example.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-03-01T21:58:11.455340Z",
     "start_time": "2021-03-01T21:57:42.322665Z"
    }
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "\n",
    "import tensorflow as tf\n",
    "from tensorflow import keras\n",
    "from tensorflow.keras.optimizers import Adam\n",
    "from tensorflow.keras.layers import Dense, Input, BatchNormalization\n",
    "from tensorflow.keras.models import Model\n",
    "from tensorflow.keras import regularizers\n",
    "from tensorflow.keras import backend as K\n",
    "from tensorflow.keras.datasets import mnist\n",
    "\n",
    "from realx import REALX"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Load Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-03-01T21:58:17.997673Z",
     "start_time": "2021-03-01T21:58:17.426915Z"
    }
   },
   "outputs": [],
   "source": [
    "#%% Parameters\n",
    "# MNIST    \n",
    "num_classes = 10\n",
    "\n",
    "# the data, split between train and test sets\n",
    "(x_train, y_train), (x_test, y_test) = mnist.load_data()\n",
    "\n",
    "x_train = x_train.reshape(60000, 784)\n",
    "x_test = x_test.reshape(10000, 784)\n",
    "x_train = x_train.astype('float32')\n",
    "x_test = x_test.astype('float32')\n",
    "x_train /= 255\n",
    "x_test /= 255\n",
    "\n",
    "# convert class vectors to binary class matrices\n",
    "y_train = tf.keras.utils.to_categorical(y_train, num_classes)\n",
    "y_test = tf.keras.utils.to_categorical(y_test, num_classes)\n",
    "\n",
    "input_shape = x_train.shape[1]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Set Training Parameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-03-01T21:58:24.352950Z",
     "start_time": "2021-03-01T21:58:24.344539Z"
    }
   },
   "outputs": [],
   "source": [
    "lamda = 5.\n",
    "epochs = 500\n",
    "batch_size = 1000\n",
    "optimizer = Adam(1e-4)\n",
    "loss = 'categorical_crossentropy'\n",
    "metrics = ['acc', 'AUC']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Create Selector Architecture"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-03-01T21:58:31.504854Z",
     "start_time": "2021-03-01T21:58:30.158379Z"
    }
   },
   "outputs": [],
   "source": [
    "model_input = Input(shape=(input_shape,), dtype='float32')\n",
    "out = Dense(100, activation='relu', kernel_regularizer=regularizers.l2(1e-3))(model_input)\n",
    "out = Dense(100, activation='relu', kernel_regularizer=regularizers.l2(1e-3))(out)\n",
    "select_prob = Dense(input_shape, kernel_regularizer=regularizers.l2(1e-3))(out)\n",
    "\n",
    "selector_model = Model(model_input, select_prob)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-03-01T18:01:21.722327Z",
     "start_time": "2021-03-01T18:01:21.718402Z"
    }
   },
   "source": [
    "# Create Predictor Architecture"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-03-01T21:58:33.908092Z",
     "start_time": "2021-03-01T21:58:33.726709Z"
    }
   },
   "outputs": [],
   "source": [
    "model_input = Input(shape=(input_shape,), dtype='float32')\n",
    "out= Dense(200, activation='relu', kernel_regularizer=regularizers.l2(1e-3))(model_input)\n",
    "out = BatchNormalization()(out)\n",
    "out= Dense(200, activation='relu', kernel_regularizer=regularizers.l2(1e-3))(out)\n",
    "out = BatchNormalization()(out)\n",
    "prob = Dense(num_classes, activation ='softmax', kernel_regularizer=regularizers.l2(1e-3))(out)\n",
    "\n",
    "predictor_model = Model(model_input, prob)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Initialize REAL-X"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-03-01T21:58:38.210813Z",
     "start_time": "2021-03-01T21:58:37.268520Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model_2\"\n",
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "input_3 (InputLayer)            [(None, 784)]        0                                            \n",
      "__________________________________________________________________________________________________\n",
      "flatten (Flatten)               (None, 784)          0           input_3[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "lambda (Lambda)                 (None, 784)          0           flatten[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "random__bernoulli__sampler (Ran (None, None)         0           lambda[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "resize_mask (ResizeMask)        (None, None)         0           random__bernoulli__sampler[0][0] \n",
      "__________________________________________________________________________________________________\n",
      "multiply (Multiply)             (None, 784)          0           input_3[0][0]                    \n",
      "                                                                 resize_mask[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "predictor (Model)               (None, 10)           200810      multiply[0][0]                   \n",
      "==================================================================================================\n",
      "Total params: 200,810\n",
      "Trainable params: 200,010\n",
      "Non-trainable params: 800\n",
      "__________________________________________________________________________________________________\n",
      "Predictor Built!\n",
      "Model: \"model_3\"\n",
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "input_4 (InputLayer)            [(None, 784)]        0                                            \n",
      "__________________________________________________________________________________________________\n",
      "flatten_1 (Flatten)             (None, 784)          0           input_4[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "lambda_1 (Lambda)               (None, 784)          0           flatten_1[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "random__bernoulli__sampler_1 (R (None, None)         0           lambda_1[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "resize_mask (ResizeMask)        (None, None)         0           random__bernoulli__sampler_1[0][0\n",
      "__________________________________________________________________________________________________\n",
      "multiply_1 (Multiply)           (None, 784)          0           input_4[0][0]                    \n",
      "                                                                 resize_mask[1][0]                \n",
      "__________________________________________________________________________________________________\n",
      "evalx (Model)                   (None, 10)           200810      multiply_1[0][0]                 \n",
      "==================================================================================================\n",
      "Total params: 200,810\n",
      "Trainable params: 200,010\n",
      "Non-trainable params: 800\n",
      "__________________________________________________________________________________________________\n",
      "Evaluator Built!\n"
     ]
    }
   ],
   "source": [
    "realx = REALX(selector_model, predictor_model, lamda)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-03-01T18:39:45.225887Z",
     "start_time": "2021-03-01T18:39:45.220716Z"
    }
   },
   "source": [
    "# Train the REAL-X Predictor"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-03-01T22:03:39.765604Z",
     "start_time": "2021-03-01T21:59:12.022445Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/500\n",
      "WARNING:tensorflow:From /misc/vlgscratch4/RanganathGroup/neil/anaconda3/envs/tf/lib/python3.7/site-packages/tensorflow/python/ops/resource_variable_ops.py:1817: calling BaseResourceVariable.__init__ (from tensorflow.python.ops.resource_variable_ops) with constraint is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "If using Keras pass *_constraint arguments to layers.\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 2.3294 - acc: 0.4051 - auc: 0.8022\n",
      "Epoch 2/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 1.4793 - acc: 0.6844 - auc: 0.9453\n",
      "Epoch 3/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 1.2275 - acc: 0.7674 - auc: 0.9678\n",
      "Epoch 4/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 1.0964 - acc: 0.8046 - auc: 0.9764\n",
      "Epoch 5/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 1.0141 - acc: 0.8291 - auc: 0.9805\n",
      "Epoch 6/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.9500 - acc: 0.8458 - auc: 0.9835\n",
      "Epoch 7/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.9004 - acc: 0.8563 - auc: 0.9857\n",
      "Epoch 8/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.8581 - acc: 0.8656 - auc: 0.9873\n",
      "Epoch 9/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.8194 - acc: 0.8752 - auc: 0.9887\n",
      "Epoch 10/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.7930 - acc: 0.8820 - auc: 0.9895\n",
      "Epoch 11/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.7632 - acc: 0.8879 - auc: 0.9905\n",
      "Epoch 12/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.7367 - acc: 0.8940 - auc: 0.9913\n",
      "Epoch 13/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.7179 - acc: 0.8972 - auc: 0.9917\n",
      "Epoch 14/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.6931 - acc: 0.9025 - auc: 0.9927\n",
      "Epoch 15/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.6766 - acc: 0.9059 - auc: 0.9930\n",
      "Epoch 16/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.6588 - acc: 0.9093 - auc: 0.9936\n",
      "Epoch 17/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.6435 - acc: 0.9126 - auc: 0.9937\n",
      "Epoch 18/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.6259 - acc: 0.9163 - auc: 0.9942\n",
      "Epoch 19/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.6122 - acc: 0.9176 - auc: 0.9945\n",
      "Epoch 20/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.5971 - acc: 0.9204 - auc: 0.9948\n",
      "Epoch 21/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.5836 - acc: 0.9230 - auc: 0.9951\n",
      "Epoch 22/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.5740 - acc: 0.9234 - auc: 0.9952\n",
      "Epoch 23/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.5607 - acc: 0.9268 - auc: 0.9955\n",
      "Epoch 24/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.5480 - acc: 0.9291 - auc: 0.9956\n",
      "Epoch 25/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.5378 - acc: 0.9304 - auc: 0.9959\n",
      "Epoch 26/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.5258 - acc: 0.9330 - auc: 0.9961\n",
      "Epoch 27/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.5144 - acc: 0.9346 - auc: 0.9963\n",
      "Epoch 28/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.5056 - acc: 0.9360 - auc: 0.9964\n",
      "Epoch 29/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.4961 - acc: 0.9374 - auc: 0.9966\n",
      "Epoch 30/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.4870 - acc: 0.9376 - auc: 0.9968\n",
      "Epoch 31/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.4744 - acc: 0.9406 - auc: 0.9969\n",
      "Epoch 32/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.4691 - acc: 0.9403 - auc: 0.9968\n",
      "Epoch 33/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.4593 - acc: 0.9421 - auc: 0.9971\n",
      "Epoch 34/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.4505 - acc: 0.9434 - auc: 0.9972\n",
      "Epoch 35/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.4426 - acc: 0.9443 - auc: 0.9973\n",
      "Epoch 36/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.4376 - acc: 0.9445 - auc: 0.9972\n",
