2.2 - EDA (Temporal Anomaly Detection).ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# EDA (Temporal Anomaly Detection)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "MdHy_Al6Lb15"
   },
   "outputs": [],
   "source": [
    "import math\n",
    "import datetime\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "\n",
    "import tensorflow.keras as tfk\n",
    "\n",
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt\n",
    "from matplotlib import animation\n",
    "sns.set(style=\"whitegrid\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "0Kq4l3lWLb2A"
   },
   "source": [
    "## The Data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "52dZUXIxLb2C"
   },
   "source": [
    "### Load the density images"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 35
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 5794,
     "status": "ok",
     "timestamp": 1559838800993,
     "user": {
      "displayName": "Franz Louis Cesista",
      "photoUrl": "https://lh3.googleusercontent.com/-NquKeLt1770/AAAAAAAAAAI/AAAAAAAAAMo/RHnx0T1cGIk/s64/photo.jpg",
      "userId": "08273419339046628946"
     },
     "user_tz": -480
    },
    "id": "QOxTP6cWLb2E",
    "outputId": "cfca012a-f2fe-40a4-e37b-799365f62a47"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(5856, 48, 40)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "images = np.load('density_images.npy')\n",
    "images.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "RuxKBBfulmW0"
   },
   "source": [
    "### Load the mask"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 35
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 5760,
     "status": "ok",
     "timestamp": 1559838800994,
     "user": {
      "displayName": "Franz Louis Cesista",
      "photoUrl": "https://lh3.googleusercontent.com/-NquKeLt1770/AAAAAAAAAAI/AAAAAAAAAMo/RHnx0T1cGIk/s64/photo.jpg",
      "userId": "08273419339046628946"
     },
     "user_tz": -480
    },
    "id": "f4Wxd5sQllmH",
    "outputId": "1197dae2-c437-4f2b-caa0-e02f25ef2a8a"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(48, 40)"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "mask = np.load('density_mask.npy')\n",
    "mask.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "9ARIS8GJLb2Z"
   },
   "source": [
    "### Extend the images"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 35
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 5733,
     "status": "ok",
     "timestamp": 1559838800996,
     "user": {
      "displayName": "Franz Louis Cesista",
      "photoUrl": "https://lh3.googleusercontent.com/-NquKeLt1770/AAAAAAAAAAI/AAAAAAAAAMo/RHnx0T1cGIk/s64/photo.jpg",
      "userId": "08273419339046628946"
     },
     "user_tz": -480
    },
    "id": "iA9w9ujpLb2f",
    "outputId": "ae25cc76-dbc7-4aad-ac3f-e6ef291f9e09"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1920,)"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "flat_images = images.reshape(images.shape[0], -1).astype('float32')\n",
    "flat_mask   = mask.reshape(-1).astype('float32')\n",
    "flat_mask.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "vu-EMLB0Lb2o"
   },
   "source": [
    "## The Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "eqbg0VbWLb2q"
   },
   "outputs": [],
   "source": [
    "original_dim = flat_images.shape[1]\n",
    "intermediate_dim = 256\n",
    "latent_dim = 2"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "G41cg_6hLb2u"
   },
   "source": [
    "### Build the model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "Pdeu0I0VLb2v"
   },
   "source": [
    "#### Build the encoder"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "iJcCvLAmLb2w"
   },
   "outputs": [],
   "source": [
    "encoder = tfk.Sequential([\n",
    "    tfk.layers.Dense(512, input_shape=(original_dim,), activation='relu'),\n",
    "    tfk.layers.Dense(256, activation='relu'),\n",
    "    tfk.layers.Dense(128, activation='relu'),\n",
    "    tfk.layers.Dense(64, activation='relu'),\n",
    "    tfk.layers.Dense(32, activation='relu'),\n",
    "    tfk.layers.Dense(16, activation='relu'),\n",
    "    tfk.layers.Dense(8, activation='relu'),\n",
    "    tfk.layers.Dense(4, activation='relu'),\n",
    "    tfk.layers.Dense(latent_dim)\n",
    "], name=\"encoder\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 490
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 6845,
     "status": "ok",
     "timestamp": 1559838802132,
     "user": {
      "displayName": "Franz Louis Cesista",
      "photoUrl": "https://lh3.googleusercontent.com/-NquKeLt1770/AAAAAAAAAAI/AAAAAAAAAMo/RHnx0T1cGIk/s64/photo.jpg",
      "userId": "08273419339046628946"
     },
     "user_tz": -480
    },
    "id": "woXgDtBULb20",
    "outputId": "8c2efb67-59ec-41aa-f60a-cf49aa67cbbe"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"encoder\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "dense (Dense)                (None, 512)               983552    \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 256)               131328    \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 128)               32896     \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 64)                8256      \n",
      "_________________________________________________________________\n",
      "dense_4 (Dense)              (None, 32)                2080      \n",
      "_________________________________________________________________\n",
      "dense_5 (Dense)              (None, 16)                528       \n",
      "_________________________________________________________________\n",
      "dense_6 (Dense)              (None, 8)                 136       \n",
      "_________________________________________________________________\n",
      "dense_7 (Dense)              (None, 4)                 36        \n",
      "_________________________________________________________________\n",
      "dense_8 (Dense)              (None, 2)                 10        \n",
      "=================================================================\n",
      "Total params: 1,158,822\n",
      "Trainable params: 1,158,822\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "encoder.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "4ELVl-XSLb2_"
   },
   "source": [
    "#### Build the decoder"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "U2zHMfKjLb3A"
   },
   "outputs": [],
   "source": [
    "decoder = tfk.Sequential([\n",
