vae_binary.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "vae-binary.ipynb",
      "version": "0.3.2",
      "provenance": [],
      "collapsed_sections": [],
      "machine_shape": "hm"
    },
    "kernelspec": {
      "name": "python2",
      "display_name": "Python 2"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "eqLpp_XSyj0p",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "!pip install deepdish"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "IwSK8ZwnGd8e",
        "colab_type": "code",
        "outputId": "7d68a379-e265-47df-f23e-e8b96dc55f7c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "import tensorflow as tf\n",
        "import keras\n",
        "from keras.layers import Lambda, Input, Dense\n",
        "from keras.models import Model\n",
        "from keras.datasets import mnist\n",
        "from keras.callbacks import EarlyStopping\n",
        "from keras.losses import mse, binary_crossentropy\n",
        "from keras.utils import plot_model\n",
        "from keras import backend as K\n",
        "\n",
        "import matplotlib.pyplot as plt\n",
        "import numpy as np\n",
        "import os\n",
        "import deepdish.io as io\n",
        "from sklearn.metrics import roc_curve, auc\n",
        "import random"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Using TensorFlow backend.\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "mqoGQPY2J34-",
        "colab_type": "code",
        "outputId": "445c9e53-3b0f-44a7-db3c-fad5652473db",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "# verification of tensorflow\n",
        "tf.__version__"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "'1.14.0'"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 3
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "UQ04BVk7x-Et",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "f3d37At9W39A",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# reparameterization trick\n",
        "# instead of sampling from Q(z|X), sample eps = N(0,I)\n",
        "# z = z_mean + sqrt(var)*eps\n",
        "def sampling(args):\n",
        "    \"\"\"Reparameterization trick by sampling fr an isotropic unit Gaussian.\n",
        "\n",
        "    # Arguments\n",
        "        args (tensor): mean and log of variance of Q(z|X)\n",
        "\n",
        "    # Returns\n",
        "        z (tensor): sampled latent vector\n",
        "    \"\"\"\n",
        "\n",
        "    z_mean, z_log_var = args\n",
        "    batch = K.shape(z_mean)[0]\n",
        "    dim = K.int_shape(z_mean)[1]\n",
        "    # by default, random_normal has mean=0 and std=1.0\n",
        "    epsilon = K.random_normal(shape=(batch, dim))\n",
        "    return z_mean + K.exp(0.5 * z_log_var) * epsilon\n",
        "  "
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "aD6cYa_CHsLi",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# data from debo's data for example\n",
        "data_Mix_Train = io.load(os.path.join('DarkMachineTrain.h5'))\n",
        "data_Mix_Test = io.load(os.path.join('DarkMachineTest.h5'))\n",
        "data_bkg_Train = io.load(os.path.join('X_bkgTrain.h5'))\n",
        "data_bkg_Test = io.load(os.path.join('X_bkgTest.h5'))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Fs34Kp2-Jd59",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# setup of data for training/testing \n",
        "X_Mix_train = data_Mix_Train['X']\n",
        "y_Mix_train = data_Mix_Train['y']\n",
        "X_Mix_test = data_Mix_Test['X']\n",
        "y_Mix_test = data_Mix_Test['y']\n",
        "\n",
        "X_bkg_train = data_bkg_Train['X']\n",
        "X_bkg_test = data_bkg_Test['X']\n",
        "\n",
        "X_Mix_train_N = keras.utils.normalize(X_Mix_train, axis=1)\n",
        "X_Mix_test_N = keras.utils.normalize(X_Mix_test, axis=1)\n",
        "\n",
        "y_Mix_train_N = keras.utils.normalize(y_Mix_train, axis=1)\n",
        "y_Mix_test_N = keras.utils.normalize(y_Mix_test, axis=1) \n",
        "\n",
        "X_bkg_train_N = keras.utils.normalize(X_bkg_train, axis=1)\n",
        "X_bkg_test_N = keras.utils.normalize(X_bkg_test, axis=1)\n",
        "\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "mSDMwEewKCQu",
        "colab_type": "code",
        "outputId": "f24dff54-864d-4f47-8108-6f272a92283f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 121
        }
      },
      "source": [
        "# verification of proper data\n",
        "print(\"X_Mix_train_N: \",X_Mix_train_N.shape)\n",
        "print(\"X_Mix_test_N: \",X_Mix_test_N.shape)\n",
        "print(\"y_Mix_train_N: \",y_Mix_train.shape)\n",
        "print(\"y_Mix_test_N: \",y_Mix_test.shape)\n",
        "\n",
        "print(\"X_bkg_train_N: \",X_bkg_train_N.shape)\n",
        "print(\"X_bkg_test_N: \",X_bkg_test_N.shape)\n"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "('X_Mix_train_N: ', (2420718, 113))\n",
            "('X_Mix_test_N: ', (806907, 113))\n",
            "('y_Mix_train_N: ', (2420718, 1))\n",
            "('y_Mix_test_N: ', (806907, 1))\n",
            "('X_bkg_train_N: ', (1739535, 113))\n",
            "('X_bkg_test_N: ', (434884, 113))\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Qw8WOQJq56CJ",
        "colab_type": "code",
        "outputId": "dea8ad02-03ec-4cad-90e4-041c24f5a56c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 139
        }
      },
      "source": [
        "print y_Mix_train_N"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[[1.]\n",
            " [0.]\n",
            " [0.]\n",
            " ...\n",
            " [0.]\n",
            " [1.]\n",
            " [0.]]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "liF8krK5KjxX",
        "colab_type": "code",
        "outputId": "dff3e857-e0b8-4017-9dca-b166b40f046a",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "# network parameters\n",
        "input_shape = X_Mix_train_N.shape[1]\n",
        "high_dim = 113\n",
        "mid_dim = 75\n",
        "low_dim = 30\n",
        "batch_size = 256\n",
        "latent_dim = 16\n",
        "epochs = 50\n",
        "\n",
        "print(input_shape)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "113\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "mfwX_ApiiAYd",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# early stopping\n",
        "es = EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=3)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "JytcDuKUnrjl",
        "colab_type": "text"
      },
      "source": [
        "\n",
        "# Variational Autoencoder Model\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "O8oouFFqRMdj",
        "colab_type": "code",
        "outputId": "4cebc398-7ba7-486a-df97-4ecbef7f2f37",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 159
        }
      },
      "source": [
        "# Variational Autoencoder (VAE) Model Start\n",
        "\n",
        "# Encoder Model\n",
        "inputs = Input(shape=(input_shape,), name='encoder_input')\n",
        "eh = Dense(high_dim, activation='relu')(inputs)\n",
        "em = Dense(mid_dim, activation='relu')(eh)\n",
        "el = Dense(low_dim, activation='relu')(em)\n",
        "z_mean = Dense(latent_dim, name='z_mean')(el)\n",
        "z_log_var = Dense(latent_dim, name='z_log_var')(el)\n"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING: Logging before flag parsing goes to stderr.\n",
            "W0829 18:03:58.582310 140486976563072 deprecation_wrapper.py:119] From /usr/local/lib/python2.7/dist-packages/keras/backend/tensorflow_backend.py:517: The name tf.placeholder is deprecated. Please use tf.compat.v1.placeholder instead.\n",
            "\n",
            "W0829 18:03:58.622654 140486976563072 deprecation_wrapper.py:119] From /usr/local/lib/python2.7/dist-packages/keras/backend/tensorflow_backend.py:74: The name tf.get_default_graph is deprecated. Please use tf.compat.v1.get_default_graph instead.\n",
            "\n",
            "W0829 18:03:58.623976 140486976563072 deprecation_wrapper.py:119] From /usr/local/lib/python2.7/dist-packages/keras/backend/tensorflow_backend.py:4138: The name tf.random_uniform is deprecated. Please use tf.random.uniform instead.\n",
            "\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "AUbDojQmSF00",
        "colab_type": "code",
        "outputId": "77a4ac38-1970-409b-f944-59d9ee04025d",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 72
        }
      },
      "source": [
        "# use reparameterization trick to push the sampling out as input\n",
        "# note that \"output_shape\" isn't necessary with the TensorFlow backend\n",
        "z = Lambda(sampling, output_shape=(latent_dim,), name='z')([z_mean, z_log_var])\n"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "W0829 18:03:58.680694 140486976563072 deprecation_wrapper.py:119] From /usr/local/lib/python2.7/dist-packages/keras/backend/tensorflow_backend.py:4115: The name tf.random_normal is deprecated. Please use tf.random.normal instead.\n",
            "\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jNCdDM89SPMq",
        "colab_type": "code",
        "outputId": "9a6df141-6561-405b-81f1-8925dffd623d",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 454
        }
      },
      "source": [
        "# instantiate encoder model\n",
        "encoder = Model(inputs, [z_mean, z_log_var, z], name='encoder')\n",
        "encoder.compile(optimizer='adam', loss='mean_squared_error')\n",
        "encoder.summary()\n",
        "plot_model(encoder, show_shapes=True) #to_file='vae_mlp_encoder.png',\n"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "W0829 18:03:58.694518 140486976563072 deprecation_wrapper.py:119] From /usr/local/lib/python2.7/dist-packages/keras/optimizers.py:790: The name tf.train.Optimizer is deprecated. Please use tf.compat.v1.train.Optimizer instead.\n",
            "\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "__________________________________________________________________________________________________\n",
            "Layer (type)                    Output Shape         Param #     Connected to                     \n",
            "==================================================================================================\n",