      "Epoch 37/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.4301 - acc: 0.9443 - auc: 0.9974\n",
      "Epoch 38/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.4202 - acc: 0.9474 - auc: 0.9976\n",
      "Epoch 39/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.4137 - acc: 0.9485 - auc: 0.9975\n",
      "Epoch 40/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.4066 - acc: 0.9490 - auc: 0.9977\n",
      "Epoch 41/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.3979 - acc: 0.9499 - auc: 0.9978\n",
      "Epoch 42/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.3915 - acc: 0.9502 - auc: 0.9978\n",
      "Epoch 43/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.3844 - acc: 0.9526 - auc: 0.9979\n",
      "Epoch 44/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.3809 - acc: 0.9513 - auc: 0.9979\n",
      "Epoch 45/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.3733 - acc: 0.9527 - auc: 0.9980\n",
      "Epoch 46/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.3669 - acc: 0.9542 - auc: 0.9980\n",
      "Epoch 47/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.3618 - acc: 0.9538 - auc: 0.9981\n",
      "Epoch 48/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.3562 - acc: 0.9551 - auc: 0.9981\n",
      "Epoch 49/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.3503 - acc: 0.9560 - auc: 0.9981\n",
      "Epoch 50/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.3429 - acc: 0.9566 - auc: 0.9983\n",
      "Epoch 51/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.3404 - acc: 0.9564 - auc: 0.9982\n",
      "Epoch 52/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.3350 - acc: 0.9578 - auc: 0.9983\n",
      "Epoch 53/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.3292 - acc: 0.9570 - auc: 0.9984\n",
      "Epoch 54/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.3243 - acc: 0.9594 - auc: 0.9983\n",
      "Epoch 55/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.3183 - acc: 0.9598 - auc: 0.9985\n",
      "Epoch 56/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.3139 - acc: 0.9601 - auc: 0.9985\n",
      "Epoch 57/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.3089 - acc: 0.9607 - auc: 0.9986\n",
      "Epoch 58/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.3056 - acc: 0.9614 - auc: 0.9987\n",
      "Epoch 59/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.3022 - acc: 0.9611 - auc: 0.9985\n",
      "Epoch 60/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.2980 - acc: 0.9616 - auc: 0.9984\n",
      "Epoch 61/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.2934 - acc: 0.9626 - auc: 0.9985\n",
      "Epoch 62/500\n",
      "60/60 [==============================] - 1s 11ms/step - loss: 0.2891 - acc: 0.9621 - auc: 0.9986\n",
      "Epoch 63/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.2860 - acc: 0.9623 - auc: 0.9987\n",
      "Epoch 64/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.2803 - acc: 0.9624 - auc: 0.9988\n",
      "Epoch 65/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.2765 - acc: 0.9640 - auc: 0.9988\n",
      "Epoch 66/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.2713 - acc: 0.9645 - auc: 0.9988\n",
      "Epoch 67/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.2706 - acc: 0.9647 - auc: 0.9987\n",
      "Epoch 68/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.2662 - acc: 0.9653 - auc: 0.9988\n",
      "Epoch 69/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.2613 - acc: 0.9657 - auc: 0.9988\n",
      "Epoch 70/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.2591 - acc: 0.9661 - auc: 0.9989\n",
      "Epoch 71/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.2557 - acc: 0.9660 - auc: 0.9989\n",
      "Epoch 72/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.2519 - acc: 0.9667 - auc: 0.9988\n",
      "Epoch 73/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.2491 - acc: 0.9675 - auc: 0.9989\n",
      "Epoch 74/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.2478 - acc: 0.9667 - auc: 0.9989\n",
      "Epoch 75/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.2438 - acc: 0.9678 - auc: 0.9989\n",
      "Epoch 76/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.2414 - acc: 0.9671 - auc: 0.9989\n",
      "Epoch 77/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.2383 - acc: 0.9679 - auc: 0.9990\n",
      "Epoch 78/500\n",
      "60/60 [==============================] - 1s 12ms/step - loss: 0.2346 - acc: 0.9683 - auc: 0.9990\n",
      "Epoch 79/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.2325 - acc: 0.9690 - auc: 0.9991\n",
      "Epoch 80/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.2290 - acc: 0.9691 - auc: 0.9991\n",
      "Epoch 81/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.2284 - acc: 0.9686 - auc: 0.9990\n",
      "Epoch 82/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.2247 - acc: 0.9697 - auc: 0.9991\n",
      "Epoch 83/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.2210 - acc: 0.9701 - auc: 0.9992\n",
      "Epoch 84/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.2197 - acc: 0.9701 - auc: 0.9991\n",
      "Epoch 85/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.2158 - acc: 0.9703 - auc: 0.9992\n",
      "Epoch 86/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.2146 - acc: 0.9699 - auc: 0.9992\n",
      "Epoch 87/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.2104 - acc: 0.9713 - auc: 0.9991\n",
      "Epoch 88/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.2105 - acc: 0.9715 - auc: 0.9992\n",
      "Epoch 89/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.2062 - acc: 0.9715 - auc: 0.9992\n",
      "Epoch 90/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.2058 - acc: 0.9720 - auc: 0.9992\n",
      "Epoch 91/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.2035 - acc: 0.9711 - auc: 0.9992\n",
      "Epoch 92/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.2008 - acc: 0.9723 - auc: 0.9992\n",
      "Epoch 93/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1995 - acc: 0.9716 - auc: 0.9992\n",
      "Epoch 94/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1962 - acc: 0.9733 - auc: 0.9993\n",
      "Epoch 95/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1963 - acc: 0.9721 - auc: 0.9993\n",
      "Epoch 96/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1922 - acc: 0.9736 - auc: 0.9993\n",
      "Epoch 97/500\n",
      "60/60 [==============================] - 1s 11ms/step - loss: 0.1909 - acc: 0.9733 - auc: 0.9993\n",
      "Epoch 98/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1890 - acc: 0.9733 - auc: 0.9993\n",
      "Epoch 99/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1858 - acc: 0.9738 - auc: 0.9994\n",
      "Epoch 100/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.1862 - acc: 0.9730 - auc: 0.9992\n",
      "Epoch 101/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1818 - acc: 0.9748 - auc: 0.9994\n",
      "Epoch 102/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1809 - acc: 0.9748 - auc: 0.9994\n",
      "Epoch 103/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1803 - acc: 0.9744 - auc: 0.9993\n",
      "Epoch 104/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1791 - acc: 0.9735 - auc: 0.9994\n",
      "Epoch 105/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1784 - acc: 0.9740 - auc: 0.9993\n",
      "Epoch 106/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1745 - acc: 0.9750 - auc: 0.9994\n",
      "Epoch 107/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.1733 - acc: 0.9752 - auc: 0.9994\n",
      "Epoch 108/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1719 - acc: 0.9758 - auc: 0.9994\n",
      "Epoch 109/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1700 - acc: 0.9751 - auc: 0.9993\n",
      "Epoch 110/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.1685 - acc: 0.9752 - auc: 0.9995\n",
      "Epoch 111/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1679 - acc: 0.9753 - auc: 0.9995\n",
      "Epoch 112/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.1651 - acc: 0.9764 - auc: 0.9995\n",
      "Epoch 113/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.1629 - acc: 0.9770 - auc: 0.9995\n",
      "Epoch 114/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.1641 - acc: 0.9764 - auc: 0.9994\n",
      "Epoch 115/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1628 - acc: 0.9758 - auc: 0.9994\n",
      "Epoch 116/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.1606 - acc: 0.9765 - auc: 0.9994\n",
      "Epoch 117/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1582 - acc: 0.9766 - auc: 0.9995\n",
      "Epoch 118/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1570 - acc: 0.9771 - auc: 0.9995\n",
      "Epoch 119/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1574 - acc: 0.9768 - auc: 0.9995\n",
      "Epoch 120/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1550 - acc: 0.9772 - auc: 0.9995\n",
      "Epoch 121/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.1513 - acc: 0.9785 - auc: 0.9995\n",
      "Epoch 122/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1537 - acc: 0.9772 - auc: 0.9995\n",
      "Epoch 123/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1498 - acc: 0.9779 - auc: 0.9996\n",
      "Epoch 124/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1503 - acc: 0.9776 - auc: 0.9995\n",
      "Epoch 125/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.1478 - acc: 0.9782 - auc: 0.9996\n",
      "Epoch 126/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1477 - acc: 0.9779 - auc: 0.9996\n",
      "Epoch 127/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.1492 - acc: 0.9777 - auc: 0.9994\n",
      "Epoch 128/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1462 - acc: 0.9786 - auc: 0.9995\n",
      "Epoch 129/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.1458 - acc: 0.9783 - auc: 0.9995\n",
      "Epoch 130/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.1413 - acc: 0.9798 - auc: 0.9996\n",
      "Epoch 131/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1412 - acc: 0.9790 - auc: 0.9996\n",
      "Epoch 132/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1412 - acc: 0.9784 - auc: 0.9996\n",
      "Epoch 133/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1399 - acc: 0.9792 - auc: 0.9995\n",
      "Epoch 134/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1397 - acc: 0.9790 - auc: 0.9996\n",
      "Epoch 135/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1372 - acc: 0.9792 - auc: 0.9997\n",
      "Epoch 136/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1368 - acc: 0.9791 - auc: 0.9996\n",
      "Epoch 137/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1359 - acc: 0.9793 - auc: 0.9996\n",
      "Epoch 138/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1359 - acc: 0.9791 - auc: 0.9996\n",
      "Epoch 139/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1332 - acc: 0.9807 - auc: 0.9996\n",
      "Epoch 140/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.1331 - acc: 0.9800 - auc: 0.9996\n",
      "Epoch 141/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1317 - acc: 0.9801 - auc: 0.9996\n",
      "Epoch 142/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1307 - acc: 0.9801 - auc: 0.9997\n",
      "Epoch 143/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1287 - acc: 0.9807 - auc: 0.9997\n",
      "Epoch 144/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1294 - acc: 0.9799 - auc: 0.9996\n",