    "    tfk.layers.Dense(8, input_shape=(latent_dim,), activation='relu'),\n",
    "    tfk.layers.Dense(16, activation='relu'),\n",
    "    tfk.layers.Dense(32, activation='relu'),\n",
    "    tfk.layers.Dense(64, activation='relu'),\n",
    "    tfk.layers.Dense(128, activation='relu'),\n",
    "    tfk.layers.Dense(256, activation='relu'),\n",
    "    tfk.layers.Dense(512, activation='relu'),\n",
    "    tfk.layers.Dense(original_dim, activation='sigmoid')\n",
    "], name=\"decoder\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 454
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 6831,
     "status": "ok",
     "timestamp": 1559838802143,
     "user": {
      "displayName": "Franz Louis Cesista",
      "photoUrl": "https://lh3.googleusercontent.com/-NquKeLt1770/AAAAAAAAAAI/AAAAAAAAAMo/RHnx0T1cGIk/s64/photo.jpg",
      "userId": "08273419339046628946"
     },
     "user_tz": -480
    },
    "id": "ZcUsjNyGLb3E",
    "outputId": "4b352ec0-107b-4549-d7ff-55586072270f"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"decoder\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "dense_9 (Dense)              (None, 8)                 24        \n",
      "_________________________________________________________________\n",
      "dense_10 (Dense)             (None, 16)                144       \n",
      "_________________________________________________________________\n",
      "dense_11 (Dense)             (None, 32)                544       \n",
      "_________________________________________________________________\n",
      "dense_12 (Dense)             (None, 64)                2112      \n",
      "_________________________________________________________________\n",
      "dense_13 (Dense)             (None, 128)               8320      \n",
      "_________________________________________________________________\n",
      "dense_14 (Dense)             (None, 256)               33024     \n",
      "_________________________________________________________________\n",
      "dense_15 (Dense)             (None, 512)               131584    \n",
      "_________________________________________________________________\n",
      "dense_16 (Dense)             (None, 1920)              984960    \n",
      "=================================================================\n",
      "Total params: 1,160,712\n",
      "Trainable params: 1,160,712\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "decoder.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "zAlHWwX6Lb3J"
   },
   "source": [
    "#### Build the autoencoder"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "8wlJD56HLb3K"
   },
   "outputs": [],
   "source": [
    "autoencoder = tfk.Sequential([\n",
    "    encoder,\n",
    "    decoder\n",
    "], name=\"autoencoder\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 235
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 6803,
     "status": "ok",
     "timestamp": 1559838802150,
     "user": {
      "displayName": "Franz Louis Cesista",
      "photoUrl": "https://lh3.googleusercontent.com/-NquKeLt1770/AAAAAAAAAAI/AAAAAAAAAMo/RHnx0T1cGIk/s64/photo.jpg",
      "userId": "08273419339046628946"
     },
     "user_tz": -480
    },
    "id": "UrBlu1nWLb3O",
    "outputId": "2f988c69-2685-46fa-cfcd-f040cbcfe01b"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"autoencoder\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "encoder (Sequential)         (None, 2)                 1158822   \n",
      "_________________________________________________________________\n",
      "decoder (Sequential)         (None, 1920)              1160712   \n",
      "=================================================================\n",
      "Total params: 2,319,534\n",
      "Trainable params: 2,319,534\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "autoencoder.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "0jNqGL-ZLb3V"
   },
   "source": [
    "### Train the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1054
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 43130,
     "status": "ok",
     "timestamp": 1559838838497,
     "user": {
      "displayName": "Franz Louis Cesista",
      "photoUrl": "https://lh3.googleusercontent.com/-NquKeLt1770/AAAAAAAAAAI/AAAAAAAAAMo/RHnx0T1cGIk/s64/photo.jpg",
      "userId": "08273419339046628946"
     },
     "user_tz": -480
    },
    "id": "_8op4OG-Lb3W",
    "outputId": "6b6c461b-5b0d-43c4-94d4-3e0fdabbefcd"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 5270 samples, validate on 586 samples\n",
      "Epoch 1/1000\n",
      "5270/5270 [==============================] - 7s 1ms/sample - loss: 21.5745 - val_loss: 5.1391\n",
      "Epoch 2/1000\n",
      "5270/5270 [==============================] - 7s 1ms/sample - loss: 3.7395 - val_loss: 3.1129\n",
      "Epoch 3/1000\n",
      "5270/5270 [==============================] - 8s 1ms/sample - loss: 2.0707 - val_loss: 1.7272\n",
      "Epoch 4/1000\n",
      "5270/5270 [==============================] - 8s 1ms/sample - loss: 1.6945 - val_loss: 1.6790\n",
      "Epoch 5/1000\n",
      "5270/5270 [==============================] - 8s 1ms/sample - loss: 1.6402 - val_loss: 1.7221\n",
      "Epoch 6/1000\n",
      "5270/5270 [==============================] - 7s 1ms/sample - loss: 1.6213 - val_loss: 1.6779\n",
      "Epoch 7/1000\n",
      "5270/5270 [==============================] - 6s 1ms/sample - loss: 1.6045 - val_loss: 1.7119\n",
      "Epoch 8/1000\n",
      "5270/5270 [==============================] - 7s 1ms/sample - loss: 1.5791 - val_loss: 1.6310\n",
      "Epoch 9/1000\n",
      "5270/5270 [==============================] - 7s 1ms/sample - loss: 1.5873 - val_loss: 1.6380\n",
      "Epoch 10/1000\n",
      "5270/5270 [==============================] - 6s 1ms/sample - loss: 1.5599 - val_loss: 1.6046\n",
      "Epoch 11/1000\n",
      "5270/5270 [==============================] - 6s 1ms/sample - loss: 1.5568 - val_loss: 1.6348\n",
      "Epoch 12/1000\n",
      "5270/5270 [==============================] - 7s 1ms/sample - loss: 1.5415 - val_loss: 1.6140\n",
      "Epoch 13/1000\n",
      "5270/5270 [==============================] - 6s 1ms/sample - loss: 1.5529 - val_loss: 1.6197\n",
      "Epoch 14/1000\n",
      "5270/5270 [==============================] - 7s 1ms/sample - loss: 1.5472 - val_loss: 1.5903\n",
      "Epoch 15/1000\n",
      "5270/5270 [==============================] - 8s 1ms/sample - loss: 1.5327 - val_loss: 1.5581\n",