            "encoder_input (InputLayer)      (None, 113)          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "dense_1 (Dense)                 (None, 113)          12882       encoder_input[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_2 (Dense)                 (None, 75)           8550        dense_1[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "dense_3 (Dense)                 (None, 30)           2280        dense_2[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "z_mean (Dense)                  (None, 16)           496         dense_3[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "z_log_var (Dense)               (None, 16)           496         dense_3[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "z (Lambda)                      (None, 16)           0           z_mean[0][0]                     \n",
            "                                                                 z_log_var[0][0]                  \n",
            "==================================================================================================\n",
            "Total params: 24,704\n",
            "Trainable params: 24,704\n",
            "Non-trainable params: 0\n",
            "__________________________________________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jc7cIZ-rSZtY",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# build decoder model\n",
        "latent_inputs = Input(shape=(latent_dim,), name='z_sampling')\n",
        "dl = Dense(low_dim, activation='relu')(latent_inputs)\n",
        "dm = Dense(mid_dim, activation='relu')(dl)\n",
        "dh = Dense(high_dim, activation='relu')(dm)\n",
        "\n",
        "outputs = Dense(input_shape, activation='sigmoid')(dh) # original_dim\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4ETgmg5MSgVm",
        "colab_type": "code",
        "outputId": "fc08f703-96c4-4140-8677-0b279de9b140",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 312
        }
      },
      "source": [
        "# instantiate decoder model\n",
        "decoder = Model(latent_inputs, outputs, name='decoder')\n",
        "decoder.summary()\n",
        "plot_model(decoder, show_shapes=True) #to_file='vae_mlp_decoder.png',\n"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "z_sampling (InputLayer)      (None, 16)                0         \n",
            "_________________________________________________________________\n",
            "dense_4 (Dense)              (None, 30)                510       \n",
            "_________________________________________________________________\n",
            "dense_5 (Dense)              (None, 75)                2325      \n",
            "_________________________________________________________________\n",
            "dense_6 (Dense)              (None, 113)               8588      \n",
            "_________________________________________________________________\n",
            "dense_7 (Dense)              (None, 113)               12882     \n",
            "=================================================================\n",
            "Total params: 24,305\n",
            "Trainable params: 24,305\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Gm-pc2NSSzpx",
        "colab_type": "code",
        "outputId": "11bc951a-6977-49be-ec7a-2789c216b586",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 243
        }
      },
      "source": [
        "# instantiate VAE model\n",
        "outputs = decoder(encoder(inputs)[2])\n",
        "vae = Model(inputs, outputs, name='vae_mlp')\n",
        "vae.compile(optimizer='adam', loss='mean_squared_error',metrics=['accuracy'])\n",
        "#encoder.compile(optimizer='adam', loss='binary_crossentropy')\n",
        "# VAE Model End\n",
        "vae.summary()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "encoder_input (InputLayer)   (None, 113)               0         \n",
            "_________________________________________________________________\n",
            "encoder (Model)              [(None, 16), (None, 16),  24704     \n",
            "_________________________________________________________________\n",
            "decoder (Model)              (None, 113)               24305     \n",
            "=================================================================\n",
            "Total params: 49,009\n",
            "Trainable params: 49,009\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "HMWVtJQjn7cV",
        "colab_type": "text"
      },
      "source": [
        "# Binary Classifier Model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vm9XR5zaUi4v",
        "colab_type": "code",
        "outputId": "3de1c4e7-5dfa-4cd9-f8d7-fefcb0252652",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 350
        }
      },
      "source": [
        "# Binary Classifier (Functional API) Model Start\n",
        "\n",
        "binputs = Input(shape=(latent_dim,), name='binary_input')\n",
        "bi = Dense(latent_dim, activation= 'relu')(binputs)\n",
        "bioutputs = Dense(1, activation='sigmoid')(bi)\n",
        "\n",
        "biclass = Model(binputs, bioutputs, name='binary_classifer')\n",
        "biclass.compile(optimizer='adam', loss='binary_crossentropy', metrics=['accuracy'])\n",
        "\n",
        "biclass.summary()\n"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "W0829 18:03:59.783804 140486976563072 deprecation_wrapper.py:119] From /usr/local/lib/python2.7/dist-packages/keras/backend/tensorflow_backend.py:3376: The name tf.log is deprecated. Please use tf.math.log instead.\n",
            "\n",
            "W0829 18:03:59.787472 140486976563072 deprecation.py:323] From /usr/local/lib/python2.7/dist-packages/tensorflow/python/ops/nn_impl.py:180: where (from tensorflow.python.ops.array_ops) is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Use tf.where in 2.0, which has the same broadcast rule as np.where\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "binary_input (InputLayer)    (None, 16)                0         \n",
            "_________________________________________________________________\n",
            "dense_8 (Dense)              (None, 16)                272       \n",
            "_________________________________________________________________\n",
            "dense_9 (Dense)              (None, 1)                 17        \n",
            "=================================================================\n",
            "Total params: 289\n",
            "Trainable params: 289\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "dUCWbZjzoOxI",
        "colab_type": "text"
      },
      "source": [
        "\n",
        "# Training Variational Autoencoder with Pure Background\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "H_b78w2903vC",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "##########################################\n",
        "# Start Model Combination & Verification #\n",
        "##########################################"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "gE2PGfM6aJ16",
        "colab_type": "code",
        "outputId": "3be752ce-1caa-40f6-b070-d22620c06fe7",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 434
        }
      },
      "source": [
        "# Variational Autoencoder (VAE) Training/Testing [BACKGROUND]\n",
        "history = vae.fit(X_bkg_train_N,X_bkg_train_N,\n",
        "                  epochs=epochs,\n",
        "                  batch_size=batch_size,\n",
        "                  validation_split=0.2,\n",
        "                  callbacks=[es])\n",
        "                  #validation_data=(X_bkg_test_N, X_bkg_test_N)) #shuffle=True\n",
        "\n",
        "\n",
        "encoder.save_weights('ve-weights.h5')\n",
        "vae.save_weights('vae-weights.h5')\n",
        "\n",
        "vae.save('DarkMachine_autoencoder.model')\n",
        "encoder.save('DarkMachine_encoder.model')"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train on 1391628 samples, validate on 347907 samples\n",
            "Epoch 1/50\n",
            "1391628/1391628 [==============================] - 36s 26us/step - loss: 0.0013 - acc: 0.9045 - val_loss: 2.2258e-04 - val_acc: 0.9505\n",
            "Epoch 2/50\n",
            "1391628/1391628 [==============================] - 31s 22us/step - loss: 1.1486e-04 - acc: 0.9629 - val_loss: 2.4332e-05 - val_acc: 0.9603\n",
            "Epoch 3/50\n",
            "1391628/1391628 [==============================] - 31s 22us/step - loss: 1.2970e-05 - acc: 0.9762 - val_loss: 1.2326e-05 - val_acc: 0.9768\n",
            "Epoch 4/50\n",
            "1391628/1391628 [==============================] - 31s 22us/step - loss: 8.2141e-06 - acc: 0.9801 - val_loss: 1.0606e-05 - val_acc: 0.9804\n",
            "Epoch 5/50\n",
            "1391628/1391628 [==============================] - 31s 23us/step - loss: 5.7102e-06 - acc: 0.9835 - val_loss: 5.6830e-06 - val_acc: 0.9808\n",
            "Epoch 6/50\n",
            "1391628/1391628 [==============================] - 32s 23us/step - loss: 4.5287e-06 - acc: 0.9851 - val_loss: 5.0488e-06 - val_acc: 0.9922\n",
            "Epoch 7/50\n",
            "1391628/1391628 [==============================] - 31s 23us/step - loss: 3.7661e-06 - acc: 0.9860 - val_loss: 5.6775e-06 - val_acc: 0.9740\n",
            "Epoch 8/50\n",
            "1391628/1391628 [==============================] - 31s 22us/step - loss: 3.2213e-06 - acc: 0.9866 - val_loss: 2.8895e-06 - val_acc: 0.9885\n",
            "Epoch 9/50\n",
            "1391628/1391628 [==============================] - 31s 23us/step - loss: 2.7957e-06 - acc: 0.9873 - val_loss: 5.1219e-06 - val_acc: 0.9871\n",
            "Epoch 10/50\n",
            "1391628/1391628 [==============================] - 32s 23us/step - loss: 2.5323e-06 - acc: 0.9876 - val_loss: 6.4830e-06 - val_acc: 0.9841\n",
            "Epoch 11/50\n",
            "1391628/1391628 [==============================] - 32s 23us/step - loss: 2.2897e-06 - acc: 0.9882 - val_loss: 3.5744e-06 - val_acc: 0.9807\n",
            "Epoch 00011: early stopping\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YXVu3-jynXLg",