      "Epoch 145/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1280 - acc: 0.9802 - auc: 0.9996\n",
      "Epoch 146/500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1263 - acc: 0.9808 - auc: 0.9997\n",
      "Epoch 147/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.1261 - acc: 0.9810 - auc: 0.9996\n",
      "Epoch 148/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.1232 - acc: 0.9818 - auc: 0.9996\n",
      "Epoch 149/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1261 - acc: 0.9803 - auc: 0.9996\n",
      "Epoch 150/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1215 - acc: 0.9824 - auc: 0.9997\n",
      "Epoch 151/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1233 - acc: 0.9806 - auc: 0.9997\n",
      "Epoch 152/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.1218 - acc: 0.9814 - auc: 0.9997\n",
      "Epoch 153/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1213 - acc: 0.9816 - auc: 0.9997\n",
      "Epoch 154/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.1210 - acc: 0.9811 - auc: 0.9996\n",
      "Epoch 155/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1215 - acc: 0.9807 - auc: 0.9996\n",
      "Epoch 156/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1197 - acc: 0.9815 - auc: 0.9997\n",
      "Epoch 157/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1193 - acc: 0.9816 - auc: 0.9997\n",
      "Epoch 158/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.1183 - acc: 0.9819 - auc: 0.9997\n",
      "Epoch 159/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1175 - acc: 0.9820 - auc: 0.9997\n",
      "Epoch 160/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.1171 - acc: 0.9816 - auc: 0.9997\n",
      "Epoch 161/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.1145 - acc: 0.9821 - auc: 0.9997\n",
      "Epoch 162/500\n",
      "60/60 [==============================] - 1s 12ms/step - loss: 0.1159 - acc: 0.9815 - auc: 0.9996\n",
      "Epoch 163/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1136 - acc: 0.9823 - auc: 0.9997\n",
      "Epoch 164/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.1147 - acc: 0.9819 - auc: 0.9997\n",
      "Epoch 165/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.1116 - acc: 0.9830 - auc: 0.9997\n",
      "Epoch 166/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.1126 - acc: 0.9828 - auc: 0.9997\n",
      "Epoch 167/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.1121 - acc: 0.9827 - auc: 0.9997\n",
      "Epoch 168/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.1109 - acc: 0.9829 - auc: 0.9997\n",
      "Epoch 169/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1098 - acc: 0.9834 - auc: 0.9997\n",
      "Epoch 170/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1082 - acc: 0.9840 - auc: 0.9997\n",
      "Epoch 171/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.1099 - acc: 0.9832 - auc: 0.9997\n",
      "Epoch 172/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.1077 - acc: 0.9832 - auc: 0.9997\n",
      "Epoch 173/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.1065 - acc: 0.9839 - auc: 0.9997\n",
      "Epoch 174/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1073 - acc: 0.9839 - auc: 0.9996\n",
      "Epoch 175/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.1071 - acc: 0.9829 - auc: 0.9997\n",
      "Epoch 176/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1069 - acc: 0.9829 - auc: 0.9997\n",
      "Epoch 177/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1057 - acc: 0.9833 - auc: 0.9997\n",
      "Epoch 178/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.1053 - acc: 0.9836 - auc: 0.9998\n",
      "Epoch 179/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1044 - acc: 0.9835 - auc: 0.9997\n",
      "Epoch 180/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1040 - acc: 0.9836 - auc: 0.9997\n",
      "Epoch 181/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1040 - acc: 0.9833 - auc: 0.9998\n",
      "Epoch 182/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1024 - acc: 0.9842 - auc: 0.9997\n",
      "Epoch 183/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1026 - acc: 0.9835 - auc: 0.9997\n",
      "Epoch 184/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1015 - acc: 0.9840 - auc: 0.9997\n",
      "Epoch 185/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0995 - acc: 0.9844 - auc: 0.9998\n",
      "Epoch 186/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.1013 - acc: 0.9833 - auc: 0.9998\n",
      "Epoch 187/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1008 - acc: 0.9843 - auc: 0.9997\n",
      "Epoch 188/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0986 - acc: 0.9852 - auc: 0.9998\n",
      "Epoch 189/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0988 - acc: 0.9843 - auc: 0.9998\n",
      "Epoch 190/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0987 - acc: 0.9846 - auc: 0.9998\n",
      "Epoch 191/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0974 - acc: 0.9847 - auc: 0.9998\n",
      "Epoch 192/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0984 - acc: 0.9842 - auc: 0.9997\n",
      "Epoch 193/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0968 - acc: 0.9847 - auc: 0.9998\n",
      "Epoch 194/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0968 - acc: 0.9849 - auc: 0.9998\n",
      "Epoch 195/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0955 - acc: 0.9849 - auc: 0.9997\n",
      "Epoch 196/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0965 - acc: 0.9845 - auc: 0.9998\n",
      "Epoch 197/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0960 - acc: 0.9846 - auc: 0.9998\n",
      "Epoch 198/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0938 - acc: 0.9847 - auc: 0.9998\n",
      "Epoch 199/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0949 - acc: 0.9844 - auc: 0.9998\n",
      "Epoch 200/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0938 - acc: 0.9853 - auc: 0.9998\n",
      "Epoch 201/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0942 - acc: 0.9851 - auc: 0.9997\n",
      "Epoch 202/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0912 - acc: 0.9855 - auc: 0.9999\n",
      "Epoch 203/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0938 - acc: 0.9850 - auc: 0.9998\n",
      "Epoch 204/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0920 - acc: 0.9856 - auc: 0.9998\n",
      "Epoch 205/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0921 - acc: 0.9852 - auc: 0.9998\n",
      "Epoch 206/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0913 - acc: 0.9855 - auc: 0.9998\n",
      "Epoch 207/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0913 - acc: 0.9848 - auc: 0.9998\n",
      "Epoch 208/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0904 - acc: 0.9854 - auc: 0.9998\n",
      "Epoch 209/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0912 - acc: 0.9854 - auc: 0.9997\n",
      "Epoch 210/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0872 - acc: 0.9867 - auc: 0.9998\n",
      "Epoch 211/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0914 - acc: 0.9854 - auc: 0.9997\n",
      "Epoch 212/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0876 - acc: 0.9858 - auc: 0.9998\n",
      "Epoch 213/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0891 - acc: 0.9858 - auc: 0.9998\n",
      "Epoch 214/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0880 - acc: 0.9858 - auc: 0.9998\n",
      "Epoch 215/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0873 - acc: 0.9862 - auc: 0.9998\n",
      "Epoch 216/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0872 - acc: 0.9864 - auc: 0.9998\n",
      "Epoch 217/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0873 - acc: 0.9862 - auc: 0.9998\n",
      "Epoch 218/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0872 - acc: 0.9861 - auc: 0.9998\n",
      "Epoch 219/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0860 - acc: 0.9865 - auc: 0.9998\n",
      "Epoch 220/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0865 - acc: 0.9862 - auc: 0.9998\n",
      "Epoch 221/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0847 - acc: 0.9865 - auc: 0.9998\n",
      "Epoch 222/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0854 - acc: 0.9865 - auc: 0.9998\n",
      "Epoch 223/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0844 - acc: 0.9863 - auc: 0.9998\n",
      "Epoch 224/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0854 - acc: 0.9858 - auc: 0.9998\n",
      "Epoch 225/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0832 - acc: 0.9876 - auc: 0.9998\n",
      "Epoch 226/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0838 - acc: 0.9868 - auc: 0.9998\n",
      "Epoch 227/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0828 - acc: 0.9868 - auc: 0.9998\n",
      "Epoch 228/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0830 - acc: 0.9866 - auc: 0.9998\n",
      "Epoch 229/500\n",
      "60/60 [==============================] - 1s 11ms/step - loss: 0.0834 - acc: 0.9867 - auc: 0.9998\n",
      "Epoch 230/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0820 - acc: 0.9869 - auc: 0.9998\n",
      "Epoch 231/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0812 - acc: 0.9874 - auc: 0.9998\n",
      "Epoch 232/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0812 - acc: 0.9872 - auc: 0.9998\n",
      "Epoch 233/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0810 - acc: 0.9869 - auc: 0.9998\n",
      "Epoch 234/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0828 - acc: 0.9865 - auc: 0.9998\n",
      "Epoch 235/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0815 - acc: 0.9872 - auc: 0.9998\n",
      "Epoch 236/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0797 - acc: 0.9875 - auc: 0.9999\n",
      "Epoch 237/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0800 - acc: 0.9871 - auc: 0.9998\n",
      "Epoch 238/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0807 - acc: 0.9873 - auc: 0.9998\n",
      "Epoch 239/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0787 - acc: 0.9876 - auc: 0.9999\n",
      "Epoch 240/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0791 - acc: 0.9880 - auc: 0.9998\n",
      "Epoch 241/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0788 - acc: 0.9875 - auc: 0.9998\n",
      "Epoch 242/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0792 - acc: 0.9869 - auc: 0.9998\n",
      "Epoch 243/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0791 - acc: 0.9876 - auc: 0.9998\n",
      "Epoch 244/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0792 - acc: 0.9870 - auc: 0.9998\n",
      "Epoch 245/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0791 - acc: 0.9869 - auc: 0.9998\n",
      "Epoch 246/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0798 - acc: 0.9867 - auc: 0.9998\n",
      "Epoch 247/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0774 - acc: 0.9873 - auc: 0.9998\n",
      "Epoch 248/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0777 - acc: 0.9879 - auc: 0.9998\n",
      "Epoch 249/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0774 - acc: 0.9880 - auc: 0.9998\n",
      "Epoch 250/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0771 - acc: 0.9876 - auc: 0.9998\n",
      "Epoch 251/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0778 - acc: 0.9872 - auc: 0.9998\n",
      "Epoch 252/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0782 - acc: 0.9873 - auc: 0.9998\n",
      "Epoch 253/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0769 - acc: 0.9878 - auc: 0.9998\n",
      "Epoch 254/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0762 - acc: 0.9876 - auc: 0.9998\n",
      "Epoch 255/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0763 - acc: 0.9872 - auc: 0.9998\n",
      "Epoch 256/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0743 - acc: 0.9882 - auc: 0.9998\n",
      "Epoch 257/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0758 - acc: 0.9876 - auc: 0.9998\n",
      "Epoch 258/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0759 - acc: 0.9874 - auc: 0.9998\n",
      "Epoch 259/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0751 - acc: 0.9876 - auc: 0.9998\n",
      "Epoch 260/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0750 - acc: 0.9878 - auc: 0.9998\n",
      "Epoch 261/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0738 - acc: 0.9885 - auc: 0.9998\n",
      "Epoch 262/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0748 - acc: 0.9877 - auc: 0.9999\n",
      "Epoch 263/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0737 - acc: 0.9882 - auc: 0.9998\n",
      "Epoch 264/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0732 - acc: 0.9882 - auc: 0.9998\n",
      "Epoch 265/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0718 - acc: 0.9885 - auc: 0.9999\n",
      "Epoch 266/500\n",
      "60/60 [==============================] - 1s 12ms/step - loss: 0.0751 - acc: 0.9872 - auc: 0.9999\n",
      "Epoch 267/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0714 - acc: 0.9889 - auc: 0.9999\n",
      "Epoch 268/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0719 - acc: 0.9886 - auc: 0.9998\n",
      "Epoch 269/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0740 - acc: 0.9878 - auc: 0.9998\n",
      "Epoch 270/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0720 - acc: 0.9881 - auc: 0.9999\n",
      "Epoch 271/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0727 - acc: 0.9881 - auc: 0.9998\n",
      "Epoch 272/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0710 - acc: 0.9881 - auc: 0.9999\n",
      "Epoch 273/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0727 - acc: 0.9880 - auc: 0.9998\n",
      "Epoch 274/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0717 - acc: 0.9887 - auc: 0.9999\n",
      "Epoch 275/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0702 - acc: 0.9890 - auc: 0.9999\n",
      "Epoch 276/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0708 - acc: 0.9882 - auc: 0.9999\n",
      "Epoch 277/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0716 - acc: 0.9882 - auc: 0.9998\n",
      "Epoch 278/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0715 - acc: 0.9877 - auc: 0.9998\n",
      "Epoch 279/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0703 - acc: 0.9885 - auc: 0.9998\n",
      "Epoch 280/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0700 - acc: 0.9889 - auc: 0.9999\n",
      "Epoch 281/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0703 - acc: 0.9883 - auc: 0.9999\n",
      "Epoch 282/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0692 - acc: 0.9888 - auc: 0.9998\n",
      "Epoch 283/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0711 - acc: 0.9876 - auc: 0.9998\n",
      "Epoch 284/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0696 - acc: 0.9885 - auc: 0.9999\n",
      "Epoch 285/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0697 - acc: 0.9886 - auc: 0.9998\n",
      "Epoch 286/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0701 - acc: 0.9881 - auc: 0.9998\n",
      "Epoch 287/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0689 - acc: 0.9886 - auc: 0.9999\n",
      "Epoch 288/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0671 - acc: 0.9893 - auc: 0.9999\n",
      "Epoch 289/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0687 - acc: 0.9887 - auc: 0.9999\n",
      "Epoch 290/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0677 - acc: 0.9893 - auc: 0.9999\n",
      "Epoch 291/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0696 - acc: 0.9884 - auc: 0.9998\n",
      "Epoch 292/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0672 - acc: 0.9890 - auc: 0.9998\n",
      "Epoch 293/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0677 - acc: 0.9887 - auc: 0.9999\n",
      "Epoch 294/500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0690 - acc: 0.9884 - auc: 0.9998\n",
      "Epoch 295/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0678 - acc: 0.9889 - auc: 0.9999\n",
      "Epoch 296/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0685 - acc: 0.9888 - auc: 0.9998\n",
      "Epoch 297/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0676 - acc: 0.9886 - auc: 0.9999\n",
      "Epoch 298/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0665 - acc: 0.9887 - auc: 0.9999\n",
      "Epoch 299/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0682 - acc: 0.9887 - auc: 0.9998\n",
      "Epoch 300/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0668 - acc: 0.9893 - auc: 0.9999\n",
      "Epoch 301/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0673 - acc: 0.9889 - auc: 0.9999\n",
      "Epoch 302/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0677 - acc: 0.9890 - auc: 0.9998\n",
      "Epoch 303/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0666 - acc: 0.9886 - auc: 0.9999\n",
      "Epoch 304/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0669 - acc: 0.9885 - auc: 0.9999\n",
      "Epoch 305/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0674 - acc: 0.9885 - auc: 0.9999\n",
      "Epoch 306/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0654 - acc: 0.9888 - auc: 0.9999\n",
      "Epoch 307/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0661 - acc: 0.9892 - auc: 0.9999\n",
      "Epoch 308/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0662 - acc: 0.9891 - auc: 0.9998\n",
      "Epoch 309/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0664 - acc: 0.9885 - auc: 0.9999\n",
      "Epoch 310/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0666 - acc: 0.9883 - auc: 0.9999\n",
      "Epoch 311/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0664 - acc: 0.9894 - auc: 0.9998\n",
      "Epoch 312/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0638 - acc: 0.9895 - auc: 0.9999\n",
      "Epoch 313/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0655 - acc: 0.9890 - auc: 0.9999\n",
      "Epoch 314/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0652 - acc: 0.9893 - auc: 0.9998\n",
      "Epoch 315/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0650 - acc: 0.9897 - auc: 0.9999\n",
      "Epoch 316/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0641 - acc: 0.9894 - auc: 0.9999\n",
      "Epoch 317/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0651 - acc: 0.9890 - auc: 0.9999\n",
      "Epoch 318/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0639 - acc: 0.9892 - auc: 0.9999\n",
      "Epoch 319/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0639 - acc: 0.9893 - auc: 0.9999\n",
      "Epoch 320/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0638 - acc: 0.9895 - auc: 0.9999\n",
      "Epoch 321/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0628 - acc: 0.9899 - auc: 0.9999\n",
      "Epoch 322/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0641 - acc: 0.9894 - auc: 0.9999\n",
      "Epoch 323/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0642 - acc: 0.9891 - auc: 0.9999\n",
      "Epoch 324/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0638 - acc: 0.9894 - auc: 0.9999\n",
      "Epoch 325/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0630 - acc: 0.9898 - auc: 0.9999\n",
      "Epoch 326/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0636 - acc: 0.9898 - auc: 0.9998\n",
      "Epoch 327/500\n",
      "60/60 [==============================] - 1s 11ms/step - loss: 0.0627 - acc: 0.9895 - auc: 0.9999\n",
      "Epoch 328/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0634 - acc: 0.9894 - auc: 0.9999\n",
      "Epoch 329/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0634 - acc: 0.9888 - auc: 0.9999\n",
      "Epoch 330/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0618 - acc: 0.9901 - auc: 0.9999\n",
      "Epoch 331/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0616 - acc: 0.9900 - auc: 0.9999\n",
      "Epoch 332/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0634 - acc: 0.9891 - auc: 0.9999\n",
      "Epoch 333/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0626 - acc: 0.9898 - auc: 0.9998\n",
      "Epoch 334/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0633 - acc: 0.9895 - auc: 0.9999\n",
      "Epoch 335/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0625 - acc: 0.9898 - auc: 0.9999\n",
      "Epoch 336/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0623 - acc: 0.9894 - auc: 0.9999\n",
      "Epoch 337/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0624 - acc: 0.9896 - auc: 0.9998\n",
      "Epoch 338/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0613 - acc: 0.9898 - auc: 0.9999\n",
      "Epoch 339/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0617 - acc: 0.9897 - auc: 0.9998\n",
      "Epoch 340/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0595 - acc: 0.9909 - auc: 0.9999\n",
      "Epoch 341/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0625 - acc: 0.9890 - auc: 0.9999\n",
      "Epoch 342/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0614 - acc: 0.9900 - auc: 0.9999\n",
      "Epoch 343/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0607 - acc: 0.9901 - auc: 0.9999\n",
      "Epoch 344/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0610 - acc: 0.9901 - auc: 0.9999\n",
      "Epoch 345/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0614 - acc: 0.9899 - auc: 0.9999\n",
      "Epoch 346/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0609 - acc: 0.9902 - auc: 0.9999\n",
      "Epoch 347/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0595 - acc: 0.9903 - auc: 0.9999\n",
      "Epoch 348/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0607 - acc: 0.9901 - auc: 0.9999\n",
      "Epoch 349/500\n",
      "60/60 [==============================] - 1s 11ms/step - loss: 0.0621 - acc: 0.9890 - auc: 0.9998\n",
      "Epoch 350/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0591 - acc: 0.9905 - auc: 0.9999\n",
      "Epoch 351/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0599 - acc: 0.9901 - auc: 0.9999\n",
      "Epoch 352/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0591 - acc: 0.9910 - auc: 0.9999\n",
      "Epoch 353/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0603 - acc: 0.9906 - auc: 0.9999\n",
      "Epoch 354/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0598 - acc: 0.9901 - auc: 0.9999\n",
      "Epoch 355/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0588 - acc: 0.9906 - auc: 0.9999\n",
      "Epoch 356/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0591 - acc: 0.9900 - auc: 0.9999\n",
      "Epoch 357/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0607 - acc: 0.9897 - auc: 0.9999\n",
      "Epoch 358/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0600 - acc: 0.9897 - auc: 0.9999\n",
      "Epoch 359/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0596 - acc: 0.9898 - auc: 0.9999\n",
      "Epoch 360/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0595 - acc: 0.9901 - auc: 0.9999\n",
      "Epoch 361/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0591 - acc: 0.9902 - auc: 0.9999\n",
      "Epoch 362/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0587 - acc: 0.9903 - auc: 0.9999\n",