      "Epoch 16/1000\n",
      "5270/5270 [==============================] - 7s 1ms/sample - loss: 1.5442 - val_loss: 1.5945\n",
      "Epoch 17/1000\n",
      "5270/5270 [==============================] - 7s 1ms/sample - loss: 1.5444 - val_loss: 1.6155\n",
      "Epoch 18/1000\n",
      "5270/5270 [==============================] - 7s 1ms/sample - loss: 1.5407 - val_loss: 1.5741\n",
      "Epoch 19/1000\n",
      "5270/5270 [==============================] - 7s 1ms/sample - loss: 1.5459 - val_loss: 1.5878\n",
      "Epoch 20/1000\n",
      "5270/5270 [==============================] - 7s 1ms/sample - loss: 1.5323 - val_loss: 1.5838\n",
      "Epoch 21/1000\n",
      "5270/5270 [==============================] - 7s 1ms/sample - loss: 1.5006 - val_loss: 1.5376\n",
      "Epoch 22/1000\n",
      "5270/5270 [==============================] - 6s 1ms/sample - loss: 1.3781 - val_loss: 1.3256\n",
      "Epoch 23/1000\n",
      "5270/5270 [==============================] - 6s 1ms/sample - loss: 1.2623 - val_loss: 1.2921\n",
      "Epoch 24/1000\n",
      "5270/5270 [==============================] - 6s 1ms/sample - loss: 1.2015 - val_loss: 1.2667\n",
      "Epoch 25/1000\n",
      "5270/5270 [==============================] - 7s 1ms/sample - loss: 1.1738 - val_loss: 1.2843\n",
      "Epoch 26/1000\n",
      "5270/5270 [==============================] - 7s 1ms/sample - loss: 1.1238 - val_loss: 1.2145\n",
      "Epoch 27/1000\n",
      "5270/5270 [==============================] - 6s 1ms/sample - loss: 1.1004 - val_loss: 1.2443\n",
      "Epoch 28/1000\n",
      "5270/5270 [==============================] - 7s 1ms/sample - loss: 1.0678 - val_loss: 1.2843\n",
      "Epoch 29/1000\n",
      "5270/5270 [==============================] - 6s 1ms/sample - loss: 1.0600 - val_loss: 1.2105\n",
      "Epoch 30/1000\n",
      "5270/5270 [==============================] - 8s 1ms/sample - loss: 1.0300 - val_loss: 1.1806\n",
      "Epoch 31/1000\n",
      "5270/5270 [==============================] - 6s 1ms/sample - loss: 1.0211 - val_loss: 1.1859\n",
      "Epoch 32/1000\n",
      "5270/5270 [==============================] - 6s 1ms/sample - loss: 1.0139 - val_loss: 1.2160\n",
      "Epoch 33/1000\n",
      "5270/5270 [==============================] - 6s 1ms/sample - loss: 0.9983 - val_loss: 1.1972\n",
      "Epoch 34/1000\n",
      "5270/5270 [==============================] - 6s 1ms/sample - loss: 1.0030 - val_loss: 1.2115\n",
      "Epoch 35/1000\n",
      "5270/5270 [==============================] - 6s 1ms/sample - loss: 0.9826 - val_loss: 1.1905\n",
      "Epoch 36/1000\n",
      "5270/5270 [==============================] - 6s 1ms/sample - loss: 0.9922 - val_loss: 1.1815\n",
      "Epoch 37/1000\n",
      "5270/5270 [==============================] - 6s 1ms/sample - loss: 0.9815 - val_loss: 1.1589\n",
      "Epoch 38/1000\n",
      "5270/5270 [==============================] - 6s 1ms/sample - loss: 0.9727 - val_loss: 1.1750\n",
      "Epoch 39/1000\n",
      "5270/5270 [==============================] - 8s 2ms/sample - loss: 0.9660 - val_loss: 1.1545\n",
      "Epoch 40/1000\n",
      "5270/5270 [==============================] - 8s 1ms/sample - loss: 0.9513 - val_loss: 1.1737\n",
      "Epoch 41/1000\n",
      "5270/5270 [==============================] - 8s 1ms/sample - loss: 0.9493 - val_loss: 1.1613\n",
      "Epoch 42/1000\n",
      "5270/5270 [==============================] - 8s 1ms/sample - loss: 0.9768 - val_loss: 1.1276\n",
      "Epoch 43/1000\n",
      "5270/5270 [==============================] - 9s 2ms/sample - loss: 0.9601 - val_loss: 1.1474\n",
      "Epoch 44/1000\n",
      "5270/5270 [==============================] - 8s 2ms/sample - loss: 0.9672 - val_loss: 1.1649\n",
      "Epoch 45/1000\n",
      "5270/5270 [==============================] - 10s 2ms/sample - loss: 0.9414 - val_loss: 1.2072\n",
      "Epoch 46/1000\n",
      "5270/5270 [==============================] - 9s 2ms/sample - loss: 0.9498 - val_loss: 1.2483\n",
      "Epoch 47/1000\n",
      "5270/5270 [==============================] - 9s 2ms/sample - loss: 0.9506 - val_loss: 1.1679\n",
      "Epoch 48/1000\n",
      "5270/5270 [==============================] - 9s 2ms/sample - loss: 0.9400 - val_loss: 1.1935\n",
      "Epoch 49/1000\n",
      "5270/5270 [==============================] - 8s 1ms/sample - loss: 0.9351 - val_loss: 1.1637\n",
      "Epoch 50/1000\n",
      "5270/5270 [==============================] - 7s 1ms/sample - loss: 0.9196 - val_loss: 1.1847\n",
      "Epoch 51/1000\n",
      "5270/5270 [==============================] - 7s 1ms/sample - loss: 0.9580 - val_loss: 1.1821\n",
      "Epoch 52/1000\n",
      "5270/5270 [==============================] - 7s 1ms/sample - loss: 0.9204 - val_loss: 1.1620\n"
     ]
    }
   ],
   "source": [
    "class PrintDot(tfk.callbacks.Callback):\n",
    "  def on_epoch_end(self, epoch, logs):\n",
    "    if epoch % 100 == 0:\n",
    "      print('')\n",
    "    print('.', end='')\n",
    "\n",
    "mask_sum = mask.sum()\n",
    "    \n",
    "def scaled_loss(inputs, outputs):\n",
    "  outputs *= flat_mask\n",
    "  loss = tfk.metrics.mean_squared_error(inputs, outputs)\n",
    "  return mask_sum * loss\n",
    "\n",
    "early_stop = tfk.callbacks.EarlyStopping(monitor='val_loss', patience=10)\n",
    "\n",
    "autoencoder.compile(optimizer='adam', loss=scaled_loss)\n",
    "history = autoencoder.fit(x=flat_images,\n",
    "                          y=flat_images,\n",
    "                          epochs=1000, batch_size=32,\n",
    "                          validation_split=0.1, verbose=1,\n",
    "                          callbacks=[early_stop]) # PrintDot(), "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 293
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 43715,
     "status": "ok",
     "timestamp": 1559838839104,
     "user": {
      "displayName": "Franz Louis Cesista",
      "photoUrl": "https://lh3.googleusercontent.com/-NquKeLt1770/AAAAAAAAAAI/AAAAAAAAAMo/RHnx0T1cGIk/s64/photo.jpg",
      "userId": "08273419339046628946"
     },
     "user_tz": -480
    },
    "id": "3SKdJba6XuIG",
    "outputId": "54c42c18-bd1f-4715-c2a3-fb12e04ac9ce"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x2e3875ee668>"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "pd.DataFrame(history.history).plot()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "ybvTSTXTNgmf"
   },
   "source": [