        "colab_type": "code",
        "outputId": "022915ce-cd29-4cdd-f275-15ec8b3afec2",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 590
        }
      },
      "source": [
        "# list all data in history\n",
        "print(history.history.keys())\n",
        "# summarize history for accuracy\n",
        "# accuracy doesn't matter, especially for variational autoencoders\n",
        "plt.plot(history.history['acc'])\n",
        "plt.plot(history.history['val_acc'])\n",
        "plt.title('model accuracy')\n",
        "plt.ylabel('accuracy')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()\n",
        "# summarize history for loss\n",
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.title('model loss')\n",
        "plt.ylabel('loss')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "['acc', 'loss', 'val_acc', 'val_loss']\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYsAAAEWCAYAAACXGLsWAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nk5KFiIi0SclCRETa9P8B0LTLaOGlmGYAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZgAAAEWCAYAAABbgYH9AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAxTXPsoI/q3DAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DtPPpks_1fnI",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "encoder_model = keras.models.load_model('DarkMachine_encoder.model')"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Kg9oqtnJwWhU",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Encoder as input, using Mixed Data\n",
        "# Training Input\n",
        "x_train_encoded = encoder_model.predict(X_Mix_train_N, batch_size=batch_size)\n",
        "# Testing Input\n",
        "x_test_encoded = encoder_model.predict(X_Mix_test_N, batch_size=batch_size)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "qhJrTeBcoaAy",
        "colab_type": "text"
      },
      "source": [
        "## Binary Classifier with Pure Background Encoder Input\n",
        "##### (Remember: Implies classification using only latent space representation)"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "putLoqSpipAn",
        "colab_type": "code",
        "outputId": "5b15dde4-5893-4821-d352-f13015d1f2c5",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "# Binary Classifer Training/Testing \n",
        "## Classifer can be used for both VAE's \n",
        "history = biclass.fit(x_train_encoded[0], \n",
        "                      y_Mix_train_N, \n",
        "                      batch_size=batch_size, \n",
        "                      epochs=epochs,\n",
        "                      validation_split=0.2,\n",
        "                      callbacks=[es])\n",
        "                      #validation_data=(x_test_encoded[0], y_Mix_test_N))\n",
        "\n",
        "\n",
        "biclass.save_weights('bi-weights.h5')\n",
        "biclass.save('binary-class.model')"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train on 1936574 samples, validate on 484144 samples\n",
            "Epoch 1/50\n",
            "1936574/1936574 [==============================] - 25s 13us/step - loss: 0.1980 - acc: 0.9226 - val_loss: 0.1634 - val_acc: 0.9370\n",
            "Epoch 2/50\n",
            "1936574/1936574 [==============================] - 24s 12us/step - loss: 0.1569 - acc: 0.9400 - val_loss: 0.1513 - val_acc: 0.9422\n",
            "Epoch 3/50\n",
            "1936574/1936574 [==============================] - 24s 12us/step - loss: 0.1485 - acc: 0.9436 - val_loss: 0.1467 - val_acc: 0.9435\n",
            "Epoch 4/50\n",
            "1936574/1936574 [==============================] - 24s 12us/step - loss: 0.1440 - acc: 0.9456 - val_loss: 0.1421 - val_acc: 0.9455\n",
            "Epoch 5/50\n",
            "1936574/1936574 [==============================] - 25s 13us/step - loss: 0.1412 - acc: 0.9469 - val_loss: 0.1398 - val_acc: 0.9473\n",
            "Epoch 6/50\n",
            "1936574/1936574 [==============================] - 25s 13us/step - loss: 0.1392 - acc: 0.9477 - val_loss: 0.1373 - val_acc: 0.9481\n",
            "Epoch 7/50\n",
            "1936574/1936574 [==============================] - 24s 12us/step - loss: 0.1377 - acc: 0.9485 - val_loss: 0.1376 - val_acc: 0.9487\n",
            "Epoch 8/50\n",
            "1936574/1936574 [==============================] - 24s 13us/step - loss: 0.1364 - acc: 0.9492 - val_loss: 0.1349 - val_acc: 0.9490\n",
            "Epoch 9/50\n",
            "1936574/1936574 [==============================] - 24s 12us/step - loss: 0.1354 - acc: 0.9496 - val_loss: 0.1343 - val_acc: 0.9491\n",
            "Epoch 10/50\n",
            "1936574/1936574 [==============================] - 24s 12us/step - loss: 0.1346 - acc: 0.9500 - val_loss: 0.1351 - val_acc: 0.9490\n",
            "Epoch 11/50\n",
            "1936574/1936574 [==============================] - 24s 12us/step - loss: 0.1339 - acc: 0.9503 - val_loss: 0.1376 - val_acc: 0.9489\n",
            "Epoch 12/50\n",
            "1936574/1936574 [==============================] - 24s 12us/step - loss: 0.1333 - acc: 0.9506 - val_loss: 0.1322 - val_acc: 0.9507\n",
            "Epoch 13/50\n",
            "1936574/1936574 [==============================] - 24s 13us/step - loss: 0.1329 - acc: 0.9508 - val_loss: 0.1328 - val_acc: 0.9504\n",
            "Epoch 14/50\n",
            "1936574/1936574 [==============================] - 24s 12us/step - loss: 0.1325 - acc: 0.9510 - val_loss: 0.1314 - val_acc: 0.9513\n",
            "Epoch 15/50\n",
            "1936574/1936574 [==============================] - 25s 13us/step - loss: 0.1321 - acc: 0.9512 - val_loss: 0.1308 - val_acc: 0.9512\n",
            "Epoch 16/50\n",
            "1936574/1936574 [==============================] - 24s 12us/step - loss: 0.1318 - acc: 0.9514 - val_loss: 0.1322 - val_acc: 0.9509\n",
            "Epoch 17/50\n",
            "1936574/1936574 [==============================] - 24s 13us/step - loss: 0.1315 - acc: 0.9515 - val_loss: 0.1305 - val_acc: 0.9515\n",
            "Epoch 18/50\n",
            "1936574/1936574 [==============================] - 25s 13us/step - loss: 0.1312 - acc: 0.9516 - val_loss: 0.1304 - val_acc: 0.9514\n",
            "Epoch 19/50\n",
            "1936574/1936574 [==============================] - 24s 13us/step - loss: 0.1310 - acc: 0.9517 - val_loss: 0.1300 - val_acc: 0.9512\n",
            "Epoch 20/50\n",
            "1936574/1936574 [==============================] - 24s 12us/step - loss: 0.1308 - acc: 0.9519 - val_loss: 0.1299 - val_acc: 0.9516\n",
            "Epoch 21/50\n",
            "1936574/1936574 [==============================] - 24s 12us/step - loss: 0.1305 - acc: 0.9520 - val_loss: 0.1302 - val_acc: 0.9516\n",
            "Epoch 22/50\n",
            "1936574/1936574 [==============================] - 24s 12us/step - loss: 0.1304 - acc: 0.9520 - val_loss: 0.1293 - val_acc: 0.9519\n",
            "Epoch 23/50\n",
            "1936574/1936574 [==============================] - 24s 12us/step - loss: 0.1302 - acc: 0.9520 - val_loss: 0.1347 - val_acc: 0.9507\n",
            "Epoch 24/50\n",
            "1936574/1936574 [==============================] - 24s 12us/step - loss: 0.1300 - acc: 0.9522 - val_loss: 0.1294 - val_acc: 0.9524\n",
            "Epoch 25/50\n",
            "1936574/1936574 [==============================] - 24s 12us/step - loss: 0.1298 - acc: 0.9523 - val_loss: 0.1291 - val_acc: 0.9518\n",
            "Epoch 26/50\n",
            "1936574/1936574 [==============================] - 24s 12us/step - loss: 0.1296 - acc: 0.9524 - val_loss: 0.1289 - val_acc: 0.9527\n",
            "Epoch 27/50\n",
            "1936574/1936574 [==============================] - 24s 12us/step - loss: 0.1293 - acc: 0.9525 - val_loss: 0.1302 - val_acc: 0.9513\n",
            "Epoch 28/50\n",
            "1936574/1936574 [==============================] - 24s 12us/step - loss: 0.1292 - acc: 0.9526 - val_loss: 0.1285 - val_acc: 0.9526\n",
            "Epoch 29/50\n",
            "1936574/1936574 [==============================] - 23s 12us/step - loss: 0.1289 - acc: 0.9526 - val_loss: 0.1285 - val_acc: 0.9520\n",
            "Epoch 30/50\n",
            "1936574/1936574 [==============================] - 24s 12us/step - loss: 0.1288 - acc: 0.9526 - val_loss: 0.1287 - val_acc: 0.9524\n",
            "Epoch 31/50\n",
            "1936574/1936574 [==============================] - 25s 13us/step - loss: 0.1285 - acc: 0.9528 - val_loss: 0.1280 - val_acc: 0.9527\n",
            "Epoch 32/50\n",
            "1936574/1936574 [==============================] - 24s 13us/step - loss: 0.1285 - acc: 0.9529 - val_loss: 0.1275 - val_acc: 0.9529\n",
            "Epoch 33/50\n",
            "1936574/1936574 [==============================] - 26s 13us/step - loss: 0.1283 - acc: 0.9529 - val_loss: 0.1292 - val_acc: 0.9519\n",
            "Epoch 34/50\n",
            "1936574/1936574 [==============================] - 25s 13us/step - loss: 0.1282 - acc: 0.9529 - val_loss: 0.1270 - val_acc: 0.9529\n",
            "Epoch 35/50\n",
            "1936574/1936574 [==============================] - 24s 13us/step - loss: 0.1280 - acc: 0.9530 - val_loss: 0.1266 - val_acc: 0.9532\n",
            "Epoch 36/50\n",
            "1936574/1936574 [==============================] - 24s 13us/step - loss: 0.1278 - acc: 0.9531 - val_loss: 0.1269 - val_acc: 0.9532\n",
            "Epoch 37/50\n",
            "1936574/1936574 [==============================] - 24s 13us/step - loss: 0.1276 - acc: 0.9532 - val_loss: 0.1274 - val_acc: 0.9528\n",
            "Epoch 38/50\n",
            "1936574/1936574 [==============================] - 24s 12us/step - loss: 0.1276 - acc: 0.9532 - val_loss: 0.1298 - val_acc: 0.9527\n",
            "Epoch 00038: early stopping\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "q-Yeu_3VYRqz",
        "colab_type": "code",
        "outputId": "fdad7109-92a2-46b7-d257-f602cb5e8872",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 590
        }
      },
      "source": [
        "# list all data in history\n",
        "print(history.history.keys())\n",
        "# summarize history for accuracy\n",
        "plt.plot(history.history['acc'])\n",
        "plt.plot(history.history['val_acc'])\n",