      "Epoch 363/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0602 - acc: 0.9899 - auc: 0.9999\n",
      "Epoch 364/500\n",
      "60/60 [==============================] - 1s 12ms/step - loss: 0.0590 - acc: 0.9906 - auc: 0.9998\n",
      "Epoch 365/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0600 - acc: 0.9893 - auc: 0.9999\n",
      "Epoch 366/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0603 - acc: 0.9891 - auc: 0.9999\n",
      "Epoch 367/500\n",
      "60/60 [==============================] - 1s 11ms/step - loss: 0.0575 - acc: 0.9912 - auc: 0.9999\n",
      "Epoch 368/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0592 - acc: 0.9899 - auc: 0.9999\n",
      "Epoch 369/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0596 - acc: 0.9898 - auc: 0.9999\n",
      "Epoch 370/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0606 - acc: 0.9890 - auc: 0.9999\n",
      "Epoch 371/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0580 - acc: 0.9904 - auc: 0.9999\n",
      "Epoch 372/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0574 - acc: 0.9908 - auc: 0.9999\n",
      "Epoch 373/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0599 - acc: 0.9900 - auc: 0.9998\n",
      "Epoch 374/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0591 - acc: 0.9902 - auc: 0.9999\n",
      "Epoch 375/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0584 - acc: 0.9899 - auc: 0.9999\n",
      "Epoch 376/500\n",
      "60/60 [==============================] - 1s 11ms/step - loss: 0.0589 - acc: 0.9900 - auc: 0.9999\n",
      "Epoch 377/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0574 - acc: 0.9909 - auc: 0.9999\n",
      "Epoch 378/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0564 - acc: 0.9908 - auc: 0.9999\n",
      "Epoch 379/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0579 - acc: 0.9899 - auc: 0.9999\n",
      "Epoch 380/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0588 - acc: 0.9901 - auc: 0.9999\n",
      "Epoch 381/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0580 - acc: 0.9906 - auc: 0.9999\n",
      "Epoch 382/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0586 - acc: 0.9904 - auc: 0.9999\n",
      "Epoch 383/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0577 - acc: 0.9904 - auc: 0.9999\n",
      "Epoch 384/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0570 - acc: 0.9910 - auc: 0.9999\n",
      "Epoch 385/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0569 - acc: 0.9904 - auc: 0.9999\n",
      "Epoch 386/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0583 - acc: 0.9901 - auc: 0.9999\n",
      "Epoch 387/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0580 - acc: 0.9903 - auc: 0.9999\n",
      "Epoch 388/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0576 - acc: 0.9907 - auc: 0.9998\n",
      "Epoch 389/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0566 - acc: 0.9905 - auc: 0.9999\n",
      "Epoch 390/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0576 - acc: 0.9905 - auc: 0.9998\n",
      "Epoch 391/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0568 - acc: 0.9905 - auc: 0.9998\n",
      "Epoch 392/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0567 - acc: 0.9903 - auc: 0.9999\n",
      "Epoch 393/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0555 - acc: 0.9912 - auc: 0.9999\n",
      "Epoch 394/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0577 - acc: 0.9901 - auc: 0.9999\n",
      "Epoch 395/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0564 - acc: 0.9907 - auc: 0.9999\n",
      "Epoch 396/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0569 - acc: 0.9900 - auc: 0.9999\n",
      "Epoch 397/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0569 - acc: 0.9905 - auc: 0.9999\n",
      "Epoch 398/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0561 - acc: 0.9908 - auc: 0.9999\n",
      "Epoch 399/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0573 - acc: 0.9905 - auc: 0.9999\n",
      "Epoch 400/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0580 - acc: 0.9898 - auc: 0.9999\n",
      "Epoch 401/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0563 - acc: 0.9906 - auc: 0.9999\n",
      "Epoch 402/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0574 - acc: 0.9898 - auc: 0.9999\n",
      "Epoch 403/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0575 - acc: 0.9901 - auc: 0.9999\n",
      "Epoch 404/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0566 - acc: 0.9902 - auc: 0.9999\n",
      "Epoch 405/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0555 - acc: 0.9911 - auc: 0.9999\n",
      "Epoch 406/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0583 - acc: 0.9894 - auc: 0.9999\n",
      "Epoch 407/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0563 - acc: 0.9905 - auc: 0.9999\n",
      "Epoch 408/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0564 - acc: 0.9900 - auc: 0.9999\n",
      "Epoch 409/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0566 - acc: 0.9906 - auc: 0.9999\n",
      "Epoch 410/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0547 - acc: 0.9910 - auc: 0.9999\n",
      "Epoch 411/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0563 - acc: 0.9903 - auc: 0.9999\n",
      "Epoch 412/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0563 - acc: 0.9905 - auc: 0.9998\n",
      "Epoch 413/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0546 - acc: 0.9911 - auc: 0.9999\n",
      "Epoch 414/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0544 - acc: 0.9909 - auc: 0.9999\n",
      "Epoch 415/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0562 - acc: 0.9904 - auc: 0.9999\n",
      "Epoch 416/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0551 - acc: 0.9905 - auc: 0.9999\n",
      "Epoch 417/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0562 - acc: 0.9906 - auc: 0.9998\n",
      "Epoch 418/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0560 - acc: 0.9905 - auc: 0.9998\n",
      "Epoch 419/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0550 - acc: 0.9907 - auc: 0.9999\n",
      "Epoch 420/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0539 - acc: 0.9914 - auc: 0.9999\n",
      "Epoch 421/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0552 - acc: 0.9907 - auc: 0.9999\n",
      "Epoch 422/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0539 - acc: 0.9912 - auc: 0.9999\n",
      "Epoch 423/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0550 - acc: 0.9907 - auc: 0.9999\n",
      "Epoch 424/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0550 - acc: 0.9906 - auc: 0.9999\n",
      "Epoch 425/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0546 - acc: 0.9908 - auc: 0.9999\n",
      "Epoch 426/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0555 - acc: 0.9903 - auc: 0.9999\n",
      "Epoch 427/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0554 - acc: 0.9908 - auc: 0.9999\n",
      "Epoch 428/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0553 - acc: 0.9907 - auc: 0.9999\n",
      "Epoch 429/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0551 - acc: 0.9902 - auc: 0.9999\n",
      "Epoch 430/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0556 - acc: 0.9905 - auc: 0.9999\n",
      "Epoch 431/500\n",
      "60/60 [==============================] - 1s 12ms/step - loss: 0.0542 - acc: 0.9908 - auc: 0.9999\n",
      "Epoch 432/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0553 - acc: 0.9906 - auc: 0.9999\n",
      "Epoch 433/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0545 - acc: 0.9906 - auc: 0.9999\n",
      "Epoch 434/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0558 - acc: 0.9902 - auc: 0.9999\n",
      "Epoch 435/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0540 - acc: 0.9910 - auc: 0.9999\n",
      "Epoch 436/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0537 - acc: 0.9912 - auc: 0.9999\n",
      "Epoch 437/500\n",
      "60/60 [==============================] - 1s 11ms/step - loss: 0.0541 - acc: 0.9910 - auc: 0.9999\n",
      "Epoch 438/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0540 - acc: 0.9913 - auc: 0.9999\n",
      "Epoch 439/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0530 - acc: 0.9914 - auc: 0.9999\n",
      "Epoch 440/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0551 - acc: 0.9908 - auc: 0.9999\n",
      "Epoch 441/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0543 - acc: 0.9910 - auc: 0.9999\n",
      "Epoch 442/500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0530 - acc: 0.9911 - auc: 0.9999\n",
      "Epoch 443/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0553 - acc: 0.9904 - auc: 0.9999\n",
      "Epoch 444/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0529 - acc: 0.9912 - auc: 0.9999\n",
      "Epoch 445/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0539 - acc: 0.9910 - auc: 0.9999\n",
      "Epoch 446/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0536 - acc: 0.9913 - auc: 0.9999\n",
      "Epoch 447/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0531 - acc: 0.9911 - auc: 0.9999\n",
      "Epoch 448/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0537 - acc: 0.9906 - auc: 0.9999\n",
      "Epoch 449/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0548 - acc: 0.9910 - auc: 0.9999\n",
      "Epoch 450/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0545 - acc: 0.9907 - auc: 0.9999\n",
      "Epoch 451/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0538 - acc: 0.9908 - auc: 0.9999\n",
      "Epoch 452/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0534 - acc: 0.9911 - auc: 0.9999\n",
      "Epoch 453/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0526 - acc: 0.9913 - auc: 0.9999\n",
      "Epoch 454/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0532 - acc: 0.9912 - auc: 0.9999\n",
      "Epoch 455/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0552 - acc: 0.9901 - auc: 0.9999\n",
      "Epoch 456/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0531 - acc: 0.9915 - auc: 0.9999\n",
      "Epoch 457/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0526 - acc: 0.9914 - auc: 0.9999\n",
      "Epoch 458/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0521 - acc: 0.9916 - auc: 0.9999\n",
      "Epoch 459/500\n",
      "60/60 [==============================] - 1s 11ms/step - loss: 0.0543 - acc: 0.9904 - auc: 0.9999\n",
      "Epoch 460/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0522 - acc: 0.9916 - auc: 0.9999\n",
      "Epoch 461/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0530 - acc: 0.9911 - auc: 0.9998\n",
      "Epoch 462/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0528 - acc: 0.9908 - auc: 0.9999\n",
      "Epoch 463/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0530 - acc: 0.9908 - auc: 0.9999\n",
      "Epoch 464/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0538 - acc: 0.9906 - auc: 0.9999\n",
      "Epoch 465/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0534 - acc: 0.9908 - auc: 0.9999\n",
      "Epoch 466/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0530 - acc: 0.9909 - auc: 0.9999\n",
      "Epoch 467/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0526 - acc: 0.9911 - auc: 0.9999\n",
      "Epoch 468/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0512 - acc: 0.9918 - auc: 0.9999\n",
      "Epoch 469/500\n",
      "60/60 [==============================] - 1s 11ms/step - loss: 0.0531 - acc: 0.9908 - auc: 0.9999\n",
      "Epoch 470/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0518 - acc: 0.9916 - auc: 0.9999\n",