    "### Test the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 35
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 44161,
     "status": "ok",
     "timestamp": 1559838839570,
     "user": {
      "displayName": "Franz Louis Cesista",
      "photoUrl": "https://lh3.googleusercontent.com/-NquKeLt1770/AAAAAAAAAAI/AAAAAAAAAMo/RHnx0T1cGIk/s64/photo.jpg",
      "userId": "08273419339046628946"
     },
     "user_tz": -480
    },
    "id": "UuFaVRox4MiB",
    "outputId": "5620c6c3-f6b4-400f-9566-e8a435253b7d"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.07798811"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_pred = autoencoder.predict(flat_images)\n",
    "tfk.metrics.mean_squared_error(flat_images, y_pred).numpy().mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 54
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 44131,
     "status": "ok",
     "timestamp": 1559838839573,
     "user": {
      "displayName": "Franz Louis Cesista",
      "photoUrl": "https://lh3.googleusercontent.com/-NquKeLt1770/AAAAAAAAAAI/AAAAAAAAAMo/RHnx0T1cGIk/s64/photo.jpg",
      "userId": "08273419339046628946"
     },
     "user_tz": -480
    },
    "id": "peS2RM4JNie9",
    "outputId": "b9f25196-2376-4261-f0ff-1a299f694293"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "5856/5856 [==============================] - 4s 660us/sample - loss: 0.9478\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.9478113012887089"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "autoencoder.evaluate(flat_images, flat_images)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Save model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "autoencoder.save('temporal_anomaly_detection/autoencoder.h5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "autoencoder_ = tfk.models.load_model('temporal_anomaly_detection/autoencoder.h5',\n",
    "                                     custom_objects={'scaled_loss':scaled_loss})"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "T3NpeesRMgaE"
   },
   "source": [
    "### Latent space visualizations"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "kwq8IClHLb3b"
   },
   "outputs": [],
   "source": [
    "latent_vars = autoencoder.get_layer('encoder')(flat_images).numpy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 275
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 44744,
     "status": "ok",
     "timestamp": 1559838840205,
     "user": {
      "displayName": "Franz Louis Cesista",
      "photoUrl": "https://lh3.googleusercontent.com/-NquKeLt1770/AAAAAAAAAAI/AAAAAAAAAMo/RHnx0T1cGIk/s64/photo.jpg",
      "userId": "08273419339046628946"
     },
     "user_tz": -480
    },
    "id": "v7sG37_OMkyT",
    "outputId": "a1a63c8c-3905-4cb9-e521-9a03d43014d6"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "points = plt.scatter(latent_vars[:, 0], latent_vars[:, 1], alpha=0.5, linewidth=0)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 293
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 44722,
     "status": "ok",
     "timestamp": 1559838840208,
     "user": {
      "displayName": "Franz Louis Cesista",
      "photoUrl": "https://lh3.googleusercontent.com/-NquKeLt1770/AAAAAAAAAAI/AAAAAAAAAMo/RHnx0T1cGIk/s64/photo.jpg",
      "userId": "08273419339046628946"
     },
     "user_tz": -480
    },
    "id": "1fKrkoWSGBzO",
    "outputId": "7c797e8b-7828-4a22-9a70-684468c304d8"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x2e392579c50>]"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(latent_vars[:4*36, 0], latent_vars[:4*36, 1], '-o')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Calculate Mahalanobis Distances"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[ 0.02868951 -1.9072311 ]]\n"
     ]
    }
   ],
   "source": [
    "train_mu = latent_vars.mean(axis=0)\n",
    "train_mu = train_mu.reshape(1, -1)\n",
    "print(train_mu)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[0.5830229 2.27596  ]]\n"
     ]
    }
   ],
   "source": [
    "train_std = latent_vars.std(axis=0)\n",
    "train_std = train_std.reshape(1, -1)\n",
    "print(train_std)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(5856,)"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "mahalanobis_dist = (((latent_vars - train_mu)/train_std)**2).sum(axis=1)**(0.5)\n",
    "mahalanobis_dist.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Save the data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [],
   "source": [
    "np.save('temporal_anomaly_detection/mahalanobis_dist.npy', mahalanobis_dist)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [],
   "source": [
    "np.save('temporal_anomaly_detection/train_mu.npy', train_mu)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [],
   "source": [
    "np.save('temporal_anomaly_detection/train_std.npy', train_std)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Inspect"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x2e387825f28>]"
      ]
     },
     "execution_count": 60,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXIAAAD7CAYAAAB37B+tAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjAsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+17YcXAAAgAElEQVR4nO3dd3xb5b348Y8k2/KKkzh7JyTxkwQHAiJQRiAto0AJLZcCl1HKZfYy2h83FMoqUFqgtBTaS+mlI0ChjFIKhL0DCWQhCIkJeTJw9h624yXLln5/HB3pyFpHsmRb8H2/XnnFOtaRvj46+p7nPNMRDAYRQgiRv5w9HYAQQoiukUQuhBB5ThK5EELkOUnkQgiR5ySRCyFEnivo7jf0er1uYDqwDejo7vcXQog85QKGAUs9Ho/P+otuT+QYSXx+D7yvEEJ8FcwAFlg39EQi3wZQVVVFUVFR2jvX1NRQXV2d9aC6SuJKj8SVnt4aF/Te2L5qcbW1tbF69WoI5VCrnkjkHQBFRUW43e6MXiDT/XJN4kqPxJWe3hoX9N7YvqJxxVRJS2OnEELkOUnkQgiR5ySRCyFEnpNELoQQeU4SuRBC5DlbvVaUUhXAR8BpWuv1SqkjgfuBPsBy4Ida67bchSmEECKRlCVypdQRGJ3Pq0KPK4B/A5drrQ8MPe2SnEUohPhaW7u5jlmzX+RH97zNix+sIxCQNRQ6s1O1chlwFbA19PhEYKHWenno8TXA8zmITQgh0Ov3ArBlVxN/fbGGLbsaezii3sdhd4UgpdR6YCZwDnAgUARMAj4EZmutW+28jtfrHQvUph2pEOJr6YtNLTwzf0/48ZWnDmFwv8IejKjHjfN4POutGzIZ2VkAfBv4BrAR+BvwM+D2dF6kuro6o9FNXq8Xj8eT9n65JnGlR+JKT2+NC3IfW1vRVrAk8ikHTmHM0IoejytTmcbl8/moqamJ+7tMeq1sBxZprWu11h3AP4HDM3gdIYQQWZBJIn8T8CilRoUenwZ4sxeSEEJExNT+SltnjLQTudZ6E3AF8JJSahVQCdyd7cCEEELYY7uOXGs91vLzK8AruQhICCFEemRkpxBC5DlJ5EKIvCJV5LEkkQshejVJ3KlJIhdCiDwniVwIIfKcJHIhRF6xO63I14kkciGEyHOSyIUQvZsUwFOSRC6EEHlOErkQQuQ5SeRCCJHnJJELIUSek0QuhOjVgtLamZIkciFEXpFu5LEkkQshRJ6zlciVUhVKqRql1NhO269WSs3LRWBCCCHsSZnIlVJHAAuAqk7bp2AsuiyEEKIH2SmRXwZcBWw1Nyil3MDDwM9zFJcQQgCxdeIy10qslEu9aa0vBVBKWTffDcwBanMTlhBCCLscdq9uSqn1wExgInCF1vr7SqmZwO1a65l239Dr9Y5FLgBCCJtqNjTzrw/3hh9fcfJghlUW9WBEPW6cx+NZb91ge/Fli3OBA5VSy4ByYKhS6hmt9TnpvEh1dTVutzvtN/d6vXg8nrT3yzWJKz0SV3p6a1yQ+9iaXVvAksgnT57M+JH9ejyuTGUal8/no6amJu7v0k7kWuuLzZ8tJfK0krgQQmRKashjST9yIUTvJpk7Jdslcq312Djb5mHUmwshhOghUiIXQog8J4lcCJFfpKolhiRyIYTIc5LIhRC9mkxjm5okciGEyHOSyIUQeUVK6LEkkQshRJ6TRC6EEHlOErkQoleTWWtTk0QuhMgrkthjSSIXQog8J4lcCCHynCRyIYTIc5LIhRC9mlSJpyaJXAgh8pwkciGEyHO2FpZQSlUAHwGnaa3XK6UuB36McdfzMcZizG25C1MIIUQiKUvkSqkjgAVAVehxFfBT4CjgoNBrXJXDGIUQX2edOo4HpSN5DDtVK5dhJOqtocc+4EqtdYPWOgisAEbnKD4hhBAppKxa0VpfCqCUMh9vADaEtg0CrgYuylmEQgghknLYvU1RSq0HZmqt14cejwBeA57VWt9p9w29Xu9YoDbNOIUQX1PLa5v598K94ceXnjSIkQPdPRhRjxvn8XjWWzfYauzsTCk1CXgD+IPW+r5MXqO6uhq3O/0Pw+v14vF4MnnLnJK40iNxpae3xgW5j20/m8CSyNWkSUwaU9njcWUq07h8Ph81NTVxf5d2IldK9QHeBG7WWj+edjRCCJEGadpMLZMS+aXAEGC2Ump2aNtcrfXPsxeWEEIIu2wncq312NCP94f+CSGE6AVkZKcQIr9IXUsMSeRCCJHnJJELIXo1GciZmiRyIYTIc5LIhRB5RUrosSSRCyFEnpNELoQQeU4SuRCil5O6lFQkkQsh8kpQEnsMSeRCCJHnJJELIUSek0QuhBB5ThK5EKJX69xvXPqRx5JELoQQeU4SuRBC5DlJ5EIIkedsLSyhlKoAPgJO01qvV0qdAPwOKAGe0VrfksMYhRBCJJGyRK6UOgJYAFSFHpcAc4DvApOB6UqpU3IZpBDi60saN1OzU7VyGXAVsDX0+HBgjda6VmvdDjwBnJWj+IQQQqSQsmpFa30pgFLK3DQc2GZ5yjZgZNYjE0IIYYvtxZctnETPYuMAAum+SE1NTQZvbfB6vRnvm0sSV3okrvT01rggt7Gt39AU9XiV1rTuc9vat7ces2zHlUki3wwMszweSqTaxbbq6mrcbnsfhpXX68Xj8aS9X65JXOmRuNLTW+OC3Me2r2MjLNoXfqyqqqgeP7DH48pUpnH5fL6EBeBMEvliQCmlJgC1wHkYjZ9CCJED0tqZStr9yLXWrcBFwHPASmAV8K/shiWEEMIu2yVyrfVYy8/vAAfnIiAhhEhGyuexZGSnEELkOUnkQgiR5ySRCyF6NRnZmZokciFEfpHEHkMSuRBC5DlJ5EIIkeckkQshRJ6TRC6E6NU6V4kHpZI8hiRyIYTIc5LIhRAiz0kiF0KIPCeJXAiRV2SAUCxJ5EKIXk0Sd2qSyIUQIs9JIhdCiDwniVwIkV+kqiVGJku9hSmlLgBuDD18TWt9XddDEkIIkY6MS+RKqVLgD8BxGKsFzVBKnZCtwIQQwiBF8FS6UrXiCu1fBhSG/rVkIyghhBD2OYJd6NujlLoGuBdoBt4HztRaJ31Br9c7FqjN+E2FEF8r3rWNvLSkLvz4B98ayPihxT0YUY8b5/F41ls3ZFxHrpQ6CLgYGAPUA08A1wG/sbN/dXU1brc77ff1er14PJ6098s1iSs9Eld6emtckPvYdvvXgyWRV02cyLSqwT0eV6Yyjcvn81FTUxP3d12pWvk28I7WeqfW2gc8CszswusJIYTIQFd6rXwG3KuUKsOoWpkFLM1KVEIIESIjO1PLuESutX4TeArwAssxGjvvyVJcQggRlyT2WF3qR661/jXw6yzFIoQQIgMyslMIIfKcJHIhhMhzksiFEL1a7JqdojNJ5EIIkeckkQshRJ6TRC6EEHlOErkQonfr3HFcKsljSCIXQog8J4lcCCHynCRyIYTIc5LIhRB5JSiV5DEkkQshejVJ26lJIhdCiDwniVwIIfKcJHIhRF6R+chjdWk+cqXULOA2oAx4U2v9k6xEJYQQwraMS+RKqQOA/wO+BxwEHKqUOiVbgQkhBEgJ3I6ulMjPAJ7RWm8GUEqdA7RmJSohhBC2dSWRTwDalFJzgdHAy8CtWYlKCCGEbY5ghvctSqm/AEcBM4FGYC7wD631o8n283q9Y4HajN5UCPG1s2R1I69+XBd+fN5xA6gaUdKDEfW4cR6PZ711Q1dK5NuBt7XWuwCUUs8DhwOP2tm5uroat9ud9pt6vV48Hk/a++WaxJUeiSs9vTUuyH1s21trwZLIJ06ciGfykB6PK1OZxuXz+aipqYn7u64k8peBx5RS/YD9wCnAC114PSGEiCWtnSll3GtFa70YuBdYAKwENgCPZCkuIYSIK9Pq4K+yLvUj11rPAeZkKRYhhBAZkJGdQgiR5ySRCyFEnpNELoTo1TrXiEsNeSxJ5EIIkeckkQshRJ6TRC6EEHlOErkQIr9IJXkMSeRCiF5Nxv+kJolcCCHynCRyIYTIc5LIhRB5ReZaiSWJXAgh8pwkciFErxaUbiopSSIXQog8J4lcCJFXpHweSxK5EELkuawkcqXUb5VSj2bjtYQQQqSny4lcKXU88MMsxCKEELGkLiWlLiVypVQl8CvgruyEI4QQyUk38lhdLZE/DNwM7MtCLEIIITLgyHSUlFLqUmCK1vp/lFIXATO11hel2s/r9Y4FajN6UyHE186iVft5/ZP68ONzZgxg8qiSHoyox43zeDzrrRsKuvBi5wDDlFLLgEqgXCl1v9b6Wjs7V1dX43a7035Tr9eLx+NJe79ck7jSI3Glp7fGBbmPbUvTOrAk8vHjx+OZOqzH48pUpnH5fD5qamri/i7jRK61PtH82VIit5XEhRDCrtg6A6kk70z6kQshRJ7rStVKmNb6UeDRbLyWEEKI9EiJXAgh8pwkciFEXglIFXkMSeRCiF6tcw/pQIdk8s4kkQsh8kp7INDTIfQ6ksiFED1m+56mtPfp6JBE3pkkciFEj/iidi+X3fU2L7y/Lq392qVqJYYkciGSaG7109zq7+kwvpJ27GsG4O0lG9LaT0rksSSRC5HEOTe/ykW/eKOnw/hK8rV1AFBQkCoNRZfA2wNBVqzdzWdrduUosvyTlQFBQnyVtfg6ejqEr6S6xtaM9nvslZX4241S+Qv3zsLlkvKoHAGRd557dw0LV2zN+ftYZwbdGaoGENnzxGurAGhubU9rPzOJA+yqa8lqTPlKErnIO4++spK7Hl1KplMw29XiiySYBctyf+H4umpo9GW8796GzEr1XzWSyEXemvfJZlvPa2hqY099+iW3hqa28M/PvrM6qiQoIuYu3sd1v/8g7f2cDuP/ptZ228e2qFN9uvVi+3UmiTyLVqzbzZLPt/d0GN2mvtHHrNkv4l21o0fef+sue32Qr7r3XS76xZt0dAR46k3N5Xe/TcDGOG9rIm9s8fPMWzrjWPNFY4ufH93zDivW7ra9zyfrmtAb9+Hzd6R1l1TsLggn5h/c9hq3/t9HcZ9nfclrzp4W9bt0q2W+qiSRZ9FND33InXMW93QY3abmyz0A3PPY0m57z0079od/brfZDa0udOt++18W8eQbq9i2u8lWSc6ayAE272xMI9L8tHF7A1t2NXLTnz5Me99r73+f06+bS4eNi6S/PUBzazsjBpcDRql8mY1eKOWlRVGPpWuoQRJ5ltg5eRPu2xHgtYXr8+7WvaTI6PRUWlyY8Ws0t/r5ckt96ieGmF3WAAb2LU7rvayJ2U5Jbn9zdCJvbeve0t9Hy7fyRe3ejPZdWbuH1RvTX0q3KyVc8yL7yoIvUz7XrNuePLYyrfdwOKIfP/jsZ2nt35POuvFlHn/ti5y89lc+kS9fuyvtAQdg9Fi469ElLKrZZuv5Ta2RJPz2kg3Mmv1iTCJI5JGXV/LQvz7jjUXr046zJ5lzXmTS4LSz3s9PfjePn/1xAT/53TzbgzyClj7FzTbrR8uKjQvOoP6RdR4v/uWbKfczE/+kMf0BqN3aYOv9suXux5Zy/YPzMyp13vDgAmb//gPbF5/6Rh8btjewdVfkrsNO4SReVcpfXoy/HJnV7lBvk6kTBqZ8PStH50wOfB66M+zN/O0dtLZ18M+3V7OzLvt3EV1K5Eqp25RSn4f+3ZutoLLp5j99xO+fWZb2fr62Dhau2MavHlli6/lNrZGS4t9fNa66azbV2dr3xQ+MIcqxp2jvZk2+6ZZW561o4Mst9eHkaDcpW5OLndJjMBikNVSK79xVLVWiamhqw+mAu686BjAuWD1x15TOHUtn/3x7ta3n3fHXRVz9m/eikrDPxmfalGEJ3izkDBtQFrV9+ZrYunlrbo/3HfnZHxfYavPIlq27Gpk1+0XWbrb3/QZYtznyGZaXuLIeU8aJXCl1AnAScAgwDfAopc7IVmA9rd1yYvjbUw8IabSUyEtDJcCX5qe+xbRqacuvgSd76iMl8aaW9EoZJUXRp57d/Ts6rIk89T4NTW3hhN0Yeo/KCmPR710p+obvb2qjvLSIAsuAk/nL7PWU6apWy4Ut3Z4Z1lKt3X3jFTpabZyP8boOmncwyZgX4c7Vcrvqkn8mzjglcoBzb3015Xtmy8dfGI37by6yf6f/0/+dH/65pCj7RbaulMi3AbO11m1aaz/wBTA6O2H1PGtps3OjVzzWEnlZiXFyJjjnYqjQib8ujSt8T9u6q5GHn18Rfry/Ob1EXurOLJFbq7q+WL835a249WKzc6+RJI6fbpymW3Ylb7xsaGqjosxoXBs7rAKA+5/61FacXWUdgGQnoVrVWZJrcVHmg7ft3GXVN8Z+N+x8X1pCF+HS4gIGWNo6Ut5lJfhONbe24/N3U0EoFIPdxnarn19yRNzqoa7K+FPWWn9u/qyUmgicDRxtd/+amtT1aIl4vd6091m4eGlMH9REGpo7okrk3k+XM6BP8kNlLZHv3GM0+rQ0NdiKtaHBSCi1m3dl9Lelku3X7AgEufPpLdHv8WkNe7a6bb9G5zvhT5d/Tt2O5I2XT87bzeqtkcRcu7WBvz47n0PHlyXc55WlRoPfiAGFbNljJI/W/UbviM9qNDTFlrDN47V5+24cQePxIWNdrN8W/ftc2ro3kgz/9sIyrpk11Pb77mmIXBS/3LAVrzez0Y+fLlvBtv5FSZ+zarPx2sMrC9m613jfvQ3NKWNd86VRpbZq5QqKCyLfnbW1G/GWRRdoNm+J9FRasyZxVdG8BUsZWBFdws/FZ7XwU6NOvr5uj63Xt87W2Fq3kVK3K+txdXmuFaXUgcArwE+11mvs7lddXY3bbf+Lb/J6vXg8HlvPDQSC8KTxRa2t68N5356Ucp9/vbuGx15ZGbVtYtUkxg3vm3S/Nz55J/xzXZNRMigurbAVq+vtdwA/W/b4maCq6Vue/nFJJNXx2r6nib++WMMPTpnMmFCpM5V4Le/DR43FM3W4rf0bm9u4/cnXABjQt5g99a1UDh6NxzMq6X63P/li+OdRQ/qwacd+5i7ex2VnH5tyn+qJw9myx7gVPmzaZF5aspDKQSPweMZHPd96vO546kU8k4bg8XjYzyZYaFwU7J5/XVFauxfYCUAHrrTe12jcN27/C4v7pNyvvSMAT25m1JByNu2I3KWMG1/FlHEDku67278B2MONFx/DNb99D4DWtiAHTzskqkqqs5rtKylwNXLE4Yexdu8qng710e9fORiP58Co525oWAufGnXMSikK5u2hvSPIHZcfyW1/Xhh+3uixVUweF+kFYydX1Df6cDod9ClNfsGyeuKDeUALS1Y3cesVJ6R8/luLNwBGwWfGUYenlcOsfD5fwgJwVxs7jwbeAX6mtX6sK6+VC9ZbPLPeOpXOSRyiu7wl0tgaCN+Gm+z2Nmjzd1DiNr6sdhunsuWXcxaz+PPtXP3b92zfmpoxThrTnxsuPAywX7XS0RGIqtKYc8tJFLgcUf3DEymzfIbXnnuIrfcbNcTop3zcoSPD2wZUFONyOmhoSjw0PBAIEgzC0MpSY0MOboeTabO0y0yrGmR7v9qt9eHGfZfTwb79qXsUmdVaJx85Nmq7rTry0DEcOqA0anuqHlvNrf7wd9LpjBzb+B0EIiVaBzDn1pN4+MbjGT2kT9xY0nHBba9z3q2v2X5+faOPtZaGSzs9e9pCDeS3XfqNtOOzqyuNnaOAF4DztNZPZy+k7Hlj8frwz8lKB6nYSeRNrR0xJ7PdPrlt/gCjhxql4bnzv+SsG1+2Vc+YDdYv3Be1qbtxWeukLzx1Cp5JQwCjlG3HpXe9zXV/iDT8OJ0O+lcUR3Vh7AgE49Y/Wu9Uhg4o4/jpo1L2JXcXFeCZNJgRg8ot21xUlBXFrd81+UPvP6Cf0WWxu3sUWXvH2Dn/TNYL4oxpI9i4PfUF0jxPy4oLo3qR2Om1Ut/YRqHLEVMXvzvFZFbNvnZLIo9sX7Eu+YhSh8NB/z7FDB9YzsB+JfzwO1MYN9z47tQl+Tyz5eehOwAzn9i5eJjf5YMn2r8gp6srJfLrgGLgd0qpZaF/P8pSXFkx94NIr5GuzMlgp9GnsTVA/z7RScVulzqfv52B/SJ9nFvbOnh7ycb0gsxAR0eAvQ2RE9HOyEVrgp06YSDFRS4KXA5bJXLvqh1xv+BlxYXhUmEwGOQ/bniJq+59N+Z5jS1+TjlqLC/d9136lBZRVlKYsvtbQ6OPirKicAM0QJ/SIvqWu6lPMllTW+juxG67SraZCaK8pDCtRG52c3M5HYwcUk5rWwfrtyXv/94Ubngs5M4fHcXpxx4AwF2PLk3arS8YDPLiB+vwx1mx538eSD73SktrO6Vu4zPp3BMlWQN25xuj739rIvf9xKhaq7Nx99FVZldQ826gbn/qRF7f6KO8pJDCHJ5LGb+y1vonWus+Wutpln//l83guqqirIjpU4bgdNi7TbSWgn774xncfaXRdmunyqGptYN+fdz8+cZInVmLjaqVYDCIzx9gkCWRh36Tct+u2ro7eq4SO8nYLL1d/r2pgFFCKi8tsjX4af6y6AbSY6eNAIxePubrfrR8G4FAkK27jbp7UzAYpLHFT7klIZcXF9Lia086mMjoeeKOSsilxQWhi0DivzecyAtdob8z8rvu6Etu9o4ZNaRPWn309ze3UVzk4oXfnB6u9zXrrhMxqwDLSgoYUlnK2cdXhX/XlqTr7fY9mU/t29TqpyRUIj/8wKFRv3vPm7iLpyPOvVFhgYs+pUVRPZTsyGSgVd9y45geM81oD9pnI5HXNfqy2u4Vz1d2ZOeSldvZvLOR0UP6UOwuiOqXm4j5wV7+vamoMZUMqSwLbU++b0cgSFNrgH7lboYNLOOS06vD+6XqHre/2U8gEGRQvxKuOGNqeHuOZ2gFouv3StwFNLakTsbmBdGs0wfoU1pIo42LgMvppG95EX3Lixg9qIif/sCoX7eWyDftjFQFLF1pTEC2bnMdp183l0AgSHlJpB3CLGUnuvPx+Y3RdBVlRTgcDp6/dxZP/fJU4+JTkjxm8+8sLoodvNGd83uUlRSm1a1ux97mcFfJAw9I3lBpamqJ7tNtTTrtSS5anUf0njFzAsccHGnwTtaltLm1nRK3kcjHDK3gpfu+G/5dsrnfHQkyltlono5XP1qf1vMBpo4fSGlxAcccbBRC9tiYD72+0Ue/PpLIM/Lux5sA+NZhoyguctkqkZulyj6hRsv+FW6cDlKeILVbjdstcwj4944bz4WnTqYjEAw3dCRifhn6VxRz2jEHMPe3p+N02K+W6Yr6UGniktOrKXG72JDiFhwiVVQl7kjJeNOORj5cvjXlxbLF105ZcSF/velEzpsZGZpdVlJAYyg5mg3GwwaUsXV3Exu3N/CrRyOjazdsj8RoJp79oTrI1rZ26ht94RK6ud0sRRW4nOESfVlJYXiAUDxmtUtFmTv0GpEvYrKSfDZ0dARwF7k4Yfpo3EWutKpWtu1pChdAxgyt4LDJQ6IuuvGYK9lXVkSqBgtcRsk32flrJupZhxvjIC6edSA3XDida889FIC1SUY2t7S2x3RAMC+abZ0uXKlGdkIokTek183SWtVhdzqNBZ9tZfig8nBV6KKa1LOd1jf66Ccl8swMqCimtLiA0UMrKC4q4M3FG2Ju7TtbHPpQ+pQaX/YCl5PS4sKUJVWzoW/k4EgruplknkgxSY6ZyM0vkcPhoLS4kOY0R0pmYutuo078mIOHs7fBx2drdsd8iTq782+LACiOkxx+99QnSffdta+Zyr7FFLsLKC6MnHpllr/XLAGedsw4wPii+P2RZHLW8RPDP5v1yE+8bqw0c9aNr3DBba/zvetfYu2munAjU+feRADlpYU0Jflcw/uGLgIHTxzEt78xBoBV69OfjCod+/b78LV1UDW6n+1CCBgNjLv2tTBhVL/wtqnjB9Di60h6F7FhewOVFcVRifzqs4zpYpOVqs0L4dgh0UnKM2kwAOu3Jy4YNLa0RVWTAdx00eEAvJxk0q1Eg2kG9C1hb5olcmsi32GjmsgsIJS6C8L7Llm5PeX0AHX728KFiVz5yibyxhZ/+NbbHbrS3/v4x0l7gzwa6npo7VPa2OLn5QW1SUdxmWs6mreKEEnQL7y/Lum++8Il8siXobSkMFxCzaWGcInVzWGTjd4nN6eYvnTnPqPUY+2lcOY3JwDJG36CwSCbdjZGXexMRh25UcVkHqsTDh9NgcvBxu37w4vzzppxQNT+xx5idCksKow9je+csyic6M1StdXAviW0+DrCja8PPfcZs2a/SH2TcVdh9mixXgS+Gernfn+KC1ZXmaXDinI37kL7JXKzSmL00MgxGh7qreNdtTPhftt2NzFsYPSgKjOBXxmn0bnzc4oLo5NrRVkR7iJXwiqSjo4A+5v9MfXGE0cbJftka6Qm6gVaWVFMXaMvrdGW1qohO50hzDakI6cOA+DEw40Rwtv3Jp4X3/hb26REnqm99a1UhnqRWD+wT3X8E9p6VY03OOCM619K+F7mSWAtpZ561Njwz58keE9rbJWWHi/lxYU0t6RftbJ64z5+OWex7ca4hqY2SouN0sV//8dBAKzasI8Pl6de1my45Yt/0WkHctRBw9i8szFh6aSu0UdTi59Rg8tjfldaXEggaFSNmN3+igpdzJg2gvc/3czuuhaOPWREuIHVNLBfCf36uMNdwZxOBwdPNKps9jb4+EQbIzjjlciHhPqH729uo26/j9dC9aWLVzeGjo0vZt9c13OarHcSxUUFtroBAuE7F2vD7vTJQ3AXuVi1PvF0uNt2N0V9nhDd7z7R6kpmFVNxp3lzHA4Hg/uXhqdE6MxMiJ0/l/KSQo44cCgj45wjAFPGVTJ2WPyBeQP6FhMMwr6G1I2P7R0Bnn5L88xbkTEbdhJ5sy/Suwfg9GONwWTzkjTOLvjM+C71lTryzOyub2FAPyM5Wkdlrt4U/7bYOj+FdbrTqNdM0LAR+eJFPqwBfUt4/PaT6VNayLtLNyWMc29DK6XFBRRbSvOlJQUZ1cPO/v0HLP58e7jKJBXrXCKDK0v5f/9pDLL5KEEiDwaDOB1w9MHD6V8R3dXyiAOHsr+5LaoO2xHFKp0AABZLSURBVMrs2pioRA5Go5t5EXI5HXyjepjlveP/DWZDaXtHgEAgyNQJA8MDW56ftxaIn8jN3ig+fwevflQb3v7RF410BII0NLXhLnJF3XlY+6LnUvh8KjVKtm3tAQI2Wr/NHibm3wbgcjkZUlmacJHiFl87+/b7Ykrkfcvd3HrJEUDiqqTGZj8lbhcuZ2wxefjAspheUab60EWyb5w7pVFD+rB5Z2PcqZHvuPzIhF34zPla7NSTf1G7l3+EquPCHRNsdYYwG4WNc8JsVH7qTR23ALN64z5++w9jKP4BI5KPDO+qvErk9Y0+1m23Vw+2p76VgX2NhHzDDw7jyjMPYvLYStZsjN8AY86Ed9n3qqMGDz1xx8lcd74xnDbRtLR7G1pxOaNHHoJRgpsxbQRLV27n0Zc/jzvt5b4GX0z/84qyorRHqS22TCa13+ZgImsiB2MyqerxA9i1L/GXPhCEKksdrGlSaIGARA1c5jzX8UpbZSXGcWtq9fNeqJHa4TAS+aFqcNK/qbzESOSRft8uLp4VPcS786oyAO7CSMPal1vqGdS/JPzFnPNSDS+8vy6tYdvZZG10N+Nsj9NXuzOzYbJzsqusKE5Yf2w2dHZO5AAHheYKt/Yksmpq8VOWYFGRYQPL2LG3OW6vrQ8+Ndqq4l1gZ3qMO4FbLMu+2enAZTY+frQ89foB1gU3ZpjdCG3Mqd/SKZEDHBpqD+h8oVyycjuzLeuYThgZ+53JprxK5I+8/DmPv7s7ZSf85lY/Lb728FW6vLSIU44ax5hhFazasDfuh2aOiDtoQvToq77lbo6oHorTEX9e6EAgyPPz1lJU6IzbEDOtajBt7QGee28tj7z0eczv9za0RjUyAfQrd9u6RTT9+721/NIyb7qdvq1gVB90rj8eMaicTTv2x73VTNZ4OKSyjAKXk4U12+LeuWzd1URhgTNq4JPJTAZNLf6okr7T6eCCU4z5cRLdJZm9T3yWft8l7gL+dsuJnHNiFXdcdmTcEqNZr97mD7C3oZWRg8p54H9mUup2hgeSxfs7zO51due/3lPfwq8eWcy1989j4YrUVVYQuWj1KS2K9ORot7d8GsRP5LvqmuOOvjVLzZ3nBQejHWRQ/xL+8foqlny+Pab+ua7RR0WCut/B/UvxtXVE3emazCke4jWYjwmNcN60Yz8PPhu9jkCyWQPNkdHbUtyNtra1h9vCbvzh9HC71pw4383OzAbjUkuPrXNPVAAxPb4654qujCy3I68S+fQpxsCBVMN/zdb0ziWxscMqCAbhwjveiOmba85PETswxzihA0F4+i3Na5bbcIB1W4wSaIsvfr101ejIlXj5WuMiZJ1rZN/+1qiGTmObj8YWf7hbYzLLVu/kkZejT0I782tAbIkcjBFrjS1+zr7pFTbt2G8MzgnFa8ZdWRF7jFxOByMGlbF05Q7+6843YwbpvLF4AyMGlUfNq2EKV620+ilwOaP6P08c1Z/rLziMyzrVj5v6lhexr6GVtlD9sDuUoAf3L+WCkyeHS0ydmX/Dzn3NNLb46VNWhMvp4NTDIp+XOfTbSo2pDO+XzLbdTTz9luaGBxewqGY7azfXc+/jHyfdx9TU2k5RoYvCAme4od5vI5FvDFVrWRvdwZgvZm+Dj3NvfS2mm93joUVQhieoNjLPjzvnLOacm18N3/m0+tr5+IsdMdNSmMwG1wtvfyN8Hq3dXEcgEKSsuIDiIhdVo+PPW37pd43qjjcWbUhYVdeZy+lgyrjKuIOUNu/czwvvG4u3WBeuOOqg4WlN82tWv1hL5Obf2TnOwlDirh4/gPNPTj1ZX1flVSIfHCqV7U7Q+GKK9HWO/pCOP2xUuER4798/psXXzi/+tohn31kdfn5ZSfxbRbOF+qHnlvOXFyLzcOsNxm3aBd8cGHe/AX1LmHnoyMhq4be/zo/ueYdn31lNIBBk+57mmBK5WeqzzkmSyFuWofx3X3m0MVGSzdL8/ubYRG5deuv/3f8+9/x9KVfc8w6vfVTLG4s2UFZcQPX4+ANNJltmyrvmvnls2dXIzn3NfLmlnqYWf9JSNRglUX97R7g6wTTjkBExn6Vp5OA+7K5vDd9lFRXaW31lYKj95E/PLWfb7qZwNcqBo0u44/IjeeDa47jziqNiYw19iZ97b23C1w4Gg/zqkcX84/VV7LA0+LV3BG0tUtLia6c09Pe6C43/7STyZ99ZE4ox+hwebZnVcvYDkeXfdu1rCV+cEx3fq78/jQtCiajN38EP73iDFet2hxvwD58yNO5+UyyzEP78zwt59+ONXHv/+9z68Ec0tbZzyenVCUvYpx1zAD8KNb7f/9QnLPgsebdh0+Sxlazf1hBTNTj79x/wt7k1NLf6w4ujm+1BTqeD044Zh8vpSDl4z+wiW2JJ5KXFhQzqX8IHn24JLzm3dVcjtVsbKHEXcPeVx/CfoVJ7LnV5GtvuNCBU551qgI5Zl9X55Cx2FzDnlhOZ89LnvPD+Os6+6RUAlq40pvy8PjTSMJ5rzp7GRacdyN2PLWHu/C+ZO/9LigqctLUHGNC3mAnDEk/eNPt8D9eeeyg/vu89NoQmMfr7q1+El4TrnEyPPWQkv3nCS5u/gy9q90ZNzWkVDAZZvnY33/SM5H/OM+rx+/Vx2yqRt/k7aPF1xLz3uOF9+dZho9i+p4mVtXtZuMKoc3zoueWAMUNeomT5X6dNYca04bzw/jqWrtzBj+55J+r3iU5oswfJA08bw9KnTxmSMn6TWSIyV2Cxm8jDoztDX07z1tfhcITr5eOZ6RnJH/65jNcXrmfTjv3cecVR4aqM7XuaKHEXcMFtryfcf+GKbeFuk4m0+trD1Q5m0vClMS2Au9NoVM+kwVx46mT0hn0s/nw7Z91onPcTRhoNcL+4/MiErzVhVD8mjOrH2SdUcdldb7NjbzM3PWR0UXU4jN4tny3bFbNfYYGLu/77aG7604fsrmsJ3/EuX2uUiJN9xi6ng+8cPY4V63bz4Wf2qqPAqBZ97r21XPzLN/n3r2eFt5uNlOfcHFlFyFqq7lfupiMQ5O0lGznxiDEJX98skXe+UE4Y2Y+FK7bxsz8uwOEwzqFAIBi3GjFX8qpE3rfcjctprHFpNu61+tqjJlyC+LdAJofDwSWnV4eHw5vTnEKkcSceh8NBRVkRP7twenib2bg065gDUsbudDq4/bIjmTy2kpsumh7Vle47R4+Lef6Fp04G4PoH53Pd7z/g5j99yHPvruHD5VtpbvXz9Fua06+bS91+H8pyi1pZUZx0rgpTuEEtToPeteceyq+vnsH1FxzGsYeMiJrH/bQ4sZpKiws5aMIgfn7JN7jxh9NDowqNz2Dc8AomxmkkBSOJmoNtIHJhtWPq+IH0t3TtijekPpFf/XdkHRS785kUFrjCA1c+/3IP/3HDS1x7/zxmzX6Ry+56OyaJz5pxAE/ccTL/+MUpgL11XFt8keHr5jnc5jfO7Y5QVVdzq5+d+5rD9dbmrIFTxlXGlHSLiwo46/gqbv6vwzl+emTO97Wb6xk2oMzWrHwOh4OHf3Y855xYxbTQ80+YPjpp3e/UCQN56PpvhR/fduk3qB4/gDNmTggXypK54QeHhY81pJ6B0jrd73/c8BL3PLuFh59fHvO8iaP6RfWIOimUvJcnmHnR397Bs++sZlddCwUuR0wbxFXfPzg8CCoYjLSfWPvz51pelchdTgdFhU627W6KatwDo4HH3x6gqNAVrsdLdLsIxu3bKUeNC8/Z7PcHbJXm+pa7eem+74Y/rE079jNmWIWtFT8G9ivh3mtmhB+ffOQYCgviv+dZx1ext76Vlz+sRYda2c3STGeeyZHSTVOLMbCmdms9BS4ndU2RBLV2cx2LVmxj7vx14UEX8RouTTMOGcGMQ0YQCAQpcRewv7nN9uITRx00nKMOMqqIOjqM7nPJGquuPmsagUCQt5ZsjJkXO5mykkL+fvvJPPzv5bz8YS0HjLDfO+CAEX15+pen8ucXVthadMR05NRhvPib03nuvTW8tXhj3JGXpxw5lvNPnhQ16GXssApeeH8dRx80nDHDKqLOzzcWbWDx59voU1pEi689XHdrjn58/L3dLFwzjy+3NsQ0tPYtj0zJm6wU6HA4+H//eSg/PvsQ/B0BVqzdzcRR/eK2W8Tjcjm54GSjgNEcas9IZdSQPsz97ek0t7ZTVlIYHnhmh8Ph4Mipw3jmV6dSu7Uh5ffT6XTwwm9O54Kfv0Zji59Wf5CXF0TatNxFLh6//eSYvNC/opjB/Uvijt1Yu6mOn/95YVTBp/N53Lfcze2XHUlzq5/3P9nMcYeOZNe+lqTfrWzLq0QO8RsVD544kMqKYvztgXAHfEieyIFwb4bO3f/sME9+u4ktnkRJ3HTJd6s5/MChTBzVj8YWPx2BIOs21/HnF1bgdDg49yTF6KEVDLX0OLjyzIO55eGP+PF988LbHnjxRYYNKGPbnth+vXZONqfTwfeOG5/yeYm4XE7slJPP+/Yk6hvbuOg7U9J+j4tPr+bsE6vS/vKUlRSG5wZJh9Pp4KzjqzjLMlPgstU7+UTv4sxvTog7292hajDrtzWEq4HMBr94F2izemfEoPLwMH3rggZg9J12ALstVY1Xff9gW7G7na60kmpnnRdNTsbhcCRse7L7XnYnAHM5HTz1y1Opb/Txznwv3zz6UN5eupEWXzvnnzw5bg8mMEbRLlm5nXsf/5jDJg8Oz6FiVi2GX9+V+KJXWlzIKUcZd6xjhmX+92Yi7xL5FScPZtkmJ8dMG8Fxh4wgECTqwzlj4z5m//4DClzOqASXjwpcTg4JfaHNHjgjBpUnrWOdOmFgzJJdk8b0p7DAxUzPSE6YPtqo5wwNxe/OUkMqA/uVhAehpKuwwJnRBTmbplUNZlpV4vr1c09S/HtepJG0ta2D1Rv3MbBvMWpMJRecMonr/jCfphZ/uBDicjn57U+O5a35y5isJuAujE3An63eRUlxQcJeIF9HfcvdjBnspn9FcdTFNpGrzjyYax94n/nLtsTMyTSkspTzT57E7578hIYkc9j3pLxL5MMqizjtxMh6d50vkFWj+3PXfx8dNXHQ14nT6eDB676F3rCPCaP6sfyzT2PWBxxcGekyFm8eEpEbxe4CHrn1JPqWFyW8G5s8tpKPv9gRtbL8mKEVHHJAGZ6D4q+JenAaS8GJ+CaM6sc/7/oO/vYAG7c3sHV3E55Jg+lX7sblcqI3GFMc2Bw+0O26lMiVUucBtwCFwANa6z9mJaoumpqk0fLrwOl0JOzpYjrr+Ils39Oc81nZRLRUPRnMOU86zwwocq/EXUCJG6rHD6R6fHQOMe92zMVQepuME7lSagTwK8AD+ICPlFLvaa1jVy8Wvc6Fp6ZfDy1y7xtThzF3/pfhmQBF7+BwOHjyzlNStrv1lK50PzwBeFdrvVdr3QT8C/h+dsIS4utp6viB/OMXp3SpIVLkRp/SopwPtc+UI9VopkSUUjcCZVrrW0KPLwUO11pfnmw/r9c7FqhN9hwhhBAJjfN4POutG7pyn+AkemIyB2B7+Fl1dTVud/oNbV6vN6bxrjeQuNIjcaWnt8YFvTe2r1pcPp+PmpqauL/ryn3CZmCY5fFQwP54WiGEEFnRlRL528DtSqlBQBNwJpC0WkUIIUT2ZVwi11pvAW4G3gOWAU9qrZck30sIIUS2dakvjdb6SeDJLMUihBAiA72zL40QQgjbeqJ3uwugrc3eupLx+Hy9c74DiSs9Eld6emtc0Htj+yrFZcmZMfM7ZNyPPFNer/cYIPXSN0IIIeKZ4fF4Flg39ESJfCkwA9gGpF73SgghBBgl8WEYOTRKt5fIhRBCZJc0dgohRJ6TRC6EEHlOErkQQuQ5SeRCCJHnJJELIUSek0QuhBB5ThK5EELkud65AF0CPbnYs1LqNuDs0MNXtNbXK6UeAY7BmMYX4A6t9fNKqWnAX4EK4APgR1rr9hzG9h4wGPCHNl0B9AF+B5QAz1hWcuqW2EIrRl1t2TQOeBwooweOmVKqAvgIOE1rvV4pdQJpHB+l1GjgCYzjrIHztdaNOYjrcuDHGIu2fAxcobVuC51/FwP7Qrv+RWv9x26MK61zPVdxdY4NmALcZfn1CGCx1vq07jxmCfJDt51jeVMityz2fAwwDbhcKdUtKwiHPpCTgENC7+1RSp0BHAYcq7WeFvr3fGiXJ4CrtdZVGCsnXZbD2BxAFXCwGQewHJgDfBeYDExXSp3SnbFprf9qied8YCdwOz1wzJRSRwALMI4TSqkS0j8+DwEPaa0nYSTYW3MQVxXwU+Ao4CCM7+dVoacfBvyn5biZhZicx2V5/3Q+t6zHFS82rfWrlvPsZKABuNYSc86PWYL8cC7deI7lTSKnZxd73gbM1lq3aa39wBfA6NC/OUqp5UqpO5RSTqXUGKBEa70otO+jwFk5jE2F/n9TKfWZUupq4HBgjda6NlSqfQI4qwdiM/0JuAlopmeO2WUYCdFcwSqt46OUKgSOxTjnshlf57h8wJVa6watdRBYgXG8wEhKN4WO24NKqeLuikspVUoan1sO44qJrZPfAP+ntV4TetxdxyxefqiiG8+xfKpaGY5xwEzbML6QOae1/tz8WSk1EeMWagYwE7gSqAdeBi4BauLEOTKH4fUH3gGuwahymgf8OkEM8Y5hLmMzSyslWutnlVIHAO/SzcdMa31pKBZzU6LjkGj7QKDBUtWTlfg6x6W13gBsCG0bhFE1dZFSqhz4FKO0vhbjS34r8GB3xIWxjGM6n1tOjleC2Ag9nojxfTR/323HLEF++F+68RzLp0TepcWes0EpdSDwCvBTrbUGzrD87n+BC4GV3Rmn1nohsNASx9+AX2DcfnaOoSeO4RUY9YRorb+kFxwzEh8Hu9vJZXyhasTXgL9preeFNp9q+f19GLftD3VHXBl8bt16vEIux6iW8AGE6pa79ZhZ8wPQTnTVVE7PsXyqWunRxZ6VUkdjlHx/prV+TCk1VSl1puUpDozGxm6NUyl1jFLq+E5xrE8QQ3fHVgQcB8wNPe4VxyzJ+yXavhPoq5Qy54Eelqv4lFKTMBryHtNa3xnaNlopdbHlaeZx65a4Mvjcuu14WXwPeNp80N3HrHN+oJvPsXxK5G8DxyulBoXq7M4EXu+ON1ZKjQJeAM7TWpsniwN4QCnVP1S/dTnwfOj2uDX0wQL8AKN0lSv9gN+E6v/6AD/EqI9WSqkJoRPjPOC1HojtIGB1qE0Des8xW0waxydU7zkfOCe0/cJcxBf6/N4EbtFa32f5VQtwr1JqXKhx+yqM49YtcZHm59aNcQGglBqIUX1Xa9ncbccsQX7o1nMsbxJ5Dy/2fB1QDPxOKbVMKbUMo2fB3cCHGLeYy7TWT4Wefz5wv1JqFVAO/CFXgWmtX8a4nfsU8AJzQtUtFwHPhWJbRaQRpdtiAw7AKIGYsS6ndxyzVtI/Pldi9JRaidE+cksOQrsUGALMNs8zpdQvtNa7MKqoXsLoluYAzESf87gy/Ny643iZos6zUMzdeczi5YeL6MZzTOYjF0KIPJc3JXIhhBDxSSIXQog8J4lcCCHynCRyIYTIc5LIhRAiz0kiF0KIPCeJXAgh8pwkciGEyHP/H+4zXJ/JkgslAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(mahalanobis_dist[:2000])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([4778., 1057.,    9.,    0.,    0.,    0.,    0.,    0.,    0.,\n",
       "          12.]),\n",
       " array([ 0.08191028,  1.686772  ,  3.2916336 ,  4.8964953 ,  6.501357  ,\n",
       "         8.106219  ,  9.711081  , 11.315943  , 12.920804  , 14.525666  ,\n",
       "        16.130527  ], dtype=float32),\n",
       " <a list of 10 Patch objects>)"
      ]
     },
     "execution_count": 53,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAX8AAAD+CAYAAADS3wWuAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjAsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+17YcXAAASRUlEQVR4nO3dfZBddX3H8XdIwpICgVKkBHnoODFfkB0TuwbHYiCtYCczUnQqZkwKZBQig1jbAZ9qImhHW60Gh05TGR4appkWWrAqYmao8QGsQuVWwDXmO0wHU9KkQ2f6B6Jm80D6xzlrrptN9t7NzZ7r/t6vmUzu/e7v5HzP7t3PPed3zj2ZsX//fiRJZTmm6QYkSVPP8JekAhn+klQgw1+SCmT4S1KBDH9JKtCsTgZFxDeA04A9dendwInAOmAOcF9mrqnHLgLuBOYCjwDXZebeiDgb2Fj/OwmszMwXe7gtkqQOTbjnHxEzgAXAwsxclJmLgKeBu4HLgfOAxRGxrF5kI3BDZi4AZgDX1vX1wPrMPBd4Aljb0y2RJHWskz3/qP9+OCJ+A7gD+AHwTGY+CxARG4ErImILMCczH6uX2QB8LCLuBC4C3tJW/xbwwYlW3mq1BoDFwE5gXwf9SpJgJjAP+N7Q0NDI2C92Ev6/DmwG3gvMBr4JfIoqjEftBM4EzjhE/VTghczcO6beicXAox2OlST9siXAt8cWJwz/zPwu8N3R5xFxF/DxMf/YDOAlqmmk/R3Uqeud2DnxEEnSIYyboROGf0S8ARjIzM11aQbwY6rDiVGnAzuA7YeoPw+cFBEzM3NfPWZHh43vAxgcHGRgYKDDRSqtVouhoaGulpkK9tUd++pOP/bVjz3B9O5rZGSE4eFhOMR0eSeXep4M/FVEHBcRJwJXA38GRETMj4iZwApgU2ZuA3ZFxIX1slfW9T1UUzfL6/pVwKbJbpQk6chMGP6Z+RXgIeD7QAu4u54KWgU8AGwBtgL314usBG6NiK3ACcBtdf16YHV9UngJsKZ3myFJ6kZH1/ln5lrGXJpZTwMtHGfsU8AF49S3AUsn1aUkqaf8hK8kFcjwl6QCGf6SVCDDX5IKNO3Df/ee5u4I0eS6JelwOrra51fZsbNnctmNX2pk3Q9+9vJG1itJE5n2e/6SpIMZ/pJUIMNfkgpk+EtSgQx/SSqQ4S9JBTL8JalAhr8kFcjwl6QCGf6SVCDDX5IKZPhLUoEMf0kqkOEvSQUy/CWpQIa/JBXI8JekAhn+klQgw1+SCmT4S1KBDH9JKpDhL0kFMvwlqUCGvyQVyPCXpAIZ/pJUIMNfkgpk+EtSgQx/SSrQrE4HRsRngFMzc1VEXAKsA+YA92XmmnrMIuBOYC7wCHBdZu6NiLOBjcBpQAIrM/PF3m6KJKlTHe35R8Qbgavrx3OAu4HLgfOAxRGxrB66EbghMxcAM4Br6/p6YH1mngs8Aazt2RZIkro2YfhHxCnAJ4BP1qULgGcy89nM3EsV+FdExDnAnMx8rB63oa7PBi4C7m+v92wLJEld62Ta53bgI8BZ9fMzgJ1tX98JnHmY+