        "plt.title('model accuracy')\n",
        "plt.ylabel('accuracy')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()\n",
        "# summarize history for loss\n",
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.title('model loss')\n",
        "plt.ylabel('loss')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "['acc', 'loss', 'val_acc', 'val_loss']\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZIAAAEWCAYAAABMoxE0AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nrWKJxBhjTKtYIjHGGNMq/x9QRhaoyyDUGwAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYsAAAEWCAYAAACXGLsWAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nFiIi0i6FhYiItEthISIi7VJYiIhIu/4/KWf5Pta6ASsAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "aWRAlSP92U1h",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "bi_classifier = keras.models.load_model('binary-class.model')\n",
        "\n",
        "pred_x = bi_classifier.predict(x_test_encoded[0])"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "RiF8gGtUmmhB",
        "colab_type": "text"
      },
      "source": [
        "## ROC Curve for Binary Classifier"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "kTWz2KhBlF3w",
        "colab_type": "code",
        "outputId": "b4d1c3e3-5da5-4be8-a387-383609c37cd7",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 283
        }
      },
      "source": [
        "from sklearn.metrics import roc_curve, auc\n",
        "fpr, tpr, _ = roc_curve(y_Mix_test_N, pred_x)\n",
        "\n",
        "roc_auc = auc(fpr, tpr)\n",
        "\n",
        "plt.plot(fpr,tpr,color='darkorange',label='ROC curve (area = %0.2f)' % roc_auc)\n",
        "plt.legend(loc=\"lower right\")\n",
        "plt.xlabel('False Positive Rate')\n",
        "plt.ylabel('True Positive Rate')\n",
        "\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEKCAYAAAD9xUlFAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nhkWVGAAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "XDx9ZI6T5LuU",
        "colab_type": "text"
      },
      "source": [
        "## Histogram & ROC Curve of VAE"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DISsVyYUnkyv",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "vae_predict_x = vae.predict(X_Mix_test_N)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "VkTb4eBjpjoX",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Mean Square Root for prediction of loss\n",
        "predict_loss_mse = np.mean(np.power(X_Mix_test_N - vae_predict_x, 2), axis=1)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "T1MW4Kv3qW4F",
        "colab_type": "code",
        "outputId": "055b492b-a505-4619-85c0-3c47b42669d9",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "predict_loss_mse = predict_loss_mse.reshape(806907,1)\n",
        "predict_loss_mse.shape"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(806907, 1)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 30
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "QNYMmcAQxAOf",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from sklearn.preprocessing import StandardScaler, RobustScaler\n",
        "\n",
        "predict_loss=predict_loss_mse\n",
        "tf_loss=StandardScaler().fit(predict_loss.reshape(-1,1))\n",
        "predict_loss=tf_loss.transform(predict_loss.reshape(-1,1))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "syFEqUsRxpgv",
        "colab_type": "code",
        "outputId": "2de42f13-6016-42c3-9022-0d61bc9eee61",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 336
        }
      },
      "source": [
        "\n",
        "sigPred = predict_loss[np.where(y_Mix_test==1)]\n",
        "bkgPred = predict_loss[np.where(y_Mix_test==0)]\n",
        "fig = plt.hist(sigPred, bins=100, histtype='step', label='signal_loss', normed= 1)\n",
        "fig = plt.hist(bkgPred, bins=100, histtype='step', color= 'red', label='background_loss', normed= 1)\n",
        "fig = plt.yscale('log')\n",
        "fig = plt.legend(loc= 'upper right')\n",
        "fig = plt.title('loss distribution')"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python2.7/dist-packages/matplotlib/axes/_axes.py:6571: UserWarning: The 'normed' kwarg is deprecated, and has been replaced by the 'density' kwarg.\n",
            "  warnings.warn(\"The 'normed' kwarg is deprecated, and has been \"\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYAAAAEICAYAAABWJCMKAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "y3ge4tDv0HYU",
        "colab_type": "code",
        "outputId": "8f5c44b7-59e1-42b5-8d4f-3cc10c243093",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 323
        }
      },
      "source": [
        "from sklearn.metrics import roc_auc_score\n",
        "\n",
        "fpr_vae, tpr_vae, thresholds_vae = roc_curve(y_Mix_test, predict_loss)\n",
        "roc_auc = roc_auc_score(y_Mix_test, predict_loss)\n",
        "print(\"roc_auc =\", roc_auc_score(y_Mix_test, predict_loss))\n",
        "\n",
        "plt.plot(tpr_vae, 1/fpr_vae, color = 'darkorange', label='ROC curve (area = %0.2f)' % roc_auc)\n",
        "plt.xlim(0.1, 1.0)\n",
        "plt.ylim(1, 10000)\n",
        "plt.yscale(\"log\")\n",
        "plt.grid()\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "('roc_auc =', 0.9031671738324631)\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python2.7/dist-packages/ipykernel_launcher.py:7: RuntimeWarning: divide by zero encountered in divide\n",
            "  import sys\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAX4AAAD+CAYAAAA9HW6QAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nSTP/B/o5/zbUq+qYAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "x4fA6bnk2zM_",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "fce1o5X1NmrP",
        "colab_type": "text"
      },
      "source": [
        "\n",
        "# Training VAE with Mixed Data\n",
        "### [ Mixed = Background + Signal ]\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "5QtnLiN8aKEt",
        "colab_type": "code",
        "outputId": "689a8a27-ef56-4fd4-c60e-69f24dc286e5",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 746
        }
      },
      "source": [
        "# Variational Autoencoder (VAE) Training/Testing [MIXED]\n",
        "history = vae.fit(X_Mix_train_N,X_Mix_train_N,\n",
        "                  epochs=epochs,\n",
        "                  batch_size=batch_size,\n",
        "                  validation_split=0.2,\n",
        "                  callbacks=[es])\n",
        "                  #validation_data=(X_Mix_test_N, X_Mix_test_N)) #shuffle=True\n",
        "\n",
        "\n",
        "encoder.save_weights('ve-mix-weights.h5')\n",
        "vae.save_weights('vae-mix-weights.h5')\n",
        "\n",
        "vae.save('DarkMachine_autoencoder-mix.model')\n",
        "encoder.save('DarkMachine_encoder-mix.model')"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train on 1936574 samples, validate on 484144 samples\n",
            "Epoch 1/50\n",
            "1936574/1936574 [==============================] - 44s 23us/step - loss: 9.2752e-06 - acc: 0.9868 - val_loss: 4.6483e-06 - val_acc: 0.9817\n",
            "Epoch 2/50\n",
            "1936574/1936574 [==============================] - 44s 23us/step - loss: 4.0526e-06 - acc: 0.9880 - val_loss: 3.0279e-06 - val_acc: 0.9901\n",
            "Epoch 3/50\n",
            "1936574/1936574 [==============================] - 45s 23us/step - loss: 3.2400e-06 - acc: 0.9889 - val_loss: 2.6770e-06 - val_acc: 0.9913\n",
            "Epoch 4/50\n",
            "1936574/1936574 [==============================] - 44s 23us/step - loss: 2.8181e-06 - acc: 0.9893 - val_loss: 2.7181e-06 - val_acc: 0.9954\n",
            "Epoch 5/50\n",
            "1936574/1936574 [==============================] - 45s 23us/step - loss: 2.5455e-06 - acc: 0.9894 - val_loss: 2.3107e-06 - val_acc: 0.9938\n",
            "Epoch 6/50\n",
            "1936574/1936574 [==============================] - 45s 23us/step - loss: 2.3356e-06 - acc: 0.9901 - val_loss: 2.5873e-06 - val_acc: 0.9928\n",
            "Epoch 7/50\n",
            "1936574/1936574 [==============================] - 45s 23us/step - loss: 2.1619e-06 - acc: 0.9903 - val_loss: 2.1160e-06 - val_acc: 0.9866\n",
            "Epoch 8/50\n",
            "1936574/1936574 [==============================] - 44s 23us/step - loss: 2.0157e-06 - acc: 0.9906 - val_loss: 2.1952e-06 - val_acc: 0.9851\n",
            "Epoch 9/50\n",
            "1936574/1936574 [==============================] - 45s 23us/step - loss: 1.8328e-06 - acc: 0.9908 - val_loss: 1.7886e-06 - val_acc: 0.9902\n",
            "Epoch 10/50\n",
            "1936574/1936574 [==============================] - 45s 23us/step - loss: 1.7002e-06 - acc: 0.9909 - val_loss: 1.7756e-06 - val_acc: 0.9860\n",
            "Epoch 11/50\n",
            "1936574/1936574 [==============================] - 45s 23us/step - loss: 1.5812e-06 - acc: 0.9910 - val_loss: 1.3713e-06 - val_acc: 0.9880\n",
            "Epoch 12/50\n",
            "1936574/1936574 [==============================] - 44s 23us/step - loss: 1.4776e-06 - acc: 0.9911 - val_loss: 1.5400e-06 - val_acc: 0.9851\n",
            "Epoch 13/50\n",
            "1936574/1936574 [==============================] - 44s 23us/step - loss: 1.3877e-06 - acc: 0.9915 - val_loss: 1.4267e-06 - val_acc: 0.9910\n",
            "Epoch 14/50\n",
            "1936574/1936574 [==============================] - 45s 23us/step - loss: 1.3348e-06 - acc: 0.9916 - val_loss: 1.3079e-06 - val_acc: 0.9928\n",
            "Epoch 15/50\n",
            "1936574/1936574 [==============================] - 44s 23us/step - loss: 1.2632e-06 - acc: 0.9918 - val_loss: 9.9484e-07 - val_acc: 0.9965\n",
            "Epoch 16/50\n",
            "1936574/1936574 [==============================] - 45s 23us/step - loss: 1.2227e-06 - acc: 0.9918 - val_loss: 1.3124e-06 - val_acc: 0.9893\n",
            "Epoch 17/50\n",
            "1936574/1936574 [==============================] - 45s 23us/step - loss: 1.1720e-06 - acc: 0.9920 - val_loss: 9.1151e-07 - val_acc: 0.9893\n",
            "Epoch 18/50\n",
            "1936574/1936574 [==============================] - 44s 23us/step - loss: 1.1421e-06 - acc: 0.9920 - val_loss: 9.3833e-07 - val_acc: 0.9935\n",