      "Epoch 471/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0537 - acc: 0.9908 - auc: 0.9999\n",
      "Epoch 472/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0530 - acc: 0.9909 - auc: 0.9999\n",
      "Epoch 473/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0523 - acc: 0.9910 - auc: 0.9999\n",
      "Epoch 474/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0521 - acc: 0.9911 - auc: 0.9999\n",
      "Epoch 475/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0527 - acc: 0.9910 - auc: 0.9999\n",
      "Epoch 476/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0522 - acc: 0.9914 - auc: 0.9999\n",
      "Epoch 477/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0526 - acc: 0.9912 - auc: 0.9999\n",
      "Epoch 478/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0519 - acc: 0.9912 - auc: 0.9999\n",
      "Epoch 479/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0541 - acc: 0.9905 - auc: 0.9999\n",
      "Epoch 480/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0509 - acc: 0.9916 - auc: 0.9999\n",
      "Epoch 481/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0507 - acc: 0.9919 - auc: 0.9999\n",
      "Epoch 482/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0530 - acc: 0.9908 - auc: 0.9999\n",
      "Epoch 483/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0517 - acc: 0.9914 - auc: 0.9999\n",
      "Epoch 484/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0527 - acc: 0.9908 - auc: 0.9999\n",
      "Epoch 485/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0531 - acc: 0.9909 - auc: 0.9999\n",
      "Epoch 486/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0512 - acc: 0.9915 - auc: 0.9999\n",
      "Epoch 487/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0517 - acc: 0.9913 - auc: 0.9999\n",
      "Epoch 488/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0517 - acc: 0.9911 - auc: 0.9999\n",
      "Epoch 489/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0528 - acc: 0.9910 - auc: 0.9999\n",
      "Epoch 490/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0520 - acc: 0.9915 - auc: 0.9999\n",
      "Epoch 491/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0519 - acc: 0.9912 - auc: 0.9999\n",
      "Epoch 492/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0516 - acc: 0.9915 - auc: 0.9999\n",
      "Epoch 493/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0522 - acc: 0.9910 - auc: 0.9999\n",
      "Epoch 494/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0521 - acc: 0.9913 - auc: 0.9999\n",
      "Epoch 495/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0529 - acc: 0.9910 - auc: 0.9999\n",
      "Epoch 496/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0505 - acc: 0.9919 - auc: 0.9999\n",
      "Epoch 497/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0511 - acc: 0.9915 - auc: 0.9999\n",
      "Epoch 498/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0514 - acc: 0.9913 - auc: 0.9999\n",
      "Epoch 499/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0518 - acc: 0.9911 - auc: 0.9999\n",
      "Epoch 500/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0521 - acc: 0.9909 - auc: 0.9999\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<tensorflow.python.keras.callbacks.History at 0x7f7ff8c38d90>"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "realx.predictor.compile(loss=loss,\n",
    "                        optimizer=optimizer,\n",
    "                        metrics=metrics)\n",
    "realx.predictor.fit(x_train,\n",
    "                    y_train,\n",
    "                    epochs=epochs,\n",
    "                    batch_size=batch_size,\n",
    "                    verbose=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-03-01T18:54:19.645234Z",
     "start_time": "2021-03-01T18:54:19.636391Z"
    }
   },
   "source": [
    "# Train the REAL-X Selector"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-03-01T22:04:48.608417Z",
     "start_time": "2021-03-01T22:04:48.409519Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Selector Built!\n"
     ]
    }
   ],
   "source": [
    "# Build the Selector\n",
    "realx.build_selector()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-03-01T22:12:00.659651Z",
     "start_time": "2021-03-01T22:04:49.806682Z"
    },
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/500\n",
      "WARNING:tensorflow:Calling GradientTape.gradient on a persistent tape inside its context is significantly less efficient than calling it outside the context (it causes the gradient ops to be recorded on the tape, leading to increased CPU and memory usage). Only call GradientTape.gradient inside the context if you actually want to trace the gradient in order to compute higher order derivatives.\n",
      "60/60 [==============================] - 1s 15ms/step - loss: -1.3925 - sel%: 0.2442 - acc: 0.9456 - auc: 0.9975\n",
      "Epoch 2/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.5990 - sel%: 0.0900 - acc: 0.9561 - auc: 0.9983\n",
      "Epoch 3/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.5363 - sel%: 0.0808 - acc: 0.9624 - auc: 0.9989\n",
      "Epoch 4/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.5198 - sel%: 0.0778 - acc: 0.9637 - auc: 0.9989\n",
      "Epoch 5/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.5128 - sel%: 0.0754 - acc: 0.9633 - auc: 0.9988\n",
      "Epoch 6/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.5085 - sel%: 0.0751 - acc: 0.9639 - auc: 0.9989\n",
      "Epoch 7/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.5074 - sel%: 0.0752 - acc: 0.9641 - auc: 0.9989\n",
      "Epoch 8/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.5044 - sel%: 0.0753 - acc: 0.9650 - auc: 0.9990\n",
      "Epoch 9/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.5031 - sel%: 0.0750 - acc: 0.9653 - auc: 0.9989\n",
      "Epoch 10/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.5026 - sel%: 0.0745 - acc: 0.9640 - auc: 0.9989\n",
      "Epoch 11/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.5008 - sel%: 0.0742 - acc: 0.9653 - auc: 0.9989\n",
      "Epoch 12/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4995 - sel%: 0.0741 - acc: 0.9646 - auc: 0.9989\n",
      "Epoch 13/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4997 - sel%: 0.0737 - acc: 0.9641 - auc: 0.9989\n",
      "Epoch 14/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4978 - sel%: 0.0733 - acc: 0.9643 - auc: 0.9989\n",
      "Epoch 15/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4960 - sel%: 0.0730 - acc: 0.9639 - auc: 0.9989\n",
      "Epoch 16/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4955 - sel%: 0.0727 - acc: 0.9640 - auc: 0.9988\n",
      "Epoch 17/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4939 - sel%: 0.0724 - acc: 0.9640 - auc: 0.9989\n",
      "Epoch 18/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4927 - sel%: 0.0723 - acc: 0.9640 - auc: 0.9988\n",
      "Epoch 19/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4917 - sel%: 0.0730 - acc: 0.9653 - auc: 0.9989\n",
      "Epoch 20/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4903 - sel%: 0.0733 - acc: 0.9658 - auc: 0.9990\n",
      "Epoch 21/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4894 - sel%: 0.0732 - acc: 0.9664 - auc: 0.9989\n",
      "Epoch 22/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4880 - sel%: 0.0729 - acc: 0.9662 - auc: 0.9990\n",
      "Epoch 23/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4873 - sel%: 0.0721 - acc: 0.9657 - auc: 0.9989\n",
      "Epoch 24/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4835 - sel%: 0.0715 - acc: 0.9660 - auc: 0.9990\n",
      "Epoch 25/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4819 - sel%: 0.0712 - acc: 0.9658 - auc: 0.9990\n",
      "Epoch 26/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4815 - sel%: 0.0717 - acc: 0.9664 - auc: 0.9990\n",
      "Epoch 27/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4802 - sel%: 0.0713 - acc: 0.9670 - auc: 0.9990\n",
      "Epoch 28/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4802 - sel%: 0.0712 - acc: 0.9666 - auc: 0.9990\n",
      "Epoch 29/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4804 - sel%: 0.0713 - acc: 0.9666 - auc: 0.9990\n",
      "Epoch 30/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4793 - sel%: 0.0716 - acc: 0.9667 - auc: 0.9990\n",
      "Epoch 31/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4782 - sel%: 0.0720 - acc: 0.9679 - auc: 0.9991\n",
      "Epoch 32/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4773 - sel%: 0.0715 - acc: 0.9674 - auc: 0.9990\n",
      "Epoch 33/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4761 - sel%: 0.0715 - acc: 0.9679 - auc: 0.9991\n",
      "Epoch 34/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4757 - sel%: 0.0715 - acc: 0.9682 - auc: 0.9991\n",
      "Epoch 35/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4743 - sel%: 0.0710 - acc: 0.9678 - auc: 0.9991\n",
      "Epoch 36/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4746 - sel%: 0.0715 - acc: 0.9684 - auc: 0.9991\n",
      "Epoch 37/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4732 - sel%: 0.0713 - acc: 0.9686 - auc: 0.9991\n",
      "Epoch 38/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4700 - sel%: 0.0707 - acc: 0.9683 - auc: 0.9991\n",
      "Epoch 39/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4677 - sel%: 0.0699 - acc: 0.9692 - auc: 0.9991\n",
      "Epoch 40/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4672 - sel%: 0.0699 - acc: 0.9681 - auc: 0.9991\n",
      "Epoch 41/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4668 - sel%: 0.0699 - acc: 0.9689 - auc: 0.9991\n",
      "Epoch 42/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4650 - sel%: 0.0692 - acc: 0.9682 - auc: 0.9992\n",
      "Epoch 43/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4643 - sel%: 0.0691 - acc: 0.9687 - auc: 0.9991\n",
      "Epoch 44/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4632 - sel%: 0.0694 - acc: 0.9691 - auc: 0.9991\n",
      "Epoch 45/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4621 - sel%: 0.0688 - acc: 0.9685 - auc: 0.9992\n",
      "Epoch 46/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4625 - sel%: 0.0691 - acc: 0.9691 - auc: 0.9991\n",
      "Epoch 47/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4612 - sel%: 0.0693 - acc: 0.9694 - auc: 0.9992\n",
      "Epoch 48/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4604 - sel%: 0.0691 - acc: 0.9694 - auc: 0.9991\n",
      "Epoch 49/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4600 - sel%: 0.0686 - acc: 0.9687 - auc: 0.9992\n",
      "Epoch 50/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4594 - sel%: 0.0687 - acc: 0.9694 - auc: 0.9992\n",
      "Epoch 51/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4578 - sel%: 0.0687 - acc: 0.9701 - auc: 0.9992\n",
      "Epoch 52/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4573 - sel%: 0.0684 - acc: 0.9691 - auc: 0.9992\n",
      "Epoch 53/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4566 - sel%: 0.0676 - acc: 0.9680 - auc: 0.9992\n",
      "Epoch 54/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4534 - sel%: 0.0675 - acc: 0.9689 - auc: 0.9992\n",
      "Epoch 55/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4519 - sel%: 0.0673 - acc: 0.9701 - auc: 0.9992\n",
      "Epoch 56/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4504 - sel%: 0.0670 - acc: 0.9696 - auc: 0.9992\n",
      "Epoch 57/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4490 - sel%: 0.0669 - acc: 0.9702 - auc: 0.9992\n",
