\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4256 - accuracy: 0.7985\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2814 - accuracy: 0.7698\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.8048 - accuracy: 0.7200\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.6198 - accuracy: 0.7149\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4658 - accuracy: 0.6824\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.5193 - accuracy: 0.6671\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4059 - accuracy: 0.7589\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5258 - accuracy: 0.7449\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1795 - accuracy: 0.8699\n",
      "1/1 [==============================] - 0s 18ms/step - loss: 0.5983 - accuracy: 0.7175\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.5082 - accuracy: 0.7494\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4061 - accuracy: 0.7321\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4214 - accuracy: 0.7685\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.8663 - accuracy: 0.7506\n",
      "1/1 [==============================] - 0s 17ms/step - loss: 0.2412 - accuracy: 0.8469\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.8661 - accuracy: 0.6773\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.7177 - accuracy: 0.6926\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4870 - accuracy: 0.7538\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3304 - accuracy: 0.8119\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.8245 - accuracy: 0.7175\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.6801 - accuracy: 0.7328\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5518 - accuracy: 0.8080\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2472 - accuracy: 0.7959\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.3802 - accuracy: 0.7538\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4137 - accuracy: 0.7577\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4667 - accuracy: 0.7857\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.2837 - accuracy: 0.8884\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3038 - accuracy: 0.8182\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3600 - accuracy: 0.8284\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.7997 - accuracy: 0.6913\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 1.1293 - accuracy: 0.6645\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3894 - accuracy: 0.7219\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.4770 - accuracy: 0.7411\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3936 - accuracy: 0.8068\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.8522 - accuracy: 0.7175\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.4824 - accuracy: 0.7481\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.8244 - accuracy: 0.6103\n",
      "1/1 [==============================] - 0s 19ms/step - loss: 0.6762 - accuracy: 0.7047\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4980 - accuracy: 0.7921\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.4421 - accuracy: 0.7793\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4129 - accuracy: 0.7417\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.3814 - accuracy: 0.7532\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.7590 - accuracy: 0.7181\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.5416 - accuracy: 0.7449\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.9898 - accuracy: 0.7372\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2835 - accuracy: 0.8272\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.7392 - accuracy: 0.7481\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3229 - accuracy: 0.7895\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.9031 - accuracy: 0.6690\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5248 - accuracy: 0.7653\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.6086 - accuracy: 0.7372\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.4012 - accuracy: 0.7634\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5486 - accuracy: 0.7851\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.6208 - accuracy: 0.7232\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3645 - accuracy: 0.8023\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.7388 - accuracy: 0.6735\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4497 - accuracy: 0.7691\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4752 - accuracy: 0.7551\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.6366 - accuracy: 0.7411\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 1.0650 - accuracy: 0.6856\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.7451 - accuracy: 0.6958\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.5874 - accuracy: 0.7117\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2920 - accuracy: 0.7723\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.4758 - accuracy: 0.7659\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4294 - accuracy: 0.7366\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.3672 - accuracy: 0.7704\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4297 - accuracy: 0.7570\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.6858 - accuracy: 0.6837\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.5860 - accuracy: 0.7672\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4080 - accuracy: 0.8087\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.6638 - accuracy: 0.6958\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3001 - accuracy: 0.8444\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.6566 - accuracy: 0.7181\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 1.0443 - accuracy: 0.5867\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5436 - accuracy: 0.7353\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3108 - accuracy: 0.7870\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4857 - accuracy: 0.7596\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2450 - accuracy: 0.8654\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5714 - accuracy: 0.7047\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.7288 - accuracy: 0.7379\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.8546 - accuracy: 0.6920\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 1.4081 - accuracy: 0.6913\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.7007 - accuracy: 0.7341\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.3996 - accuracy: 0.8087\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.7716 - accuracy: 0.7341\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 1.0367 - accuracy: 0.6977\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4566 - accuracy: 0.8393\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.4431 - accuracy: 0.7864\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.5415 - accuracy: 0.7392\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3949 - accuracy: 0.7895\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.6373 - accuracy: 0.7462\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3890 - accuracy: 0.8240\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4425 - accuracy: 0.7889\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4812 - accuracy: 0.7615\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.5059 - accuracy: 0.7213\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3996 - accuracy: 0.7800\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3631 - accuracy: 0.8425\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4956 - accuracy: 0.7385\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3441 - accuracy: 0.8673\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4800 - accuracy: 0.6958\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4976 - accuracy: 0.7379\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3483 - accuracy: 0.7723\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3811 - accuracy: 0.7379\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.7288 - accuracy: 0.7149\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3282 - accuracy: 0.7277\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4636 - accuracy: 0.7596\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3407 - accuracy: 0.7545\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.4684 - accuracy: 0.7781\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.5993 - accuracy: 0.6996\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.6242 - accuracy: 0.7902\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3740 - accuracy: 0.7844\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.7011 - accuracy: 0.7806\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4805 - accuracy: 0.6894\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4880 - accuracy: 0.8144\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5655 - accuracy: 0.7277\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.9144 - accuracy: 0.7270\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5665 - accuracy: 0.7659\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4677 - accuracy: 0.7188\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4268 - accuracy: 0.7321\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.1290 - accuracy: 0.8884\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5113 - accuracy: 0.7526\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4712 - accuracy: 0.7347\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.4037 - accuracy: 0.6958\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4865 - accuracy: 0.7781\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.7287 - accuracy: 0.7532\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.8844 - accuracy: 0.6575\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4629 - accuracy: 0.7583\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5845 - accuracy: 0.7455\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.8150 - accuracy: 0.7149\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3586 - accuracy: 0.8112\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.6633 - accuracy: 0.6747\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4484 - accuracy: 0.7698\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5006 - accuracy: 0.6665\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.3652 - accuracy: 0.7468\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.6282 - accuracy: 0.7526\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.1501 - accuracy: 0.9101\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.4925 - accuracy: 0.7577\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.8227 - accuracy: 0.7283\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4581 - accuracy: 0.8010\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.5435 - accuracy: 0.7621\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3963 - accuracy: 0.8080\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.4924 - accuracy: 0.7551\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.5605 - accuracy: 0.7385\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.5143 - accuracy: 0.7640\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4302 - accuracy: 0.7742\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.6949 - accuracy: 0.7462\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4254 - accuracy: 0.8138\n",
      "1/1 [==============================] - 0s 16ms/step - loss: 0.4750 - accuracy: 0.7908\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4712 - accuracy: 0.7991\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5960 - accuracy: 0.7124\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.4991 - accuracy: 0.7341\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.2865 - accuracy: 0.7966\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4862 - accuracy: 0.7468\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5055 - accuracy: 0.7066\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.5970 - accuracy: 0.7283\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4123 - accuracy: 0.8119\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3542 - accuracy: 0.7940\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.5139 - accuracy: 0.7411\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3833 - accuracy: 0.7557\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.6798 - accuracy: 0.7309\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.7018 - accuracy: 0.7736\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.4898 - accuracy: 0.7156\n",
      "1/1 [==============================] - 0s 14ms/step - loss: 0.2931 - accuracy: 0.8476\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.5016 - accuracy: 0.7832\n",
      "1/1 [==============================] - 0s 15ms/step - loss: 0.3048 - accuracy: 0.8093\n",
      "1/1 [==============================] - 0s 20ms/step - loss: 0.4606 - accuracy: 0.7864\n"
     ]
    }
   ],
   "source": [
    "test  = autoencoder.predict(anomaly_test_data)\n",
    "anomaly_det= 0\n",
    "test_data_nos = 500\n",
    "\n",
    "for i in range(1, test_data_nos):\n",
    "    test_reconstruction_error = autoencoder.evaluate(test[i],anomaly_data[i])\n",
    "    if test_reconstruction_error[0] > threshold[0] or test_reconstruction_error[0] < threshold[1]:\n",
    "        anomaly_det += 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "id": "16d0493a",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy of the model detecting an anomaly is:  85.6\n"
     ]
    }
   ],
   "source": [
    "accuracy = anomaly_det/test_data_nos *100\n",
    "print(\"Accuracy of the model detecting an anomaly is: \",accuracy)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "id": "f1ec97ab",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 15ms/step - loss: 0.0948 - accuracy: 0.8227\n",
      "\n",
      "The reconstruction error is:  0.0948\n",
      "Anomaly Detected\n"
     ]
    }
   ],
   "source": [
    "test  = autoencoder.predict(anomaly_data)\n",
    "\n",
    "plt.imshow(anomaly_data[0].reshape(28, 28))\n",
    "plt.show()\n",
    "\n",
    "test_reconstruction_error = autoencoder.evaluate(test[0],anomaly_data[0])\n",
    "print('\\nThe reconstruction error is: ',format(test_reconstruction_error[0],'.4f'))\n",
    "\n",
    "if test_reconstruction_error[0] > threshold[0] or test_reconstruction_error[0] < threshold[1]:\n",
    "    print(\"Anomaly Detected\")\n",
    "else:\n",
    "    print(\"Data is Pure\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "5ca4c30f",
   "metadata": {},
   "outputs": [],
   "source": [
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "0ab9fd52",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}