            "Epoch 19/50\n",
            "1936574/1936574 [==============================] - 44s 23us/step - loss: 1.1035e-06 - acc: 0.9921 - val_loss: 1.1442e-06 - val_acc: 0.9925\n",
            "Epoch 20/50\n",
            "1936574/1936574 [==============================] - 43s 22us/step - loss: 1.0715e-06 - acc: 0.9922 - val_loss: 1.0836e-06 - val_acc: 0.9948\n",
            "Epoch 00020: early stopping\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "rH-QjqjnmNRi",
        "colab_type": "code",
        "outputId": "5ed49f54-3dfd-497f-aff6-d7ec297d4c66",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 590
        }
      },
      "source": [
        "# list all data in history\n",
        "print(history.history.keys())\n",
        "# summarize history for accuracy\n",
        "plt.plot(history.history['acc'])\n",
        "plt.plot(history.history['val_acc'])\n",
        "plt.title('model accuracy')\n",
        "plt.ylabel('accuracy')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()\n",
        "# summarize history for loss\n",
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.title('model loss')\n",
        "plt.ylabel('loss')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "['acc', 'loss', 'val_acc', 'val_loss']\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZIAAAEWCAYAAABMoxE0AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi40LCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcv7US4rQAAIABJREFUeJzsnXl8VOX1/98ne0I2SAIBwhJlByO7\nu7iL2rpga9Wq1bq1tYtt9Vvt4tZabX92s1otVqzWFdeCZVVBXEBAdgwQCBCSsISErGTP+f3x3AlD\nSMIkmZuZJM/79cpr7tzluedCMp95zjnPOaKqWCwWi8XSXkICbYDFYrFYujZWSCwWi8XSIayQWCwW\ni6VDWCGxWCwWS4ewQmKxWCyWDmGFxGKxWCwdwgqJxdIKIvJvEfmdj+fuEpEL3LbJYgk2rJBYLBaL\npUNYIbFYegAiEhZoGyzdFyskli6P41K6V0Q2iEiFiDwvIv1EZL6IlInIByLS2+v8y0Vks4gUi8hS\nERntdWyCiKxxrnsDiGpyr6+JyDrn2s9FJMNHGy8TkbUiUioie0TkoSbHz3TGK3aO3+zsjxaRP4nI\nbhEpEZFPnX3niEhuM/8OFzjbD4nIWyLysoiUAjeLyFQRWe7cY6+IPCUiEV7XjxWRxSJSJCL7ReSX\nIpIqIodFJMnrvIkiUiAi4b48u6X7Y4XE0l24GrgQGAF8HZgP/BJIwfye/xhAREYArwF3O8fmAXNF\nJML5UH0P+A/QB3jTGRfn2gnALOBOIAn4JzBHRCJ9sK8CuAlIBC4Dvi8iVzrjDnHs/btj03hgnXPd\nE8Ak4HTHpv8DGnz8N7kCeMu55ytAPfBTIBk4DTgf+IFjQxzwAbAAGAAMAz5U1X3AUuAar3FvBF5X\n1Vof7bB0c6yQWLoLf1fV/aqaB3wCfKGqa1W1CngXmOCc9y3gf6q62PkgfAKIxnxQnwqEA39V1VpV\nfQtY5XWPO4B/quoXqlqvqi8C1c51raKqS1V1o6o2qOoGjJhNcw5fD3ygqq859y1U1XUiEgJ8F/iJ\nquY59/xcVat9/DdZrqrvOfesVNUvVXWFqtap6i6MEHps+BqwT1X/pKpVqlqmql84x14EbgAQkVDg\nOozYWiyAFRJL92G/13ZlM+9jne0BwG7PAVVtAPYAA51jeXp0JdPdXttDgJ87rqFiESkGBjnXtYqI\nnCIiSxyXUAnwPczMAGeMHc1cloxxrTV3zBf2NLFhhIi8LyL7HHfX732wAeC/wBgRScfM+kpUdWU7\nbbJ0Q6yQWHoa+RhBAEBEBPMhmgfsBQY6+zwM9treAzyqqolePzGq+poP930VmAMMUtUE4FnAc589\nwInNXHMQqGrhWAUQ4/UcoRi3mDdNS3s/A2wBhqtqPMb1523DCc0Z7szqZmNmJTdiZyOWJlghsfQ0\nZgOXicj5TrD45xj31OfAcqAO+LGIhIvIDGCq17XPAd9zZhciIr2cIHqcD/eNA4pUtUpEpmLcWR5e\nAS4QkWtEJExEkkRkvDNbmgX8WUQGiEioiJzmxGS2AVHO/cOBXwPHi9XEAaVAuYiMAr7vdex9oL+I\n3C0ikSISJyKneB1/CbgZuBwrJJYmWCGx9ChUdSvmm/XfMd/4vw58XVVrVLUGmIH5wCzCxFPe8bp2\nNXA78BRwCNjunOsLPwAeEZEy4AGMoHnGzQEuxYhaESbQfrJz+B5gIyZWUwT8AQhR1RJnzH9hZlMV\nwFFZXM1wD0bAyjCi+IaXDWUYt9XXgX1AFnCu1/HPMEH+Narq7e6zWBDb2MpisfiCiHwEvKqq/wq0\nLZbgwgqJxWI5LiIyBViMifGUBdoeS3BhXVsWi6VVRORFzBqTu62IWJrDzkgsFovF0iHsjMRisVgs\nHaJHFHJLTk7WoUOHBtoMi8Vi6VJ8+eWXB1W16fqkY+gRQjJ06FBWr14daDMsFoulSyEiPqV6W9eW\nxWKxWDqEFRKLxWKxdAgrJBaLxWLpED0iRtIctbW15ObmUlVVFWhTXCUqKoq0tDTCw20PIovF4g49\nVkhyc3OJi4tj6NChHF3stfugqhQWFpKbm0t6enqgzbFYLN2UHuvaqqqqIikpqduKCICIkJSU1O1n\nXRaLJbD0WCEBurWIeOgJz2ixWAJLjxYSi8USxFQchA2zj3+eJeBYIQkQxcXF/OMf/2jzdZdeeinF\nxcUuWGSxBBnLnoB3boeS47VZsQQaKyQBoiUhqaura/W6efPmkZiY6JZZFktwoAqZc812wZbA2mI5\nLlZIAsR9993Hjh07GD9+PFOmTOGss87i8ssvZ8yYMQBceeWVTJo0ibFjxzJz5szG64YOHcrBgwfZ\ntWsXo0eP5vbbb2fs2LFcdNFFVFZWBupxLBb/kr8GSp2ZSMHWwNpiOS6upv+KyHTgb0Ao8C9VfbzJ\n8SGYntQpmDaiN6hqrnPsD8Blzqm/VdU3nP0C/A74JlAPPKOqT3bEzofnbuar/NKODHEMYwbE8+DX\nx7Z4/PHHH2fTpk2sW7eOpUuXctlll7Fp06bGNN1Zs2bRp08fKisrmTJlCldffTVJSUlHjZGVlcVr\nr73Gc889xzXXXMPbb7/NDTfc4NfnsFgCQuZckFCIiIUDmYG2xnIcXBMSEQkFnsb0gc4FVonIHFX9\nyuu0J4CXVPVFETkPeAy4UUQuAyYC44FIYKmIzFfVUkyP7EHAKFVtEJG+bj1DZzJ16tSj1no8+eST\nvPvuuwDs2bOHrKysY4QkPT2d8ePHAzBp0iR27drVafZaLK6hCl/NgfSzoKHezki6AG7OSKYC21U1\nG0BEXgeuALyFZAzwM2d7CfCe1/5lqloH1InIBmA6MBv4PnC9qjYAqOqBjhra2syhs+jVq1fj9tKl\nS/nggw9Yvnw5MTExnHPOOc2uBYmMjGzcDg0Nta4tS/fgQCYU7YDT7jLbG94w4mJT2YMWN2MkA4E9\nXu9znX3erAdmONtXAXEikuTsny4iMSKSDJyLmYUAnAh8S0RWi8h8ERnu2hO4SFxcHGVlzXctLSkp\noXfv3sTExLBlyxZWrFjRydZZLAEkcw4gMOprkDISqkuhbG+grbK0QqBLpNwDPCUiNwPLgDygXlUX\nicgU4HOgAFiOiYeAcXVVqepkEZmBibGc1XRgEbkDuANg8ODBbj9Hm0lKSuKMM85g3LhxREdH069f\nv8Zj06dP59lnn2X06NGMHDmSU089NYCWWiydTOZcGHwqxPWDlFFmX8EWiB8QWLssLeKmkORxZBYB\nkObsa0RV83FmJCISC1ytqsXOsUeBR51jrwLbnMtygXec7XeBF5q7uarOBGYCTJ48OSgb07/66qvN\n7o+MjGT+/PnNHvPEQZKTk9m0aVPj/nvuucfv9lksnU7hDti/CS7+vXnfd7R5PbAFTjwvcHZZWsVN\n19YqYLiIpItIBHAtMMf7BBFJFhGPDfdjZheISKjj4kJEMoAMYJFz3nsYVxfANI4IjMVi6ep41o6M\n/rp57ZUMMUl2LUmQ49qMRFXrROSHwEJM+u8sVd0sIo8Aq1V1DnAO8JiIKMa1dZdzeTjwiVMnqhST\nFuxZqfc48IqI/BQoB25z6xksFksnkzkX+o+HRC93dMoom7kV5LgaI1HVecC8Jvse8Np+C3irmeuq\nMJlbzY1ZzJH1JRaLpbtQkgd5q+H8B47enzIKNr1lM7eCGLuy3WKxBAdb3jevoy8/en/KKKgqgfL9\nnW+TxSeskFgsluDgqzmQMhqSm2T0p4w0rzZOErRYIenO1FZClX9Lv1gsrlBeADmfHwmye+OduWUJ\nSqyQBIj2lpEH+Otf/8rhw4ePf2JJrlkhXFfdrvtYLJ3G1v+BNsCYy4891isFonvbGUkQY4UkQLgu\nJPV1UFNutg8XwuGidt3LYukUMudC76HQb9yxx0SczC0rJMFKoFe291i8y8hfeOGF9O3bl9mzZ1Nd\nXc1VV13Fww8/TEVFBddccw25ubnU19fzm9/8hv3795Ofn8+5555LcnIyS5Ysaf4G1Y5LKyENdhfA\nu9+D616HEPvdwRJkVBZD9sdw6vdbzspKGQWb37WZW0GKFRKA+ffBvo3+HTP1JLjk8RYPe5eRX7Ro\nEW+99RYrV65EVbn88stZtmwZBQUFDBgwgP/973+AqcGVkJDAn//8Z5YsWUJycnLL968uhZAwiEmG\n6ETIWgjLn4Izfuzf57RYOsq2hdBQe2y2ljcpo6CqGCoKILZbFPzuVtivp0HAokWLWLRoERMmTGDi\nxIls2bKFrKwsTjrpJBYvXswvfvELPvnkExISEnwbUNUE2SPjzbe3yDgTxPzwYdizyt2HsVjaSuYc\niOsPAye1fE5fp+aW7U0SlNgZCbQ6c+gMVJX777+fO++885hja9asYd68efz617/m/PPP54EHHmhm\nhCbUVIDWQ5SX8Fz+FPzzbHjrFrhzGcT08eMTWCztpKYCtn8AE29q3e3aWLxxK5wwrXNss/iMnZEE\nCO8y8hdffDGzZs2ivNwEx/Py8jhw4AD5+fnExMRwww03cO+997JmzZpjrm2W6hLAmYl4iE6Eb74A\nZfvgv3eZWYvFEmiyFkNdVfNpv97E9jNfjGzAPSixM5IA4V1G/pJLLuH666/ntNNOAyA2NpaXX36Z\n7du3c++99xISEkJ4eDjPPPMMAHfccQfTp09nwIABzQfbq0pNi9KQ0KP3D5wEFz4CC++HFc/AaT9w\n+zEtltbJnGuKMg4+vfXzRMxiRSskvlNRCKueg1N/AFHxrt5KtAd8M508ebKuXr36qH2ZmZmMHj06\nQBa5SF01HPgK4gc2BiWPelZVeP3bkLUIbl3Yul/aYnGTumr444kw9kq44qnjnz/nx6aMyv9lu29b\nd2DB/fDFs/CDFUeqA7QREflSVScf7zzr2upuVJWY16gWAvMi5o82LhXevNmkXlosgSB7KdSUwZgr\nfDs/ZZRZE1Vx0FWzugWHdsHK52DCDe0WkbZghaS7UVUKYZHmpyVi+sA3XoDSfJjzQxsvsQSGzDkm\nszD9bN/Ot5lbvvPRoyb9/5z7O+V2PVpIup1br6HerGaPPDIbafEZB02B8x80PuqVz3WSgRaLQ30d\nbJkHI6a3/qXHG++2u5aW2bseNs42Czw7qT1xjxWSqKgoCgsLu5eYVJcB2hhYU1UKCwuJiopq/vzT\nfgjDL4ZFv4L8dZ1np8Wy+zOoLDp+tpY3cf3NDMY2uWqdxQ9CdB848+5Ou2WPzdpKS0sjNzeXgoKC\nQJviPw4XQe1hKI5sLCMRFRVFWlpa8+eHhMBVz8KzZ5p4yZ3LXM/usFgA49YKi4ZhF/h+ja25dXx2\nfATZS+Dix1qOk7pAjxWS8PBw0tPTA22G/2hogD+NMP7mb8zy/bqYPub8Fy6FuT82sRNby8jiJg0N\nkPk+DL8AImLadm3KSNi2wB27ujoNDWY2kjgYptzaqbd21bUlItNFZKuIbBeR+5o5PkREPhSRDSKy\nVETSvI79QUQ2OT/faubaJ0Wk3E37uxT5a00dohHT237t4FPhvF+bonir2yBCFkt7yF0F5ftgtI/Z\nWt6kjDK/5xWF/rerq7P5Hdi3Ac77je9xJz/hmpCISCjwNHAJpv/6dSLStA/7E8BLqpoBPAI85lx7\nGTARGA+cAtwjIvFeY08Gertle5dk2wKQkLa5Crw5425z7YL7Ye8G/9pmsXiTOQdCwmHERW2/tq8N\nuDdLXTV8+IgpFjvuG51+ezdnJFOB7aqarao1wOtA068gY4CPnO0lXsfHAMtUtU5VK4ANwHRoFKj/\nB/yfi7Z3PbYtgEGntr+GVkgIXPVPc/2bNzuBe4vFz6gaITnhnPb58G3mVvOsngXFu+GChwPSKsLN\nOw4E9ni9z3X2ebMemOFsXwXEiUiSs3+6iMSISDJwLjDIOe+HwBxV3dvazUXkDhFZLSKru1VAvTlK\n882UdsTFHRunVzJc/Twc2gnv/9SuL7H4n30boDin+U6IvhA/ECLirJB4U1UCH//RiPOw8wNiQqDT\nf+8BponIWmAakAfUq+oiYB7wOfAasByoF5EBwDeBvx9vYFWdqaqTVXVySkqKaw8QFGxbaF7bEx9p\nytAz4JxfwsY3Yc1LHR/PYvHmqznGBTvy0vZdL2IC7lZIjvDZkyaV+oKHAmaCm0KSx5FZBECas68R\nVc1X1RmqOgH4lbOv2Hl9VFXHq+qFgADbgAnAMGC7iOwCYkRku4vP0DXYthASh/ivFMJZPzPfbub/\nH+zf7J8xLRYwC2CHnGFmv+0lZZRdS+KhdC8sf9rERQZMCJgZbgrJKmC4iKSLSARwLTDH+wQRSRYR\njw33A7Oc/aGOiwsRyQAygEWq+j9VTVXVoao6FDisqsNcfIbgp7bS1CwaMd1/abshoTDjOePDfvVa\nG3y3+IeCrXBwa+udEH2h7ygo32/WTfV0Pn4cGupM1mUAcU1IVLUOE89YCGQCs1V1s4g8IiKe36Rz\ngK0isg3oBzzq7A8HPhGRr4CZwA3OeJam7PwE6io7Hh9pSmxf0+O9oQ6evxDWvuzf8S09j0zne+To\nr3VsHO8mVz2Zgm2w5j9mzUifwK6Jc3VBoqrOw8Q6vPc94LX9FvBWM9dVYTK3jjd+rB/M7NpsWwDh\nvWDomf4fe+BEs9r97e+aZlh7voBL/h+Et1ByxWJpja/mQNqUjtd/8rhwC7bAkNM6bldX5cOHITwG\nzr430JYEPNhu6QiqJj5y4rnuLUCKTYEb34Ozfm6C77MuMiWqLZa2cGiXydjqqFsLIGGQadwW7AF3\nVSh3KWM05wvTm+WMn3Qs3uQnrJB0ZfZvgtJc/2RrtUZIKJz/gHF1Fe0yvd89mWIWiy9kzjWvbSnS\n2BIikDwi+IVkw2x4YphZKNhQ779xVWHxA6b9cJB0ObVC0pXx1Bwa3o4Vwu1h5CVw58emls+r18BH\nv/PvH4il+5I516y69pcvP2UUHAhyIfGs4P/kT/CfK6H8gH/G3Tof9qyAc+6DiF7+GbODWCHpymxb\nCAMmQly/zrtnn3S4dbHpvLbs/8HLM2zHOkvrlO418TV/uLU89B1l6nVVHvLfmP6krsZkU068Ea58\nBvasNDP53cs7Nm59HXzwECQNhwk3+cNSv2CFpKtSXgC5q913azVHeDRc8TRc/nfzh/HPs2HPqs63\nw9I12PK+efWnkDRmbm3z35j+JGe5aTI37EIYfz3c9qH5u/n3ZfD5U+2vGrH+VZNCfcGDEBo8xdut\nkHRVti8G1P9pv21h4k1w6yLT0vOFS+CLmbasiuVYMueab9D+7B3emLkVpG13sxZBaMSRNsKp4+CO\npcY9vOhXMPsm0xa7LdQchiW/h7SpMKqDKdR+xgpJV2XbAtMxrv/JgbVjwHgTNxl2Psy/F96+Dapt\ndX+Lw+Ei2PWpCbL7s89NwmCT+hqsa0m2fwBDTodIrxUKUQnwrZfhot/Blv/BzHPaVjnii2egbC9c\n+HDQ9QyyQtIVqauB7R+Z2Ugw/EJF94ZrXzN9EDa/A/86P3hdDpbOZes80Pr2F2lsiZCQ4M3cKs4x\ndjWXBCMCp/8Ibn4fairgufNh3WvHH/NwEXz6VxhxiRGoIMMKSVck53OoKQtMfKQlQkLg7HvghndM\n46HnzjWNsiw9m6/mmNlD//H+H7vv6ODM3MpabF6HXdjyOUNON4t90ybDe9+DuT+B2qqWz1/2hIm5\nXPCgf231E1ZIuiLbFkJYFKRPC7Qlx3LiuXDnJ+aP/M2bYenjgbbIEiiqSk3/cH+7tTykjISyfFNG\nPZjY/oFJkU8e3vp5cf3MYt8zfwpf/rvlxb6HdsOq50zQvu9oNyzuMFZIuhqqJo88/ey297vuLBIG\nws3zYORl8OlfTMqipeex5X9QX+OfRYjNEYyZW3XVkP2xcWv5Ip6hYab8+7WvOYt9px272HfJo6b0\n/jm/dMFg/2CFpKtRuN00ngpktpYvhEUYv3hdFRRmBdoaS2dTecisd+g7FgZNdecejUISRJlbuz+H\n2orW3VrNMepSuHMpJA4yi309q+H3bjAr5E/9vvmCFqQETyKyxTcaV7MHuZAApGaY170bgnZKbnGJ\nBfebWNn1r5sSO26QOBjCooMrc2v7BxAaCelntf3aPieYxb7z7jWr4XNXQUMDRCfCGXf731Y/Ymck\nXY1tC6HfOPPNJdhJHmFiOftsP5Mexdb5sP410yDNzWZLIaEmDnEgiGYkWYtMl9H2li4Jj4YrnjIL\nfveshN2fwln3GDEJYuyMpCtRWWymzmcG97eTRkLDoO8Y2Ls+0JZYOovKQzD3buPSOvv/3L9f39Gw\n6zP37+MLh3bDwW0w6ZaOjzXhBrNG7Ks5MOW2jo/nMnZG0pXY/oHJyQ+mtN/j0T8D9m20K957Ch6X\n1pVPmziZ26SMNBWw27pK3A22O2m//iqimnoSnPerLtH/xwpJV2LbQohJgoGTAm2J76RmQFUxlOwJ\ntCUWt9m6oHNcWt54Au4HgyBzK2sx9B4KSScG2pJOxwpJV6G+znzjGX6Re8FLN/CUcLF937s3lYfM\norq+Yzq3Y19j5laAFybWVsHOZSZbKxiqTXQyrsZIRGQ68DcgFPiXqj7e5PgQYBaQAhRherPnOsf+\nAFzmnPpbVX3D2f8KMBmoBVYCd6pqrZvPERTkrjJ/rMGe9tuUvmNMDvy+DR3v1W0JXhb88kiWllvd\nOpuj91CTJRVoIdn9GdQe9mtvIFXlcE095dV15qeqrvntFo5VVNdRVl3H2987ncFJ7q45c01IRCQU\neBq4EMgFVonIHFX9yuu0J4CXVPVFETkPeAy4UUQuAyYC44FIYKmIzFfVUuAV4Abn+leB24Bn3HqO\noGHbAlNl98TzAm1J24iIMdlbdkbSfdm20JQ3P+ueznNpeQgJNb9fgS6V4kn7HXpms4dr6hooPlzD\nocO1HDpcQ/Hh2sb3xc77Q96vlWZ/bf3xY4vhoUJsZBixUWHERoYTGxlKcmwEQ5N7ERsZRkSY+44n\nN2ckU4HtqpoNICKvA1cA3kIyBviZs70EeM9r/zJVrQPqRGQDMB2YrarzPBeLyEogzcVnCB62LTT1\neaISAm1J20nNMN/YLN0Pb5fWtE7I0mqOvqNMD3M/cLimjvV7SvhydxFf7j5E1oHy4+aJiMArVf9l\nr4zhnr+sOMqzVV+vFFfWcrim5U6iEaEhJMaE0zsmgsSYcE5I6eVsR5AQHU5cVBhxUWHERobRK9K8\net7HRoURGRZ4V7ebQjIQ8I6w5gKnNDlnPTAD4/66CogTkSRn/4Mi8icgBjiXowUIEQkHbgR+0tzN\nReQO4A6AwYMHd/RZAsuhXWb17sQbA21J++ifARtnQ0Uh9EoKtDUWf7Lgl6aF7HWvda5Ly5uUkbDx\nTdO+wLtsuw/kF1eyevch1uw+xJe7D/HV3lLqG4xyDO8by6QhvQkPPfKNvqmoKEqfmnyGbM9jTd8Z\nTO3T56jjIuKIRDgJMRH0dgQjITqc3r3M++jwUKSLx1UCvY7kHuApEbkZWAbkAfWqukhEpgCfAwXA\ncqCppP8DM2v5pLmBVXUmMBNg8uTJXTv3dNsi89qV0n69ST3JvO5b3/Vcc5aWccmlpaocKKsmc28p\ne0uqiI8yH8SJMRH07mU+iKPCvb6FpzhVEw5ubTWjsba+ga/yS/ly9yG+zDHisbfEVNyNDg9l/KBE\nvj/tRCYN7c3EQb1JiAn3zeCVq2A7XHXNLVzVAzO2wF0hyQO8l1+nOfsaUdV8zIwEEYkFrlbVYufY\no8CjzrFXgcb8PhF5EBOgv9NF+4OHbQtMh7mu+kvqXSrFCkn3wE8uraraerYfKCdzbymZe8vYsq+U\nLfvKKKqoafW6yLCQRlfQ6Igq/gK8Of9Ddqb1atzfOyaCBlXW7Snmy92HWJ9bTFVtAwADE6OZPLQP\nkwYnMnloH0alxhEW2s5YQtYiU96kq/59+gE3hWQVMFxE0jECci1wvfcJIpIMFKlqA3A/JoPLE6hP\nVNVCEckAMoBFzrHbgIuB853rujfV5bDrE5h6R6AtaT8xfSBhkC2V0p1oo0tLVdlfamYZmfsc0dhb\nSvbBikZXUlR4CCNT47loTD9GpcYxun88aX1iKKuqPSo43RiUrjBB6fyKftQQRmXeJmZmj6Ou4WgH\nRFiIMHZAPNdNHczkIX2YOCSR/gnR/vl3qK2CnZ+YttM9GNeERFXrROSHwEJM+u8sVd0sIo8Aq1V1\nDnAO8JiIKMa1dZdzeTjwieM3LMWkBXtqkT8L7AaWO8ffUdVH3HqOgJO91JTi7qpuLQ+pGTZzq7vQ\nikvLIxjZB8vZebCC7QfK2bK3jMx9pRQfPpKlPzAxmtH945g+LpVRqfGM7h/HkKRehIY0Fyvw4UP/\nmVHcFF/FjddfQll1HcUVtRRXmqynMf3jiY5wKSC9+1Ooq/Rr2m9XxNUYiZNhNa/Jvge8tt8C3mrm\nuipM5lZzYwY6rtO5bFsAkQkw+NRAW9Ix+meYtqvtCIhaggjHpVWfPJoNJ9xO9pe57DxYcdRPZe2R\ncGZ0eCgjU+O4ZFwqo/vHMyo1npGpcSRE+xh/8JWUkZC7GhEhPiqc+KhwBtMJ/XqyFpvCpEPPcP9e\nQUzP+lDuajQ0GP/rsPMh1M9/eJ1NagagsH8zDG6avGcJRurqG8g6UN4oENkFFVyW/QhnV+3nyoN3\nsemfXwIQGiIM6h1NenIvTj0hifSUXpyQ3Iv05F6kxkcR0uwsw8+kjIJNb5s+6O2tvNseshbD0LNM\n1d4ejBWSYGbvOijf3/XdWmBmJGDiJFZIgpaq2no+zTrIws37+CBzP4e83FEzYjdxXt0HfNT3Jq4Y\ndxl3J/ciPaUXg3rHdMqit1bxrrnVWYsiC3dA0Q445Xudc78gxgpJMLPlfVNeZNgFgbak48QPhOg+\nNuAehJRW1bJkywEWbt7H0q0FHK6pJy4yjPNG9+XckX0Z3i+Wob3q6PWvn0P0GM674wnOC9SakZZo\nrLm1tfOEZPsH5nV4N/j77CBWSIKVze/Cp381s5HusIhPxMxKbMA9KCgoq2bxV/tZuHkfn+84SG29\nkhIXyZUTBnLx2FROOyHp6FnGez8wWVrXvhq4hYet0ScdQsI7t8lV1mLoc6JJ/e3hWCEJRja9A2/f\nZnpdz5gZaGv8R2oGfPEs1NddgvUOAAAgAElEQVR2/ZhPFySn8DALN+9j4eZ9fJlzCFUYkhTDLWek\nc/HYfkwY1Lv5eMa2RbDuFTjr5zBwYucb7guh4aZbYme13a2tNGn5/mhi1Q2wQhJsbHob3r4dBp0C\n336ze2U49T/ZpDIXbIXUcYG2pttTV9/Atv3ljeKxZV8ZAGP6x3P3+SO4eFw/RvaLa708R2UxzP2x\nWT0+7RedZHk7SRkJ+es65167PoW6KuvWcrBCEkx4RGTwqXD97O4lInBkhfu+DVZI2kltfQOF5TUc\nLK+moLyag2XVHHTeN/6UmfdFh2tQNV7FKUP68OvLRnPx2FQG9WlDWuzCXwW3S8ublFGw+T2oOWyq\nTrtJ1iIIi4YhzVf77WlYIQkWNr4F79wOg0/rniICpoREeIyJk4y//vjnd3OqauspqaylpNKs3D6y\nXUNpZS3FlbUcOlxLQVlVo1h4L+rzJiYilOTYSJJjIxiSFMOkob1Jjo1kUO9ozh3Vl+TYdojAjiWw\n7uXgdml5kzIKUCjMOtJQzS2yFkP62V2iDW5nYIUkGGgUkdPh27M7Nw++MwkJhX7jukXmlnfTobKq\nWsqqjjQVKqsyDYXKq+oorToiFEYcahrfV9e1XOEnRCA+2tSLSo6NYHjfWE47IcmIRVyEIxqRpDjv\nYyJc+FPOnGMWwwa7S8uDd+aWm0JSuAMO7YTT7jr+uT0EKySBZsOb8O4d3V9EPKSeZEp+NzRASPB1\nelZV8oor2ZRXyqa8EnYWVhzpPlfliIbTfa7Bh5rSMRGhJEaHEx8dbnpNJMeS4Gx79iVEh5MYHXHU\n/rjIsM5ZyNcaeWtgwPjgd2l5SDrRNH9zO3Mry6nG3R3S8v2EFZJAsmE2vHsnDDkDrn+j+4sImBTg\n1c9D8a6Ap02qKnuKKtmYV8Km/BI25ZkfzyI8z4rthOhwYqPCSI6NITby6EZDcVHmWJzTZKhxf2Q4\nvSJD219RNtDUVZsqBF3pW3doOCQNcz9zK2uxqcbdJ93d+3QhrJAEivVvwHvf61kiAkeXlO9EIWlo\nUHYVVrApv7RRMDbllVBaZWqBhocKI/rFcfHYVMYOTOCkgQmMSo07uu9FT2L/Zmio7RqxEW9SRsK+\nje6NX3PYZGxNuc29e3RBrJAEgqNEZLb7GSbBRN8xIKEmTjL2SlduoarkHqpkTc4hNuQawfgqv5Sy\naiMaEaEhjOofx9dOHsC4AUY0RqTGBkXL0qAhf6157ewe7B0lZTRkzjXrPNyof7XrE6ivtmm/TfBJ\nSETkHeB5YH6P6AHiJuteg/e+D+lnwXVv9CwRAZPlkjLKryvcy6pq2ZBbwro9xazNOcS6PcUcLDeN\nkSLDQhgzIJ4rJwzkpIEJjB0Yz4h+cUe1T7U0Q/4aiEkyfWS6EikjQRugcPuRzpz+JGuxyTwc0rOr\n/TbF1xnJP4BbgCdF5E3gBVXtpCWk3Yh1r5pSE+lnw3Wv9zwR8dA/A3Z81K5L6xuUrANlrMspZm1O\nMev2FLPtQFljL+0TU3oxbURfJgxOZPygREamWtFoF/nrzGykq/US92RuHdjifyFRNYH29GldJwGh\nk/BJSFT1A+ADEUkArnO29wDPAS+ravPJ7ZYjrH0F/nsXnDANrn2t54oImDjJ+tegbD/E9Wv11IKy\n6qNmGuv3FFNRY/pdJMaEM35QIpee1J8JgxM5OS3R9z7blpapOWwyn0ZeEmhL2k7SMOM6Ldji/7EL\nt0Pxbjjjx/4fu4vjc4xERJKAG4AbgbXAK8CZwHcwnQ4tLbH2ZfjvD42IXPd6j+9dcFRJ+bgLG3eX\nVdWyMa+EDbklbMgtZv2eEvKKKwHTLnXMgHiunpTmzDZ6MzQppvXyHpb2sX8TaD0M6GKBdoCwCJMG\n7IaQZC02r8MubP28HoivMZJ3gZHAf4Cvq+pe59AbIrLaLeO6BWv+A3N+BCecY/pb93QRgUaXQ96W\nL1h8YBgbcktYn1tM9sGKRhfV4D4xTBicyM2nD2XC4ETGDUzouRlUnU1XDbR7SBnpzlqSrEWQPBJ6\nD/H/2F0cX2ckT6rqkuYOqOrkli4SkenA3zA92/+lqo83OT4EmAWkAEWY3uy5zrE/AJc5p/5WVd9w\n9qcDrwNJwJfAjapa4+NzdC6b3jYicuK5plZRDxURT6e9DbnFrM8tYf2eYp7RvmxYuYyHaieQEhfJ\nyWkJXDF+IBlpCWSkJdKnV0SgzTaJEZvfMZl1PWnmk7cGYlMhvn+gLWkfKaNhy//MWhh/xTJqKmD3\nZzD1Dv+M183wVUjGiMhaVS0GEJHewHWq+o+WLhCRUOBp4EIgF1glInNU9Suv054AXlLVF0XkPOAx\n4EYRuQyYCIwHIoGlIjJfVUuBPwB/UdXXReRZ4FbgmbY8dKexahYkj+iRIlJT18BrK3OYuz6fTfkl\nVNWaZL+4qDAy0hKoZCzn1mSz/PbzSI2PCk4X1frXYOfHxk3Sd3Sgrek88td23dkIHJ251W+sf8bc\nucxUrh5u3VrN4auQ3K6qT3veqOohEbkdk83VElOB7aqaDSAirwNXAN5CMgb4mbO9BHjPa/8yVa0D\n6kRkAzDdyRg7D/BU/HsReIhgFZKibOPS6kEioqrM27iPPy7cwu7Cw4wbGM+1UwYzflAiGWkJDE3q\nZUp/LDsDPlpCr8hakCD896mvhVzHa5u1uOcISXWZaVc77upAW9J+GjO3Mv0nJFmLIbyXKapqOQZf\nhSRURETVeLCd2cbxfA8DgT1e73OBps261wMzMO6vq4A4J6i/HnhQRP4ExADnYgQoCSh2BMYz5sDm\nbi4idwB3AAwePNiXZ/QvNYehLB+Sek73tJU7i/j9vEzW7SlmZL84XrhlCueMSGl+tpHqFNXbvwmG\nnN65hvrCvo1QWwEIbF/cczJ19m4AtOutaPcmebhpUe2vUimq5nfgBJv22xK+CskCTGD9n877O519\nHeUe4CkRuRlYBuQB9aq6SESmAJ8DBcByoL4tA6vqTGAmwOTJk30or+dnDu00rz2gDef2A+X8YcEW\nFn+1n37xkfzx6gyunpRGaGtFB/t7lUoJRiHJWW5eT/qmaXtcXQaRcYG1qTPIX2Ne+48PrB0dISzS\n/N0V+CngfnAbFOfAmT/1z3jdEF+F5BcY8fi+834x8K/jXJMHeC+LTXP2NaKq+ZgZCSISC1zticOo\n6qPAo86xV4FtQCGQKCJhzqzkmDGDhqJs89qNheRAWRV/+yCL11ftITo8lHsvHsl3z0gnOsKH7Kq4\nVOjVN3hLyucsh8QhMPFG2Dgbsj+G0V8LtFXuk7/WrGaPTQm0JR0jZZT/ZiQ27fe4+LogsQETh2hL\nLGIVMNzJssoDruVIbAMAEUkGipzx78dkcHlcZ4mqWigiGUAGsEhVVUSWAN/AZG59B/hvG2zqPAp3\nmNduKCQV1XU890k2M5dlU1PXwA2nDOZH5w9ve/Ok1JP8WirFb6hCzgo48XwYdCpExBrXRk8RkgFd\neDbiIWUUbJ0PdTVmbUlHyFpkMsESu1i5mE7E13UkwzEZVWOAxpZgqtrip6Sq1onID4GFmPTfWaq6\nWUQeAVar6hzMQsbHREQxri1Pzepw4BPHt16KSQv2xEV+AbwuIr/DLIx83sdn7VyKsiEmGaISAm2J\n36irb2D26lz+8sE2CsqqufSkVO69eBTpye2sXNw/Az7/u3/TNP1BUTZUFMCQ08yH0AnnQNYHNPat\n7a5UHjLPPuHGQFvScVJGmUWVhduh35j2j1Ndbmanp9zpP9u6Ib66tl4AHgT+ggl83wIct4CRqs4D\n5jXZ94DX9lvAW81cV4URrebGzMZkhAU3RdndZjaiqnyYeYDHF2xh+4FyJg/pzbM3TGLSkN4dGzg1\nAxrqTHZNMH0L9sRHPBk6wy6ALe93/zTgvevNa1dO/fXQ19MtcUvHhMST9mvdWq3iq5BEq+qHTubW\nbuAhEfkSeOB4F/ZYirJNccYuzro9xfx+XiYrdxZxQnIv/nnjJC4a088/6z487VD3bQg+IYnuY9YA\nwZG1A909DTjPCbQH0/9Fe0ka5p/MraxFxrVp035bxVchqRaRECDLcVflAbHumdXFqa2E0rygnpGo\nKmXVdRSUVXOgtJoDZVVmu6yaA6VV5rWsmu0HykmOjeC3V47j2imD/FtJt3c6RMQFX5wkZwUMPvWI\nGyshzfRR6e5pwPlrzf9JdAdnmsFAeDT0HtqxzC1V2P6BcW12NM7SzfFVSH6CWc/xY+C3GPfWd9wy\nqstTFBypv9sPlLE2p5gDZdWOSFQ5omHeV9Yem1EdERZC37hI+sZFMiwllqsmDOQ7pw8lNtKFHmgh\nIZA6Lrgyt8oLjF994k1H7x92Aax4pnunAeevg7QWKx51PVJGH39GUl8LtYehtsp5rTQ/dZUm5bdk\nD5z1886xtwtz3E8HJ4PqW6p6D1COiY9YWiOAqb+F5dXMXZ/PO2vz2JBb0rg/LiqMFEcgxg9KNGIR\nH0nfuKjG7ZTYKOKjwzq3XElqhqmO3FAPIUFQlHHPCvPa1JUx/EL4/EnjMx912bHXdXUqDkJJDpzS\njWpJpYyEbQvghUuPFYs6Z7uhrvUxQsJg+EWdY28X5rhCoqr1InJmZxjTbehkIamuq+ejzAO8vSaP\npVsPUNegjB0Qz2++NobzRvUlNT7Kt7UdgaB/BqysMP9mycMDbY1xa4VFHYnfePCkAWct7p5Ckr/O\nvHaHQLuH0V+H3Z8DYjIow6NNd8PwKOc12vyERXsdi/b6iYHYfpDQbPEMixe++ivWisgc4E2gwrNT\nVd9xxaquTtEO06Y0OtG1W6gqa3KKeWdNLnPX51NaVUffuEhuPTOdqyYOZFRqvGv39iupXr1JgkFI\ndn8OAycdm47sSQPe3k3TgPPXAHLk/6M7MHAi3Low0Fb0CHwVkijMqvLzvPYpYIWkOVxM/d1TdJh3\n1+bxzppcdhUeJio8hOljU5kxMY0zhiW3XpYkGEkZBSHhJuAe6EKBNRUmBfbMu5s/3p3TgPPXGiGP\n6iJfQCxBha8r221cpC0U7fRr/ajSqlrmb9zL22vyWLmzCIDTTkjirnOHcclJ/d0JhHcWYRHmQzkY\nAu65q80itpZSPbtzGnD+2m6Rrm4JDL6ubH8BMwM5ClX9rt8t6urUVkFJLvQ5scNDrdtTzPOf7mTR\n5n1U1zVwQkov7r14JFeMH0Ba727U871/BmxdEHiXUc4KQCBtSvPHE9JMJlB3SwMu3Qtle7tma11L\nUODrV9n3vbajMCXf8/1vTjfg0C5AO+Taqq1v4O8fZvHUku3ER4fzrSmDmDExjZPTEoKzAVRHST3Z\nZG6V7YX4AYGzI2c59BvXemxr+AWw4tnulQbc1VvrWgKOr66tt73fi8hrwKeuWNTV6WDG1q6DFdz9\nxjrW7SlmxsSBPHT5WOKjwv1oYBDiXVI+UEJSXwe5q+Dk61o/b/hFpj5Yd0oDzl9rVoGnnhRoSyxd\nlPYuUx4O9PWnId2GIqfqbxsbWqkqb6zK4dInPyG7oJynrp/An68Z3/1FBJwudhLYOMn+TVBTbla0\nt4Z3GnB3IX+tcdlFdCN3qaVT8TVGUsbRMZJ9mCq8lqYUZZsSE20oM1FUUcP972xg4eb9nH5iEn+6\n5mT6JwRh+1m3iIwzMzhP0cBAkNPCQsSmdLc0YFUjJCOmB9oSSxfGV9dWN3EGdwKFO9rk1lq2rYB7\n3lzPocM1/PLSUdx25gmmp3lPo38G5H0ZuPvnLIeEwb4tPmtMA956pMpsV6UkFw4fhIE2PmJpPz65\ntkTkKhFJ8HqfKCJXumdWF6Zop08ZW1W19Tw8dzM3zVpJQnQ47911BnecfWLPFBEwC+GKc0xPjM7G\n08jqeG4tD5404O3dwL3laa1rA+2WDuBrjORBVW0s3OS0w33QHZO6MHXVpsjbcWYkmXtLueKpz3jh\ns13cfPpQ5v7oTMYO6D4NsNqFJ+C+b2Pn3/vQLijf57uQeNKAu0OcJH+tWRDab1ygLbF0YXwVkubO\n68Kr4FzCk/qb1PyMpKFB+dcn2Vzx1GcUVtTw71um8NDlY4kKD9I6WJ1JqlPbKhAl5Zs2svKF4ReY\ncirVZe7Y1FnkrzWNn4KpQ6Wly+GrkKwWkT+LyInOz5+BADq0g5RWUn/3lVRx06yV/O5/mUwbmcLC\nu8/inJE28a2R2BSI6x+YzK2c5aYlckob4h3DLoSGWpMG3FXxBNqtW8vSQXwVkh8BNcAbwOtAFUf6\nq7eIiEwXka0isl1E7mvm+BAR+VBENojIUhFJ8zr2RxHZLCKZIvKkOCvxROQ6EdnoXLNARJJ9fAb3\naUFI5m/cy/S/LePL3Yd4bMZJzLxxEkmx9hvgMaRmBGhGssKk9Ya0IRt+8GldPw24KBuqSuyKdkuH\n8ekvR1UrVPU+VZ2sqlNU9ZeqWtHaNU4fk6eBSzD9168TkabNk58AXlLVDOAR4DHn2tOBM4AMYBww\nBZgmImHA34BznWs2AD/08Vndp3AHRCVCTB8AyqvruOfN9Xz/lTUM6RPDvJ+cxXVTB3fP1en+oH8G\nHNxm+kV0FhUHzT2HtLGVatM0YDeoPAS5Lk787Yp2i5/wNWtrsYgker3vLSLHq888FdiuqtmqWoOZ\nyVzR5JwxwEfO9hKv44opxRIBRALhwH5AnJ9ezgwlnmAq1eJV9bekspZrnl3OO2ty+dF5w3jr+6eT\nntwrwAYGOakZpmjiga867557vjCv7enJPewCk1zR0b7gLfHOnfD8hVCS5874+WshNLL7FaC0dDq+\nzuWTnUwtAFT1EMdf2T4Q2OP1PtfZ5816YIazfRUQJyJJqrocIyx7nZ+FqpqpqrXA94GNGAEZAzzf\n3M1F5A4RWS0iqwsKCnx5xo7jCElFdR03v7CSrANlPH/zFH5+0Uj/9jrvrniXSukscpabD9P2fCt3\nMw04azFkLTTCuuYl/48PpplV6kkQ2gOqJ1hcxddPtwYRGex5IyJDaaYacDu4B+OyWgtMA/KAehEZ\nBowG0jDic56InCUi4RghmQAMwLi27m9uYFWd6bjiJqekpPjB1ONQVwMle6hLTOe2F1ezIbeEv183\nkXNtQN13EoeYoHdnBtxzVpgGSO3JWnIrDbiuBhbcb9YjpU+DNS+aWmD+pKEB9q6zbi2LX/BVSH4F\nfCoi/xGRl4GPaeED3Is8YJDX+zRnXyOqmq+qM1R1gnMPzxqVq4AVqlququXAfOA0YLxzzg5VVWA2\n4L/GHx2heDdoAy9sCWHFzkL+9M2TmT4uNdBWdS1EOjfgXnPYfCv3df1Icwy/wMxqqsv9Z9fKmVCY\nBdMfg1O+Z6oib5vvv/HBjF9TbkTUYukgvgbbFwCTga3Aa8DPgeNFRFcBw0UkXUQigGuBOd4niEiy\niHhsuB+Y5Wzn4ATXnVnINCATI0RjRMQzxbjQ2R9w6guyAJifF82jV57ElRNsn+d2kZoB+zdDQ737\n98pfY1J42xMf8TDsQqiv8V8acPkB+PgPZtwRF5tqw/EDYVWzHtz2YwPtFj/ia7D9NuBDjIDcA/wH