      "Epoch 58/500\n",
      "60/60 [==============================] - 1s 13ms/step - loss: -0.4486 - sel%: 0.0670 - acc: 0.9700 - auc: 0.9993\n",
      "Epoch 59/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4466 - sel%: 0.0672 - acc: 0.9711 - auc: 0.9993\n",
      "Epoch 60/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4454 - sel%: 0.0671 - acc: 0.9711 - auc: 0.9993\n",
      "Epoch 61/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4448 - sel%: 0.0673 - acc: 0.9716 - auc: 0.9993\n",
      "Epoch 62/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4432 - sel%: 0.0667 - acc: 0.9717 - auc: 0.9993\n",
      "Epoch 63/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4419 - sel%: 0.0661 - acc: 0.9711 - auc: 0.9993\n",
      "Epoch 64/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4420 - sel%: 0.0656 - acc: 0.9703 - auc: 0.9993\n",
      "Epoch 65/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4401 - sel%: 0.0658 - acc: 0.9713 - auc: 0.9992\n",
      "Epoch 66/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4385 - sel%: 0.0658 - acc: 0.9714 - auc: 0.9993\n",
      "Epoch 67/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4353 - sel%: 0.0652 - acc: 0.9717 - auc: 0.9993\n",
      "Epoch 68/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4348 - sel%: 0.0646 - acc: 0.9714 - auc: 0.9993\n",
      "Epoch 69/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4346 - sel%: 0.0649 - acc: 0.9714 - auc: 0.9993\n",
      "Epoch 70/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4335 - sel%: 0.0650 - acc: 0.9716 - auc: 0.9993\n",
      "Epoch 71/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4323 - sel%: 0.0645 - acc: 0.9712 - auc: 0.9993\n",
      "Epoch 72/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4315 - sel%: 0.0648 - acc: 0.9715 - auc: 0.9993\n",
      "Epoch 73/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4309 - sel%: 0.0644 - acc: 0.9717 - auc: 0.9994\n",
      "Epoch 74/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4298 - sel%: 0.0645 - acc: 0.9719 - auc: 0.9994\n",
      "Epoch 75/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4298 - sel%: 0.0645 - acc: 0.9719 - auc: 0.9994\n",
      "Epoch 76/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4292 - sel%: 0.0646 - acc: 0.9721 - auc: 0.9993\n",
      "Epoch 77/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4283 - sel%: 0.0646 - acc: 0.9727 - auc: 0.9994\n",
      "Epoch 78/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4287 - sel%: 0.0645 - acc: 0.9720 - auc: 0.9993\n",
      "Epoch 79/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4277 - sel%: 0.0649 - acc: 0.9727 - auc: 0.9994\n",
      "Epoch 80/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4268 - sel%: 0.0643 - acc: 0.9724 - auc: 0.9994\n",
      "Epoch 81/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4259 - sel%: 0.0638 - acc: 0.9719 - auc: 0.9994\n",
      "Epoch 82/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4250 - sel%: 0.0642 - acc: 0.9730 - auc: 0.9994\n",
      "Epoch 83/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4242 - sel%: 0.0637 - acc: 0.9723 - auc: 0.9993\n",
      "Epoch 84/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4242 - sel%: 0.0638 - acc: 0.9722 - auc: 0.9994\n",
      "Epoch 85/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4218 - sel%: 0.0636 - acc: 0.9731 - auc: 0.9994\n",
      "Epoch 86/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4220 - sel%: 0.0635 - acc: 0.9729 - auc: 0.9994\n",
      "Epoch 87/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4219 - sel%: 0.0631 - acc: 0.9716 - auc: 0.9993\n",
      "Epoch 88/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4226 - sel%: 0.0631 - acc: 0.9714 - auc: 0.9993\n",
      "Epoch 89/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4208 - sel%: 0.0634 - acc: 0.9729 - auc: 0.9994\n",
      "Epoch 90/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4208 - sel%: 0.0636 - acc: 0.9729 - auc: 0.9994\n",
      "Epoch 91/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4193 - sel%: 0.0634 - acc: 0.9734 - auc: 0.9994\n",
      "Epoch 92/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4193 - sel%: 0.0635 - acc: 0.9733 - auc: 0.9994\n",
      "Epoch 93/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4188 - sel%: 0.0631 - acc: 0.9729 - auc: 0.9994\n",
      "Epoch 94/500\n",
      "60/60 [==============================] - 1s 13ms/step - loss: -0.4177 - sel%: 0.0633 - acc: 0.9727 - auc: 0.9994\n",
      "Epoch 95/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4169 - sel%: 0.0629 - acc: 0.9726 - auc: 0.9994\n",
      "Epoch 96/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4160 - sel%: 0.0628 - acc: 0.9725 - auc: 0.9994\n",
      "Epoch 97/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4146 - sel%: 0.0628 - acc: 0.9733 - auc: 0.9995\n",
      "Epoch 98/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4145 - sel%: 0.0628 - acc: 0.9734 - auc: 0.9994\n",
      "Epoch 99/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4133 - sel%: 0.0627 - acc: 0.9733 - auc: 0.9994\n",
      "Epoch 100/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4127 - sel%: 0.0622 - acc: 0.9732 - auc: 0.9994\n",
      "Epoch 101/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4111 - sel%: 0.0627 - acc: 0.9737 - auc: 0.9995\n",
      "Epoch 102/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4101 - sel%: 0.0619 - acc: 0.9739 - auc: 0.9994\n",
      "Epoch 103/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4094 - sel%: 0.0617 - acc: 0.9730 - auc: 0.9995\n",
      "Epoch 104/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4084 - sel%: 0.0615 - acc: 0.9732 - auc: 0.9994\n",
      "Epoch 105/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4088 - sel%: 0.0616 - acc: 0.9729 - auc: 0.9995\n",
      "Epoch 106/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4083 - sel%: 0.0617 - acc: 0.9735 - auc: 0.9994\n",
      "Epoch 107/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4070 - sel%: 0.0615 - acc: 0.9733 - auc: 0.9994\n",
      "Epoch 108/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4074 - sel%: 0.0613 - acc: 0.9728 - auc: 0.9994\n",
      "Epoch 109/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4065 - sel%: 0.0610 - acc: 0.9726 - auc: 0.9995\n",
      "Epoch 110/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4067 - sel%: 0.0611 - acc: 0.9719 - auc: 0.9994\n",
      "Epoch 111/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4058 - sel%: 0.0617 - acc: 0.9745 - auc: 0.9995\n",
      "Epoch 112/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4047 - sel%: 0.0613 - acc: 0.9737 - auc: 0.9995\n",
      "Epoch 113/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4045 - sel%: 0.0609 - acc: 0.9728 - auc: 0.9994\n",
      "Epoch 114/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4041 - sel%: 0.0612 - acc: 0.9736 - auc: 0.9995\n",
      "Epoch 115/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4040 - sel%: 0.0607 - acc: 0.9726 - auc: 0.9994\n",
      "Epoch 116/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4029 - sel%: 0.0608 - acc: 0.9729 - auc: 0.9995\n",
      "Epoch 117/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4026 - sel%: 0.0604 - acc: 0.9729 - auc: 0.9994\n",
      "Epoch 118/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4020 - sel%: 0.0604 - acc: 0.9733 - auc: 0.9995\n",
      "Epoch 119/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4026 - sel%: 0.0607 - acc: 0.9731 - auc: 0.9995\n",
      "Epoch 120/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4014 - sel%: 0.0607 - acc: 0.9734 - auc: 0.9994\n",
      "Epoch 121/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4005 - sel%: 0.0607 - acc: 0.9740 - auc: 0.9995\n",
      "Epoch 122/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3998 - sel%: 0.0605 - acc: 0.9742 - auc: 0.9994\n",
      "Epoch 123/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4013 - sel%: 0.0607 - acc: 0.9734 - auc: 0.9995\n",
      "Epoch 124/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.4003 - sel%: 0.0603 - acc: 0.9732 - auc: 0.9995\n",
      "Epoch 125/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3987 - sel%: 0.0604 - acc: 0.9737 - auc: 0.9995\n",
      "Epoch 126/500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3962 - sel%: 0.0591 - acc: 0.9727 - auc: 0.9995\n",
      "Epoch 127/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3954 - sel%: 0.0593 - acc: 0.9737 - auc: 0.9994\n",
      "Epoch 128/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3953 - sel%: 0.0596 - acc: 0.9744 - auc: 0.9994\n",
      "Epoch 129/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3941 - sel%: 0.0594 - acc: 0.9737 - auc: 0.9995\n",
      "Epoch 130/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3947 - sel%: 0.0592 - acc: 0.9733 - auc: 0.9995\n",
      "Epoch 131/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3947 - sel%: 0.0588 - acc: 0.9729 - auc: 0.9994\n",
      "Epoch 132/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3932 - sel%: 0.0589 - acc: 0.9737 - auc: 0.9994\n",
      "Epoch 133/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3934 - sel%: 0.0588 - acc: 0.9733 - auc: 0.9994\n",
      "Epoch 134/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3941 - sel%: 0.0586 - acc: 0.9732 - auc: 0.9994\n",
      "Epoch 135/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3920 - sel%: 0.0586 - acc: 0.9737 - auc: 0.9995\n",
      "Epoch 136/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3921 - sel%: 0.0584 - acc: 0.9733 - auc: 0.9994\n",
      "Epoch 137/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3907 - sel%: 0.0573 - acc: 0.9730 - auc: 0.9994\n",
      "Epoch 138/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3898 - sel%: 0.0579 - acc: 0.9736 - auc: 0.9994\n",
      "Epoch 139/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3900 - sel%: 0.0575 - acc: 0.9721 - auc: 0.9994\n",
      "Epoch 140/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3886 - sel%: 0.0578 - acc: 0.9731 - auc: 0.9994\n",
      "Epoch 141/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3889 - sel%: 0.0578 - acc: 0.9732 - auc: 0.9994\n",
      "Epoch 142/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3878 - sel%: 0.0577 - acc: 0.9731 - auc: 0.9994\n",
      "Epoch 143/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3864 - sel%: 0.0572 - acc: 0.9726 - auc: 0.9994\n",
      "Epoch 144/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3861 - sel%: 0.0569 - acc: 0.9726 - auc: 0.9994\n",
      "Epoch 145/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3843 - sel%: 0.0570 - acc: 0.9734 - auc: 0.9994\n",
      "Epoch 146/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3855 - sel%: 0.0574 - acc: 0.9731 - auc: 0.9995\n",
      "Epoch 147/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3857 - sel%: 0.0571 - acc: 0.9732 - auc: 0.9994\n",
      "Epoch 148/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3850 - sel%: 0.0571 - acc: 0.9730 - auc: 0.9994\n",
      "Epoch 149/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3840 - sel%: 0.0573 - acc: 0.9740 - auc: 0.9994\n",
      "Epoch 150/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3837 - sel%: 0.0572 - acc: 0.9744 - auc: 0.9995\n",
      "Epoch 151/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3828 - sel%: 0.0567 - acc: 0.9736 - auc: 0.9994\n",
      "Epoch 152/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3825 - sel%: 0.0565 - acc: 0.9737 - auc: 0.9995\n",
      "Epoch 153/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3814 - sel%: 0.0565 - acc: 0.9728 - auc: 0.9995\n",