qnAC/UbRXu9K8PDw90u0hdarVZX9abZV3fsq3P92BOU29dhwz8irgGey8zNEbGqLh8D7G8bNgN4qYs6db0rg4ODDAwMdLVMP/xQh4aGDqq1Wq1x602zr+7YV+f6sSeY3n2NjIwcdqd5oj3/5cC8iHgSOAU4ATgH2Nc25nRgB7AdmDdO/XngpIiYmZn76jE7utwOSVIPHXbOPzMvzczBzFwEfBT4MrAMiIiYHxEzgRXApszcBuyKiAvrxa+s63uAR6neSACuAjYdhW2RJHWo6+v8M3MXsAp4ANgCbOXAydyVwK0RsZXqKOG2un49sDoitgBLgDVH1rYk6Uh0fJ1/Zm6gulKHzNwMLBxnzFNUVwONrW8Dlk6yR0lSj/kJX0kqkOEvSQUy/CWpQIa/JBXI8JekAhn+klQgw1+SCmT4S1KBDH9JKpDhL0kFMvwlqUCGvyQVyPCXpAIZ/pJUIMNfkgpk+EtSgQx/SSqQ4S9JBTL8JalAhr8kFcjwl6QCGf6SVCDDX5IKZPhLUoEMf0kqkOEvSQUy/CWpQIa/JBXI8JekAhn+klQgw1+SCmT4S1KBDH9JKpDhL0kFmtXJoIj4OPA2YD9wV2aui4hLgHXAHOC+zFxTj10E3AnMBR4BrsvMvRFxNrAROA1IYGVmvtjrDZIkTWzCPf+IuBj4PeDVwGuB90bEQuBu4HLgPGBxRCyrF9kI3JCZC4AZwLV1fT2wPjPPBZ4A1vZyQyRJnZsw/DPzW8DvZuZeqr32WcDJwDOZ+Wxd3whcERHnAHMy87F68Q11fTZwEXB/e72XGyJJ6lxH0z6ZuSciPgbcBPwzcAaws23ITuDMw9RPBV6o3yja6x0bHh7uZnjfaLVaXdWbZl/dsa/O9WNPUG5fHYU/QGbeHBGfAh4EFlDN/4+aAbxEdSTRSZ263rHBwUEGBga6WaQvfqhDQ0MH1Vqt1rj1ptlXd+yrc/3YE0zvvkZGRg6709zJnP+59UlcMvNnwBeApcC8tmGnAzuA7YeoPw+cFBEz6/q8ui5JakAnl3q+ArgjIgYi4liqk7y3AxER8+tAXwFsysxtwK6IuLBe9sq6vgd4FFhe168CNvVyQyRJnevkhO9XgYeA7wMt4DuZeS+wCngA2AJs5cDJ3JXArRGxFTgBuK2uXw+sjogtwBJgTe82Q5LUjU5P+N4C3DKmthlYOM7Yp4ALxqlvo5oukiQ1zE/4SlKBDH9JKpDhL0kFMvwlqUCGvyQVyPCXpAIZ/pJUIMNfkgpk+EtSgQx/SSqQ4S9JBTL8JalAhr8kFcjwl6QCGf6SVCDDX5IKZPhLUoEMf0kqkOEvSQUy/CWpQIa/JBXI8JekAhn+klQgw1+SCmT4S1KBDH9JKpDhL0kFMvwlqUCGvyQVyPCXpAIZ/pJUIMNfkgpk+EtSgQx/SSrQrE4GRcTNwNvrpw9l5gci4hJgHTAHuC8z19RjFwF3AnOBR4DrMnNvRJwNbAROAxJYmZkv9nRrJEkdmXDPvw75NwGvARYBQxHxDuBu4HLgPGBxRCyrF9kI3JCZC4AZwLV1fT2wPjPPBZ4A1vZyQyRJnetk2mcncGNm7s7MPcCPgAXAM5n5bGbupQr8KyLiHGBOZj5WL7uhrs8GLgLub6/3bjMkSd2YcNonM384+jgiXkk1/fPXVG8Ko3YCZwJnHKJ+KvBC/UbRXpckNaCjOX+AiDgfeAh4P7CXau9/1AzgJaojif0d1KnrHRseHu5meN9otVpd1ZtmX92xr871Y09Qbl+dnvC9EHgA+JPMvDciLgbmtQ05HdgBbD9E/XngpIiYmZn76jE7uml0cHCQgYGBbhbpix/q0NDQQbVWqzVuvWn21R376lw/9gTTu6+RkZHD7jR3csL3LOCLwIrMvLcuP159KeZHxExgBbApM7cBu+o3C4Ar6/oe4FFgeV2/Ctg0mQ2SJB25Tvb8bwKOA9ZFxGjt88AqqqOB44CvcuBk7krgjoiYC/wHcFtdvx64JyLWAP8FvKMH/UuSJqGTE77vA953iC8vHGf8U8AF49S3AUu77E+SdBT4CV9JKpDhL0kFMvwlqUCGvyQVyPCXpAIZ/pJUIMNfkgpk+EtSgQx/SSqQ4S9JBTL8JalAhr8kFcjwl6QCGf6SVCDDX5IKZPhLUoEMf0kqkOEvSQUy/CWpQIa/JBXI8JekAhn+klQgw1+SCmT4S1KBDH9JKpDhL0kFMvwlqUCGvyQVyPCXpAIZ/pJUIMNfkgpk+EtSgQx/SSqQ4S9JBTL8JalAszoZFBFzge8Ab87MH0fEJcA6YA5wX2auqcctAu4E5gKPANdl5t6IOBvYCJwGJLAyM1/s+db0md179nHs7JkH1YeGhhpZrySNmjD8I+J1wB3Agvr5HOBu4GLgOeChiFiWmZuoAv6azHwsIu4CrgX+FlgPrM/MeyNiLbAW+ODR2KB+cuzsmVx245emfL0PfvbyKV+npF8tnUz7XAu8B9hRP78AeCYzn83MvVSBf0VEnAPMyczH6nEb6vps4CLg/vZ6b9qXJE3GhHv+mXkNQESMls4AdrYN2QmceZj6qcAL9RtFe70rw8PD3S5StFarNaXLHW321Z1+7Ksfe4Jy++pozn+MY4D9bc9nAC91Uaeud2VwcJCBgYGulunXH+pUmMx5hVarddTPR0yGfXWnH/vqx55gevc1MjJy2J3myVztsx2Y1/b8dKopoUPVnwdOiojRM5DzODCFJElqwGTC/3EgImJ+HegrgE2ZuQ3YFREX1uOurOt7gEeB5XX9KmDTEfYtSToCXYd/Zu4CVgEPAFuArRw4mbsSuDUitgInALfV9euB1RGxBVgCrDmytiVJR6LjOf/M/K22x5uBheOMeYrqaqCx9W3A0kl1KEnqOT/hK0kFMvwlqUCGvyQVyPCXpAIZ/pJUIMNfkgpk+EtSgQx/SSqQ4S9JBTL8JalAhr8kFcjwl6QCGf6SVCDDX5IKZPhLUoEMf0kqkOEvSQUy/CWpQIa/JBXI8JekAhn+klQgw1+SCmT4S1KBDH9JKpDhL0kFMvwlqUCGvyQVyPCXpAIZ/pJUIMNfkgpk+EtSgQx/SSqQ4S9JBTL8JalAs6ZyZRGxAlgDzAY+l5l/M5XrlyRVpmzPPyJeDnwCeAOwCFgdEa+aqvVLkg6Yyj3/S4CvZ+b/AUTE/cDbgI9PsNxMgN27d09qpSMjI5x8/MxJLXukmlr3yMhII8seTfbVnX7sqx97gunbV1tmjhtCM/bv339EK+hURHwYOD4z19TPrwEuyMzVh1uu1Wq9AXh0ClqUpOloydDQ0LfHFqdyz/8YoP2dZgbwUgfLfQ9YAuwE9h2FviRpOpoJzKPK0INMZfhvpwrxUacDOyZaaGhoaAQ46F1LkjSh/zzUF6Yy/L8G3BIRLwN+CvwhcNgpH0nS0TFlV/tk5n8DHwG+ATwJ/ENm/vtUrV+SdMCUnfCVJPUPP+ErSQUy/CWpQIa/JBXI8JekAk3pjd2mSr/eQC4ibgbeXj99KDM/0GQ/Y0XEZ4BTM3NV070ARMRlwM3A8cDDmfm+hlsCICL+CPhw/XRTZt7UcD9zge8Ab87MH0fEJcA6YA5w3+in6vugr9XAH1N92PMJ4N2ZObn7tvSop7b6DcDbMnPpVPZzqL4i4vXArcCJwNPA1b3+Xk27Pf9+vYFc/Qv5JuA1VH0NRcRbm+3qgIh4I3B1032MiohXAJ8H3gK8GvjtiFjWbFcQEb8G3AZcDCwEltQ/26b6eR3VhyAX1M/nAHcDlwPnAYub+L6N09cC4P3A71D9PI8B3tNkT231VwEfmspexqx/7PdqLvAFYHVmnl8Pe1ev1zvtwp+2G8hl5k+B0RvINW0ncGNm7s7MPcCPgLMb7gmAiDiF6g3zk0330uatVHut2+vv13Lg8YZ7guoj88dQHY3Mrv/8vMF+rqUK0dFPy18APJOZz2bmXmAjcEUf9DUCXJ+ZL2TmfuAHTP3rf2xPRMQAcDvw0Snupd3Yvi4FvpuZT9fP3wv8S69XOh2nfc6gCtpRO6l+IRqVmT8cfRwRr6Sa/rmwuY5+ye1UH8A7q+lG2swHdkfEl6lC4ivA2mZbgsz8SUSsBbYCPwO+RXW43lQ/1wBExGhpvNf/mVPc1kF9ZeY2YFtdexlwA7CqyZ5qf0F1pPTsVPbSbpy+5gMvRsS9wLnAvwE39nq903HPf7I3kJsSEXE+8K/A+zPzmT7o5xrguczc3HQvY8yiOop7F/B64HX0wbRURLwaeCdwDlXQ7gManfMfo99f/y8HNgN3ZeY3G+7lUuDszPy7JvsYxyzg96nOKw1RHWX2fFpqOob/dqo72Y3q6AZyUyEiLqR64X8oM+9pup/acuBNEfEk1f+t8AcRcWvDPQH8D/C1zPzfzPw51WFv40dwVL+UmzPz+cwcATYASxvt6Jf18+v/XKqjpHsy88+b7gd4B3B+/dq/E3htRNzXcE9QvfYfq6fu9gH/xFF47U/HaZ++vIFcRJwFfBFYnplfb7qfUZl56ejjiFgFLM3MP22uo1/4CnBPRJwM/ARYRvX9a9pTwKcj4niqaZ/LOMQtcxvyOBARMZ9qKmMF1bRGoyLiROBh4COZ+fdN9wOQme8cfRwRS4FbMnN5cx39wsPAxyLirMx8Dngz0Or1Sqbdnn8f30DuJuA4YF1EPFn/ua7ppvpVZj4OfJrqKogtVPPFjR+eZ+bDwD9S/TI+TXXC9y8bbapNZu6imkt/gOr7tpXqooemXQP8JnBj2+t/ov/Fr0h14L8beDAitgKnUJ2b6Clv7CZJBZp2e/6SpIkZ/pJUIMNfkgpk+EtSgQx/SSqQ4S9JBTL8JalAhr8kFej/AZzlqm28svgCAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.hist(mahalanobis_dist)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "16.130527\n",
      "(array([1670], dtype=int64),)\n"
     ]
    }
   ],
   "source": [
    "print(mahalanobis_dist.max())\n",
    "print(np.where(mahalanobis_dist == mahalanobis_dist.max()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.08191028\n",
      "(array([3906], dtype=int64),)\n"
     ]
    }
   ],
   "source": [
    "print(mahalanobis_dist.min())\n",
    "print(np.where(mahalanobis_dist == mahalanobis_dist.min()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x2e392661278>"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(images[1670])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x2e392705898>"
      ]
     },
     "execution_count": 48,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(images[3906])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x2e3928586d8>"
      ]
     },
     "execution_count": 56,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "x_sampled = autoencoder.get_layer('decoder')(train_mu)\n",
    "x_sampled = np.asarray(x_sampled)[0] * flat_mask\n",
    "plt.imshow(x_sampled.reshape(48, 40))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "XYF503yFWtCT"
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "collapsed_sections": [],
   "name": "2.6 - Check for Outliers Autoencoder.ipynb",
   "provenance": [],
   "version": "0.3.2"
  },
  "kernelspec": {
   "display_name": "Python [conda env:py36]",
   "language": "python",
   "name": "conda-env-py36-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}