\neKi1a1S1DpNRtRDzYT9bVTeLyCMicrlz2jnAVhHZBvQDHnX2vwXswMRC1gPrVXWuquYDDwPLRGQD\nZobye98e1T0aGpT3P/4MgKsuOJvrTxl8nCssLZJ6EtRVwsEs9+/lWYg4aGr7x/CkAfsrTvLhI1B7\n2MxGAELDYNLNkL3EZAX6i/y1EB4DySP8N6alx+Lr6vSfYFJwV6jquSIyCh8+wFV1HjCvyb4HvLbf\nwohG0+vqgTtbGPNZ4Fkf7XYdVeXhuZs5IXcr1ZGx3HiedRV0iMZSKRvcL4i4ezn0HdOmSs3HEBZh\n4hhZizteDTh/rWnwddpdpn+6h4k3wdLH4csX4KLftX/8pvfqfzKE2KoKlo7ja4ykSlWrAEQkUlW3\nACPdM6vr8MeFW3lx+W7O6lNKRL/hXb+seKBJHmGyqNwuKd9QD3tWdiw+4mH4hR1PA1aF+b+AXskw\n7f+OPhaXapporX3FtHLuKPV1Jg5lFyJa/ISvQpLrrCN5D1gsIv8FdrtnVtfg6SXbeWbpDr59ymDS\nQ/cjHWiva3EIDTe1n9zO3Nq/GWrKOhYf8eCPNOCNb5o1Lec/aDLXmjLlVqgsgq/ea/89PBRsMe5D\nGx+x+Alfg+1XqWqxqj4E/AazdqNHl5Gf9elO/t/CrcyYMJDffm0EUpzToT7tFi88mVturRgHr0ZW\nfpiRdDQNuLocFj9gPtjHf7v5c9KnQdIwWD2r+eNtwQbaLX6mzavkVPVjVZ3jrFbvkbyxKodH3v+K\n6WNT+eM3Mggp2QPa4FMfEosP9M+AqmLjLnKLnOUQnwaJfkqM6Ega8Kd/Nim+l/yx5XpfIjDpFjNr\n2bepY7bmr4XIePvFx+I37HLrNjJnfT73vbORaSNS+Nt14wkLDWm16q+lHbhdUl7VfOj7Yzbiob1p\nwEU74fOnIONbx88eG389hEXB6g6mAjcG2u2fv8U/2N+kNrD4q/387I11TBnah2dvmERkmJPxYoXE\nv/QbCxLiXpykOMfMAPwpJO1NA170awgJgwseOv65MX1g7AzYMLv9fVDqamD/JrsQ0eJXrJD4yKdZ\nB7nrlTWMHRDP89+ZTHSEV9pk0Q7jKugVPBXtuzQRMZA03L0ZSWN8xA+Bdg+NacBtqAa8Y4np/X7W\nzyB+gG/XTP6uWZG+YXb77Dyw2cycbHzE4keskPjA6l1F3P7Sak5I6cWL351KXFSTIndF2dAn3ab+\n+pP+Ge7NSHKWQ2SC/6veDr8ASnJMWfrjUV8HC+6D3kPhtDZ0QkibbBZtrp7VvmQEG2i3uIAVkuOw\nMbeEW15YRf+EKP5z6ykkxkQce5JX+XiLn0jNgNI8qCj0/9g5K0w8wt+L8YY5acC+ZG+tft6k4V70\nKIRH+X4PEZh8q3FP5a5qu435a80CzMQhbb/WYmkBKyStsG1/GTfN+oL46HBevu0UUuKaqRBbXwuH\ndtuMLX/jvcLdnxwugoJM/8ZHPCQOctKAF7V+XkUhLHnUNMYadVnb73PSNyEirn31t/Kc1rp29mzx\nI1ZIWkBV+eU7GwkPDeHV209hQGJ08ycW55ieEXZG4l9SXRKSjjSy8gVf0oCX/M4cn/6H9n2gR8bC\nyd+Cze8aYfSV2krTNMyuaLf4GSskLSAiPHX9RF69/RSGJLXS2dBmbLlDTB9IGOT/gHvOcggJdy9r\n6XhpwPs2wpf/hqm3d6yW2OTvQn01rHvF92v2bTJfemx8xOJnrJC0QmpCFMP6xrV+kkdIkqxry++k\nZkDuSpOy6i9yVpgP0vAWZpgdpbU0YE89rahEOOe+jt2n31gYdKoJujc0+HaNDbRbXMIKSUcpyjYf\nHL06oQtjT+Pka43rcNGv/DNebSXkrYEhLrm1oPU04M3vwu7P4PzfdKzisIcpt5rfv51LfTs/fy30\n6ut7qrHF4iNWSDpK4Q7j1rLBS/8z5nKTGrtypql821Hy13a8kZUvNJcGXHMYFv3GpO5O/I5/7jPm\nCohJ8r3+Vv4aG2i3uIIVko5iU3/d5YKHIf1seP+nZjbRERobWZ3Scbtao7k04M/+BqW5Tj0tP6Ud\nh0WaIo9b5kFpfuvnVpebMvd2RbvFBayQdIT6OtOr3QqJe4SGwTf+DbF94Y0boeJg+8fKWQEpo0wg\n300SB5n7eOIkxTnw2V9NeZMhp/v3XpNvMQH0NS+1ft6+DYDa+IjFFayQdISSHGios4F2t+mVBN96\nGQ4fhDdvNgLeVhoaIOcLd9aPNMewC2D352YmsOg3gMCFj/j/Pn1OgBPPhy9fbP3fxRNo7z/e/zZY\nejyuComITBeRrSKyXUSOSVMRkSEi8qGIbBCRpSKS5nXsjyKyWUQyReRJEePYFZEIEZkpIttEZIuI\nXO3mM7SKTf3tPAaMh6/9FXZ9Ynp3tJWCTKgucT8+4mH4RSYNeOljphnVmXebmYobTP4ulOXDtgUt\nn5O/FuIHQlw/d2yw9GhcExIRCQWeBi4BxgDXiciYJqc9AbykqhnAI8BjzrWnA2cAGcA4TL/4ac41\nvwIOqOoIZ9yP3XqG41K007xaIekcxl8HU++EFU/Dhjfbdq0nPtJZMxJPGvDyp8x6mNN/7N69Rkw3\nItFaefm8NdatZXENN2ckU4HtqprtNMF6HbiiyTljgI+c7SVexxWIAiKASCAc2O8c+y6O4Khqg6p2\nwGneQQp3QHgviLXf8jqNix+FwafDnB+1bbHi7uUQ17/zakx50oABLvqtqWjsFqFhJhNsx0dHZsne\nVBabCtVWSCwu4aaQDAS8W9zlOvu8WQ/McLavAuJEJElVl2OEZa/zs1BVM52+8QC/FZE1IvKmiDT7\nKS4id4jIahFZXVBQ4K9nOhpPxpZNp+w8QsPhmhfNOow3bvC9REjOCjMb6cz/qzN/CtPugzGd0JV6\n4o0gobD6hWOP7V1vXq2QWFwi0MH2e4BpIrIW47rKA+pFZBgwGkjDiM95InIWEObs+1xVJwLLMe6x\nY1DVmao6WVUnp6S4tFiwaIcpH2/pXGL7wrf+Y5pTvfVdaKhv/fziPSb1trPiIx4GTYFz7+8c8Yof\nAKMuhbUvQ23V0cfsinaLy7gpJHmAd3QxzdnXiKrmq+oMVZ2AiX2gqsWY2ckKVS1X1XJgPnAaUAgc\nBt5xhngTCExifH2dqfprM7YCQ9pkuPQJyF4CHx4nG6qxkVUnxUcCxeRbobIIMuccvT9/jXHpuZ32\nbOmxuCkkq4DhIpIuIhHAtcBRv+EikiwiHhvuBzxLdHMwM5UwEQnHzFYyVVWBucA5znnnA1+5+Awt\nU5prVknbQHvgmPQdmHSLWaOx+d2Wz8tZbsqu9xvXebYFgvRp5vexaXn5/LV2IaLFVVwTElWtA34I\nLAQygdmqullEHhGRy53TzgG2isg2oB/wqLP/LWAHsBETR1mvqnOdY78AHhKRDcCNwM/deoZWKdxh\nXm0fksByyR8gbQq8d+g9UI0AAA6vSURBVBfsb+E7hVuNrIKNkBCTCrxnBezfbPZVFJoFkdatZXER\nV2MkqjpPVUeo6omq+qiz7wFVneNsv6Wqw51zblPVamd/vareqaqjVXWMqv7Ma8zdqnq2qmao6vmq\nmuPmM7SIXUMSHIRFwjX/MT063vi2yVDypvKQ6cHR2fGRQDH+2xAaeaT+1l4bH7G4T6CD7V2Xop0Q\nHgNxqYG2xBLfH775ovnm/c7tR5dV37MK0O4fH/EQ0wfGXgXr3zCr6htXtJ8cWLss3RorJO2lyFb9\nDSqGnAbTHzdtbpc+dmR/znIICYOBkwJnW2cz5VaoKYONs01r3aRhEJUQaKss3ZiwQBvQZSnKhpSR\ngbbC4s2U2yB/HSz7oympMuoyEx/pP97dBYHBRtoU6HcSrJoFhwth6JmBtsjSzbEzkvbQUA+Hdtn4\nSLAhApf9ycQD3rnTtLXN+7LnuLU8iJiqwPs3mhpcNj5icRkrJO2hJNcU5LMZW8FHeJSpFBwWCS9+\n3fQ17ymBdm8yrjG1vsAKicV1rJC0B5uxFdwkpME3/w1VpeZ9T5uRAETGwcnXmQyu1JMCbY2lm2Nj\nJO3BCknwk34WfP1vkLcaeiUH2prAcOEjZl1JZGygLbF0c6yQtIeibAiLNtVkLcHLxBvNT08lIgb6\nNe3cYLH4H+vaag9F2aZYY4j957NYLBb7SdgeCndYt5bFYrE4WCFpKw31cGinFRKLxWJxsELSVkrz\nndRfKyQWi8UCVkjaTpFT9df2IbFYLBbACknbsam/FovFchRWSNpKUTaERUHcgEBbYrFYLEGBFZK2\nUpgNvW3qr8VisXiwn4ZtpSjburUsFovFC1eFRESmi8hWEdkuIvc1c3yIiHwoIhtEZKmIpHkd+6OI\nbBaRTBF5UuToxh8iMkdENrlp/zE0NDipv+mdeluLxWIJZlwTEhEJBZ4GLgHGANeJSNN6DU8AL6lq\nBvAI8Jhz7enAGUAGMA6YAkzzGnsGUO6W7S1Slg91VTZjy2KxWLxwc0YyFdiuqtmqWgO8DlzR5Jwx\nwEfO9hKv4wpEARFAJBAO7AcQkVjgZ8DvXLS9eWzGlsVisRyDm0IyENjj9T7X2efNemCGs30VECci\nSaq6HCMse52fhaqa6Zz3W+BPwGG3DG+RQmcNiRUSi8ViaSTQwfZ7gGkishbjusoD6kVkGDAaSMOI\nz3kicpaIjAdOVNV3jzewiNwhIqtFZHVBQYF/rC3KNv0d4tOOf67FYrH0ENwsI58HDPJ6n+bsa0RV\n83FmJI7L6mpVLRaR24EVqlruHJsPnAaUAZNFZJdje18RWaqq5zS9uarOBGYCTJ48Wf3yREXZ0Huo\nTf21WCwWL9z8RFwFDBeRdBGJAK4F5nifICLJIuKx4X5glrOdg5mphIlIOGa2kqmqz6jqAFUdCpwJ\nbGtORFyjKNsG2i0Wi6UJrgmJqtYBPwQWApnAbFXdLCKPiMjlzmnn8P/bu/dYOcoyjuPfH6ctCCVt\nobVgi+UapRiEemyAcosYbIlyE7WIiKAiERL7B9ESFEkjUfCaGiKgEos2UKmgBUu4FERJLFBqL5TS\n20mV1t7swUJBhbaPf8x76rDdPbs5s7N7Kr9PsjmzM+/sPPvuTJ/OO++8AyskrQRGAjel+bOBNcBS\nsusoiyPigbJibciuXdDtUX/NzCqV+oTEiJgLzK2Yd0NuejZZ0qhcbyfwpTqfvZasa3BrbN8IO/7l\ne0jMzCq4sb9Ru3tsuWnLzCzPiaRRvofEzKwqJ5JGdXdBxyAY4q6/ZmZ5TiSN6l6Tuv52tDsSM7N+\nxYmkUe6xZWZWlRNJIyI8fLyZWQ1OJI14dSO8+boTiZlZFU4kjXCPLTOzmpxIGtGd7iHx8ChmZntw\nImlEdxfsM9Cj/pqZVeFE0ojuLhg2BjpKHVHGzGyv5ETSiK1dHhrFzKwGJ5J63PXXzKxXTiT1bN8M\nb77mRGJmVoMTST27e2w5kZiZVeNEUo/vITEz65UTST3dXbDPABjy7nZHYmbWLzmR1LN1DQx1118z\ns1pKTSSSJkpaIWm1pKlVlo+RNE/SEkl/kDQ6t+wWScskLZc0XZn9Jf1e0otp2XfKjB9wjy0zszpK\nSySSOoBbgUnAWOBiSWMrin0PuCsijgemAd9O654CTACOJ3su+weBM3rWiYj3AicCEyRNKus7uOuv\nmVl9ZZ6RjAdWR0RXRLwB3AOcV1FmLPB4mn4itzyA/YBBwL7AQGBTRLweEU8ApM9cCJQ3bslrW+CN\n7R5jy8ysF2UmklHAS7n369K8vMXAhWn6AuBASQdHxJ/JEsuG9Ho4IpbnV5Q0FPgYMK/axiVdKWmB\npAVbtmzp2zdwjy0zs7rafbH9WuAMSX8ha7paD+yUdDRwLNnZxijgQ5JO61lJ0gDgbmB6RHRV++CI\nuCMiOiOic8SIEX2Lbmu6h8SJxMyspjK7Iq0HDsu9H53m7RYRfyedkUgaDHw8Iv4p6YvA/IjYnpY9\nBJwM/CmtegewKiJ+VGL82RmJOmCou/6amdVS5hnJs8Axko6QNAiYDMzJF5A0XFJPDNcBd6bpv5Gd\nqQyQNJDsbGV5WudbwBBgSomxZ7q7siTSMbD0TZmZ7a1KSyQRsQO4BniYLAn8OiKWSZom6dxU7Exg\nhaSVwEjgpjR/NrAGWEp2HWVxRDyQugdfT3aRfqGkRZK+UNZ34JD3wdjK/gFmZpaniGh3DKXr7OyM\nBQsWtDsMM7O9iqTnIqKzXrl2X2w3M7O9nBOJmZkV4kRiZmaFOJGYmVkhTiRmZlaIE4mZmRXiRGJm\nZoU4kZiZWSFvixsSJW0B/trH1YcD/2hiOM3m+IpxfMU4vmL6e3xjIqLuqLdvi0RShKQFjdzZ2S6O\nrxjHV4zjK6a/x9coN22ZmVkhTiRmZlaIE0l9d7Q7gDocXzGOrxjHV0x/j68hvkZiZmaF+IzEzMwK\ncSIxM7NCnEgSSRMlrZC0WtLUKsv3lTQrLX9a0uEtjO0wSU9IekHSMklfqVLmTEnb0lMjF0m6oVXx\npe2vlbQ0bXuPp4gpMz3V3xJJ41oY23ty9bJI0iuSplSUaWn9SbpT0mZJz+fmHSTpUUmr0t9hNda9\nLJVZJemyFsb3XUkvpt/vfklDa6zb675QYnw3Slqf+w3PqbFur8d6ifHNysW2VtKiGuuWXn9NFxFv\n+xfQQfZo3yOBQWSP9x1bUebLwG1pejIwq4XxHQqMS9MHAiurxHcm8GAb63AtMLyX5ecADwECTgKe\nbuNvvZHsRqu21R9wOjAOeD437xZgapqeCtxcZb2DgK70d1iaHtai+M4GBqTpm6vF18i+UGJ8NwLX\nNvD793qslxVfxfLvAze0q/6a/fIZSWY8sDoiuiLiDeAeoPJh7ecBM9L0bOAsSWpFcBGxISIWpulX\ngeXAqFZsu4nOA+6KzHxgqKRD2xDHWcCaiOjrSAdNERF/BLorZuf3sRnA+VVW/QjwaER0R8TLwKPA\nxFbEFxGPRMSO9HY+MLrZ221UjfprRCPHemG9xZf+3fgkcHezt9suTiSZUcBLuffr2PMf6t1l0sG0\nDTi4JdHlpCa1E4Gnqyw+WdJiSQ9JOq6lgUEAj0h6TtKVVZY3UsetMJnaB3A76w9gZERsSNMbgZFV\nyvSXeryC7Ayzmnr7QpmuSU1vd9ZoGuwP9XcasCkiVtVY3s766xMnkr2IpMHAb4ApEfFKxeKFZM01\n7wd+DPy2xeGdGhHjgEnA1ZJOb/H265I0CDgXuLfK4nbX31tE1sbRL/vmS7oe2AHMrFGkXfvCT4Cj\ngBOADWTNR/3RxfR+NtLvj6VKTiSZ9cBhufej07yqZSQNAIYAW1sSXbbNgWRJZGZE3Fe5PCJeiYjt\naXouMFDS8FbFFxHr09/NwP1kTQh5jdRx2SYBCyNiU+WCdtdfsqmnuS/93VylTFvrUdLngI8Cl6Rk\nt4cG9oVSRMSmiNgZEbuAn9bYbrvrbwBwITCrVpl21V8RTiSZZ4FjJB2R/tc6GZhTUWYO0NND5iLg\n8VoHUrOlNtWfA8sj4gc1yhzSc81G0niy37YliU7SAZIO7Jkmuyj7fEWxOcBnU++tk4BtuWacVqn5\nP8F21l9Ofh+7DPhdlTIPA2dLGpaabs5O80onaSLwVeDciHi9RplG9oWy4stfc7ugxnYbOdbL9GHg\nxYhYV21hO+uvkHZf7e8vL7JeRSvJenRcn+ZNIztoAPYjaxJZDTwDHNnC2E4la+ZYAixKr3OAq4Cr\nUplrgGVkvVDmA6e0ML4j03YXpxh66i8fn4BbU/0uBTpb/PseQJYYhuTmta3+yBLaBuBNsnb6z5Nd\nc5sHrAIeAw5KZTuBn+XWvSLth6uBy1sY32qy6ws9+2BPL8Z3AXN72xdaFN8v0761hCw5HFoZX3q/\nx7HeivjS/F/07HO5si2vv2a/PESKmZkV4qYtMzMrxInEzMwKcSIxM7NCnEjMzKwQJxIzMyvEicSs\nn0sjEz/Y7jjManEiMTOzQpxIzJpE0mckPZOeI3G7pA5J2yX9UNlzZOZJGpHKniBpfu7ZHsPS/KMl\nPZYGj1wo6aj08YMlzU7PA5nZqpGnzRrhRGLWBJKOBT4FTIiIE4CdwCVkd9QviIjjgCeBb6ZV7gK+\nFhHHk92N3TN/JnBrZINHnkJ2dzRkIz5PAcaS3f08ofQvZdagAe0OwOz/xFnAB4Bn08nCO8gGXdzF\n/wbo+xVwn6QhwNCIeDLNnwHcm8ZYGhUR9wNExL8B0uc9E2l8pvRkvcOBp8r/Wmb1OZGYNYeAGRFx\n3VtmSt+oKNfXMYn+k5veiY9d60fctGXWHPOAiyS9E3Y/f30M2TF2USrzaeCpiNgGvCzptDT/UuDJ\nyJ5+uU7S+ekz9pW0f0u/hVkf+H81Zk0QES9I+jrZk+32IRv19WrgNWB8WraZ7DoKZMPE35YSRRdw\neZp/KXC7pGnpMz7Rwq9h1ice/desRJK2R8TgdsdhViY3bZmZWSE+IzEzs0J8RmJmZoU4kZiZWSFO\nJGZmVogTiZmZFeJEYmZmhfwXc26zJYIgQJoAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAaQAAAEWCAYAAAApTuNLAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "81rbJgN0Ow0S",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "encoder_mix_model = keras.models.load_model('DarkMachine_encoder-mix.model')\n",
        "vae_mix_model = keras.models.load_model('DarkMachine_autoencoder-mix.model')"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "c-PiOivFUq2N",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Mixed Encoder as input, using Mixed Data\n",
        "# Training Input\n",
        "x_train_encoded_mix = encoder_mix_model.predict(X_Mix_train_N, batch_size=batch_size)\n",
        "# Testing Input\n",
        "x_test_encoded_mix = encoder_mix_model.predict(X_Mix_test_N, batch_size=batch_size)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "XYArWfOgpyjZ",
        "colab_type": "text"
      },
      "source": [
        "## Binary Classifier with Mixed Encoder Input"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "o0mEVX-RWjyN",
        "colab_type": "code",
        "outputId": "14029705-b141-4453-8e1f-85cb38e43c7a",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 885
        }
      },
      "source": [
        "# Binary Classifer Training/Testing (Sequential)\n",
        "## Classifer can be used for both VAE's \n",
        "history = biclass.fit(x_train_encoded_mix[0], \n",
        "                      y_Mix_train_N, \n",
        "                      batch_size=batch_size, \n",
        "                      epochs=epochs,\n",
        "                      validation_split=0.2,\n",
        "                      callbacks=[es])\n",
        "                      #validation_data=(x_test_encoded_mix[0], y_Mix_test_N))\n",
        "\n",
        "\n",
        "biclass.save_weights('bi-mix-weights.h5')\n",
        "biclass.save('binary-mix-class.model')"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train on 1936574 samples, validate on 484144 samples\n",
            "Epoch 1/50\n",
            "1936574/1936574 [==============================] - 25s 13us/step - loss: 0.1431 - acc: 0.9497 - val_loss: 0.1172 - val_acc: 0.9575\n",
            "Epoch 2/50\n",
            "1936574/1936574 [==============================] - 24s 12us/step - loss: 0.1135 - acc: 0.9592 - val_loss: 0.1100 - val_acc: 0.9603\n",
            "Epoch 3/50\n",
            "1936574/1936574 [==============================] - 25s 13us/step - loss: 0.1087 - acc: 0.9608 - val_loss: 0.1071 - val_acc: 0.9612\n",
            "Epoch 4/50\n",
            "1936574/1936574 [==============================] - 25s 13us/step - loss: 0.1054 - acc: 0.9620 - val_loss: 0.1046 - val_acc: 0.9618\n",
            "Epoch 5/50\n",
            "1936574/1936574 [==============================] - 25s 13us/step - loss: 0.1033 - acc: 0.9629 - val_loss: 0.1029 - val_acc: 0.9631\n",
            "Epoch 6/50\n",
            "1936574/1936574 [==============================] - 24s 12us/step - loss: 0.1022 - acc: 0.9633 - val_loss: 0.1023 - val_acc: 0.9630\n",
            "Epoch 7/50\n",
            "1936574/1936574 [==============================] - 25s 13us/step - loss: 0.1012 - acc: 0.9637 - val_loss: 0.1007 - val_acc: 0.9639\n",
            "Epoch 8/50\n",
            "1936574/1936574 [==============================] - 24s 12us/step - loss: 0.1004 - acc: 0.9641 - val_loss: 0.1003 - val_acc: 0.9640\n",
            "Epoch 9/50\n",
            "1936574/1936574 [==============================] - 24s 12us/step - loss: 0.0998 - acc: 0.9643 - val_loss: 0.0993 - val_acc: 0.9645\n",
            "Epoch 10/50\n",
            "1936574/1936574 [==============================] - 24s 13us/step - loss: 0.0993 - acc: 0.9646 - val_loss: 0.0992 - val_acc: 0.9643\n",
            "Epoch 11/50\n",
            "1936574/1936574 [==============================] - 25s 13us/step - loss: 0.0988 - acc: 0.9647 - val_loss: 0.0990 - val_acc: 0.9646\n",
            "Epoch 12/50\n",
            "1936574/1936574 [==============================] - 25s 13us/step - loss: 0.0984 - acc: 0.9649 - val_loss: 0.0981 - val_acc: 0.9649\n",
            "Epoch 13/50\n",
            "1936574/1936574 [==============================] - 25s 13us/step - loss: 0.0982 - acc: 0.9650 - val_loss: 0.0977 - val_acc: 0.9650\n",
            "Epoch 14/50\n",
            "1936574/1936574 [==============================] - 25s 13us/step - loss: 0.0979 - acc: 0.9652 - val_loss: 0.0981 - val_acc: 0.9650\n",
            "Epoch 15/50\n",
            "1936574/1936574 [==============================] - 25s 13us/step - loss: 0.0976 - acc: 0.9653 - val_loss: 0.0989 - val_acc: 0.9648\n",
            "Epoch 16/50\n",
            "1936574/1936574 [==============================] - 26s 13us/step - loss: 0.0973 - acc: 0.9654 - val_loss: 0.0972 - val_acc: 0.9653\n",
            "Epoch 17/50\n",
            "1936574/1936574 [==============================] - 26s 13us/step - loss: 0.0972 - acc: 0.9655 - val_loss: 0.0971 - val_acc: 0.9655\n",
            "Epoch 18/50\n",
            "1936574/1936574 [==============================] - 25s 13us/step - loss: 0.0971 - acc: 0.9655 - val_loss: 0.0979 - val_acc: 0.9649\n",
            "Epoch 19/50\n",
            "1936574/1936574 [==============================] - 25s 13us/step - loss: 0.0970 - acc: 0.9655 - val_loss: 0.0985 - val_acc: 0.9650\n",
            "Epoch 20/50\n",
            "1936574/1936574 [==============================] - 25s 13us/step - loss: 0.0969 - acc: 0.9655 - val_loss: 0.0970 - val_acc: 0.9653\n",
            "Epoch 21/50\n",
            "1936574/1936574 [==============================] - 25s 13us/step - loss: 0.0969 - acc: 0.9656 - val_loss: 0.0967 - val_acc: 0.9656\n",
            "Epoch 22/50\n",
            "1936574/1936574 [==============================] - 25s 13us/step - loss: 0.0967 - acc: 0.9657 - val_loss: 0.0968 - val_acc: 0.9655\n",
            "Epoch 23/50\n",
            "1936574/1936574 [==============================] - 25s 13us/step - loss: 0.0967 - acc: 0.9657 - val_loss: 0.0973 - val_acc: 0.9652\n",
            "Epoch 24/50\n",
            "1936574/1936574 [==============================] - 25s 13us/step - loss: 0.0967 - acc: 0.9657 - val_loss: 0.0968 - val_acc: 0.9655\n",
            "Epoch 00024: early stopping\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "mk1yUDEpXeBi",
        "colab_type": "code",
        "outputId": "fddcf577-2766-49fb-f9fd-6d4cb7d6ee0d",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 590
        }
      },
      "source": [
        "# list all data in history\n",
        "print(history.history.keys())\n",
        "# summarize history for accuracy\n",
        "plt.plot(history.history['acc'])\n",
        "plt.plot(history.history['val_acc'])\n",
        "plt.title('model accuracy')\n",
        "plt.ylabel('accuracy')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()\n",
        "# summarize history for loss\n",
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.title('model loss')\n",
        "plt.ylabel('loss')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "['acc', 'loss', 'val_acc', 'val_loss']\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZIAAAEWCAYAAABMoxE0AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n8UDinHOuQzyQOOec65D/H7qQX6+Y4lI6AAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYsAAAEWCAYAAACXGLsWAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "G47113s1XoWY",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "bi_mix_classifier = keras.models.load_model('binary-mix-class.model')\n",
        "\n",
        "pred_mix_x = bi_mix_classifier.predict(x_test_encoded_mix[0])"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "s1BiJFOKYSns",
        "colab_type": "text"
      },
      "source": [
        "## Mixed Input Binary Classifier ROC Curve"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "sRCdAJEtYEye",
        "colab_type": "code",
        "outputId": "22532d25-0984-44b2-cb99-e026540bdf3b",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 283
        }
      },
      "source": [
        "fpr_mb, tpr_mb, _ = roc_curve(y_Mix_test_N, pred_mix_x)\n",
        "\n",
        "roc_auc_mb = auc(fpr_mb, tpr_mb)\n",
        "\n",
        "plt.plot(fpr_mb,tpr_mb,color='darkorange',label='ROC curve (area = %0.2f)' % roc_auc_mb)\n",
        "plt.legend(loc=\"lower right\")\n",
        "plt.xlabel('False Positive Rate')\n",
        "plt.ylabel('True Positive Rate')\n",
        "\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEKCAYAAAD9xUlFAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "18Lz2G-caQWA",
        "colab_type": "text"
      },
      "source": [
        "## Histogram & ROC Curve of Mixed VAE"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "i11MIuZGYRyl",
        "colab_type": "code",
        "outputId": "8d5ac1d1-9b41-4954-e1d2-ee839098d2e3",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "vae_mix_predict_x = vae_mix_model.predict(X_Mix_test_N)\n",
        "predict_loss_mse = np.mean(np.power(X_Mix_test_N - vae_mix_predict_x, 2), axis=1)\n",
        "predict_loss_mse = predict_loss_mse.reshape(806907,1)\n",
        "predict_loss_mse.shape"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(806907, 1)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 42
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "eod24cWecM8R",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from sklearn.preprocessing import StandardScaler, RobustScaler\n",
        "\n",
        "predict_loss=predict_loss_mse\n",
        "tf_loss=StandardScaler().fit(predict_loss.reshape(-1,1))\n",
        "predict_loss=tf_loss.transform(predict_loss.reshape(-1,1))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7FPeQwv7cSdO",
        "colab_type": "code",
        "outputId": "df9cc4cf-9a00-4933-afd3-102e2faa473f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 281
        }
      },
      "source": [
        "sigPred = predict_loss[np.where(y_Mix_test==1)]\n",
        "bkgPred = predict_loss[np.where(y_Mix_test==0)]\n",
        "fig = plt.hist(sigPred, bins=100, histtype='step', label='signal_loss', normed= 1)\n",
        "fig = plt.hist(bkgPred, bins=100, histtype='step', color= 'red', label='background_loss', normed= 1)\n",
        "fig = plt.yscale('log')\n",
        "fig = plt.legend(loc= 'upper right')\n",
        "fig = plt.title('loss distribution')"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYAAAAEICAYAAABWJCMKAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "T_wvP3jMcajm",
        "colab_type": "code",
        "outputId": "c2b4b22b-ca4d-4ec1-9ec2-985ffbc83ca0",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 323
        }
      },
      "source": [
        "from sklearn.metrics import roc_auc_score\n",
        "\n",
        "fpr_vae_mix, tpr_vae_mix, thresholds_vae_mix = roc_curve(y_Mix_test, predict_loss)\n",
        "roc_auc = roc_auc_score(y_Mix_test, predict_loss)\n",
        "print(\"roc_auc =\", roc_auc_score(y_Mix_test, predict_loss))\n",
        "\n",
        "plt.plot(tpr_vae_mix, 1/fpr_vae_mix, color = 'darkorange', label='ROC curve (area = %0.2f)' % roc_auc)\n",
        "plt.xlim(0.1, 1.0)\n",
        "plt.ylim(1, 10000)\n",
        "plt.yscale(\"log\")\n",
        "plt.grid()\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "('roc_auc =', 0.6111050688110579)\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python2.7/dist-packages/ipykernel_launcher.py:7: RuntimeWarning: divide by zero encountered in divide\n",
            "  import sys\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAX4AAAD+CAYAAAA9HW6QAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nY5T4RUQy5v8DjJ/LwWsyD7wAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    }
  ]
}