      "Epoch 154/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3816 - sel%: 0.0565 - acc: 0.9733 - auc: 0.9994\n",
      "Epoch 155/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3814 - sel%: 0.0563 - acc: 0.9738 - auc: 0.9994\n",
      "Epoch 156/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3807 - sel%: 0.0563 - acc: 0.9733 - auc: 0.9994\n",
      "Epoch 157/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3802 - sel%: 0.0564 - acc: 0.9736 - auc: 0.9995\n",
      "Epoch 158/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3795 - sel%: 0.0562 - acc: 0.9735 - auc: 0.9995\n",
      "Epoch 159/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3797 - sel%: 0.0562 - acc: 0.9733 - auc: 0.9994\n",
      "Epoch 160/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3780 - sel%: 0.0561 - acc: 0.9740 - auc: 0.9995\n",
      "Epoch 161/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3788 - sel%: 0.0559 - acc: 0.9737 - auc: 0.9994\n",
      "Epoch 162/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3789 - sel%: 0.0558 - acc: 0.9728 - auc: 0.9994\n",
      "Epoch 163/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3764 - sel%: 0.0558 - acc: 0.9735 - auc: 0.9994\n",
      "Epoch 164/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3762 - sel%: 0.0559 - acc: 0.9738 - auc: 0.9995\n",
      "Epoch 165/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3765 - sel%: 0.0551 - acc: 0.9726 - auc: 0.9994\n",
      "Epoch 166/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3773 - sel%: 0.0554 - acc: 0.9729 - auc: 0.9994\n",
      "Epoch 167/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3767 - sel%: 0.0557 - acc: 0.9730 - auc: 0.9994\n",
      "Epoch 168/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3757 - sel%: 0.0549 - acc: 0.9726 - auc: 0.9994\n",
      "Epoch 169/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3756 - sel%: 0.0550 - acc: 0.9728 - auc: 0.9994\n",
      "Epoch 170/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3745 - sel%: 0.0555 - acc: 0.9740 - auc: 0.9995\n",
      "Epoch 171/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3733 - sel%: 0.0553 - acc: 0.9740 - auc: 0.9994\n",
      "Epoch 172/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3723 - sel%: 0.0553 - acc: 0.9742 - auc: 0.9995\n",
      "Epoch 173/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3735 - sel%: 0.0557 - acc: 0.9744 - auc: 0.9995\n",
      "Epoch 174/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3729 - sel%: 0.0551 - acc: 0.9739 - auc: 0.9994\n",
      "Epoch 175/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3714 - sel%: 0.0553 - acc: 0.9745 - auc: 0.9994\n",
      "Epoch 176/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3717 - sel%: 0.0551 - acc: 0.9740 - auc: 0.9995\n",
      "Epoch 177/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3708 - sel%: 0.0552 - acc: 0.9744 - auc: 0.9995\n",
      "Epoch 178/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3707 - sel%: 0.0548 - acc: 0.9737 - auc: 0.9995\n",
      "Epoch 179/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3711 - sel%: 0.0549 - acc: 0.9738 - auc: 0.9995\n",
      "Epoch 180/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3708 - sel%: 0.0549 - acc: 0.9738 - auc: 0.9995\n",
      "Epoch 181/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3705 - sel%: 0.0545 - acc: 0.9738 - auc: 0.9994\n",
      "Epoch 182/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3691 - sel%: 0.0545 - acc: 0.9738 - auc: 0.9994\n",
      "Epoch 183/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3687 - sel%: 0.0541 - acc: 0.9738 - auc: 0.9995\n",
      "Epoch 184/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3688 - sel%: 0.0539 - acc: 0.9731 - auc: 0.9994\n",
      "Epoch 185/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3687 - sel%: 0.0546 - acc: 0.9739 - auc: 0.9994\n",
      "Epoch 186/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3679 - sel%: 0.0543 - acc: 0.9737 - auc: 0.9995\n",
      "Epoch 187/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3674 - sel%: 0.0541 - acc: 0.9746 - auc: 0.9994\n",
      "Epoch 188/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3671 - sel%: 0.0542 - acc: 0.9737 - auc: 0.9994\n",
      "Epoch 189/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3669 - sel%: 0.0541 - acc: 0.9733 - auc: 0.9994\n",
      "Epoch 190/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3670 - sel%: 0.0541 - acc: 0.9735 - auc: 0.9995\n",
      "Epoch 191/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3667 - sel%: 0.0541 - acc: 0.9744 - auc: 0.9995\n",
      "Epoch 192/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3663 - sel%: 0.0541 - acc: 0.9742 - auc: 0.9994\n",
      "Epoch 193/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3651 - sel%: 0.0537 - acc: 0.9744 - auc: 0.9994\n",
      "Epoch 194/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3665 - sel%: 0.0536 - acc: 0.9735 - auc: 0.9994\n",
      "Epoch 195/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3661 - sel%: 0.0537 - acc: 0.9734 - auc: 0.9994\n",
      "Epoch 196/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3653 - sel%: 0.0538 - acc: 0.9737 - auc: 0.9995\n",
      "Epoch 197/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3647 - sel%: 0.0540 - acc: 0.9748 - auc: 0.9995\n",
      "Epoch 198/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3638 - sel%: 0.0540 - acc: 0.9747 - auc: 0.9995\n",
      "Epoch 199/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3634 - sel%: 0.0536 - acc: 0.9750 - auc: 0.9994\n",
      "Epoch 200/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3633 - sel%: 0.0537 - acc: 0.9743 - auc: 0.9995\n",
      "Epoch 201/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3631 - sel%: 0.0539 - acc: 0.9748 - auc: 0.9994\n",
      "Epoch 202/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3635 - sel%: 0.0537 - acc: 0.9743 - auc: 0.9994\n",
      "Epoch 203/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3632 - sel%: 0.0537 - acc: 0.9740 - auc: 0.9994\n",
      "Epoch 204/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3624 - sel%: 0.0539 - acc: 0.9751 - auc: 0.9994\n",
      "Epoch 205/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3622 - sel%: 0.0538 - acc: 0.9755 - auc: 0.9995\n",
      "Epoch 206/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3611 - sel%: 0.0533 - acc: 0.9745 - auc: 0.9994\n",
      "Epoch 207/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3615 - sel%: 0.0533 - acc: 0.9744 - auc: 0.9995\n",
      "Epoch 208/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3615 - sel%: 0.0533 - acc: 0.9743 - auc: 0.9995\n",
      "Epoch 209/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3590 - sel%: 0.0531 - acc: 0.9751 - auc: 0.9995\n",
      "Epoch 210/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3587 - sel%: 0.0528 - acc: 0.9752 - auc: 0.9995\n",
      "Epoch 211/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3591 - sel%: 0.0530 - acc: 0.9750 - auc: 0.9995\n",
      "Epoch 212/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3600 - sel%: 0.0529 - acc: 0.9742 - auc: 0.9994\n",
      "Epoch 213/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3588 - sel%: 0.0534 - acc: 0.9755 - auc: 0.9995\n",
      "Epoch 214/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3584 - sel%: 0.0530 - acc: 0.9748 - auc: 0.9995\n",
      "Epoch 215/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3591 - sel%: 0.0528 - acc: 0.9740 - auc: 0.9994\n",
      "Epoch 216/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3576 - sel%: 0.0527 - acc: 0.9742 - auc: 0.9995\n",
      "Epoch 217/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3583 - sel%: 0.0529 - acc: 0.9750 - auc: 0.9995\n",
      "Epoch 218/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3576 - sel%: 0.0529 - acc: 0.9751 - auc: 0.9995\n",
      "Epoch 219/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3578 - sel%: 0.0527 - acc: 0.9748 - auc: 0.9994\n",
      "Epoch 220/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3576 - sel%: 0.0529 - acc: 0.9753 - auc: 0.9994\n",
      "Epoch 221/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3566 - sel%: 0.0532 - acc: 0.9751 - auc: 0.9995\n",
      "Epoch 222/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3574 - sel%: 0.0527 - acc: 0.9741 - auc: 0.9994\n",
      "Epoch 223/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3560 - sel%: 0.0528 - acc: 0.9748 - auc: 0.9995\n",
      "Epoch 224/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3564 - sel%: 0.0528 - acc: 0.9751 - auc: 0.9995\n",
      "Epoch 225/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3562 - sel%: 0.0529 - acc: 0.9753 - auc: 0.9995\n",
      "Epoch 226/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3552 - sel%: 0.0524 - acc: 0.9749 - auc: 0.9995\n",
      "Epoch 227/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3559 - sel%: 0.0526 - acc: 0.9751 - auc: 0.9995\n",
      "Epoch 228/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3549 - sel%: 0.0527 - acc: 0.9757 - auc: 0.9995\n",
      "Epoch 229/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3555 - sel%: 0.0527 - acc: 0.9751 - auc: 0.9995\n",
      "Epoch 230/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3550 - sel%: 0.0525 - acc: 0.9751 - auc: 0.9995\n",
      "Epoch 231/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3551 - sel%: 0.0524 - acc: 0.9754 - auc: 0.9995\n",
      "Epoch 232/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3549 - sel%: 0.0525 - acc: 0.9748 - auc: 0.9995\n",
      "Epoch 233/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3537 - sel%: 0.0528 - acc: 0.9764 - auc: 0.9995\n",
      "Epoch 234/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3540 - sel%: 0.0528 - acc: 0.9760 - auc: 0.9995\n",
      "Epoch 235/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3524 - sel%: 0.0525 - acc: 0.9765 - auc: 0.9996\n",
      "Epoch 236/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3536 - sel%: 0.0522 - acc: 0.9752 - auc: 0.9995\n",
      "Epoch 237/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3528 - sel%: 0.0523 - acc: 0.9759 - auc: 0.9995\n",
      "Epoch 238/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3528 - sel%: 0.0524 - acc: 0.9760 - auc: 0.9995\n",
      "Epoch 239/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3529 - sel%: 0.0524 - acc: 0.9761 - auc: 0.9995\n",
      "Epoch 240/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3521 - sel%: 0.0524 - acc: 0.9760 - auc: 0.9996\n",
      "Epoch 241/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3527 - sel%: 0.0526 - acc: 0.9763 - auc: 0.9995\n",
      "Epoch 242/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3527 - sel%: 0.0526 - acc: 0.9761 - auc: 0.9995\n",
      "Epoch 243/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3515 - sel%: 0.0525 - acc: 0.9760 - auc: 0.9995\n",
      "Epoch 244/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3519 - sel%: 0.0522 - acc: 0.9756 - auc: 0.9996\n",
      "Epoch 245/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3511 - sel%: 0.0522 - acc: 0.9760 - auc: 0.9996\n",
      "Epoch 246/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3503 - sel%: 0.0521 - acc: 0.9765 - auc: 0.9995\n",
      "Epoch 247/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3501 - sel%: 0.0519 - acc: 0.9762 - auc: 0.9996\n",
      "Epoch 248/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3494 - sel%: 0.0518 - acc: 0.9762 - auc: 0.9995\n",
      "Epoch 249/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3500 - sel%: 0.0523 - acc: 0.9761 - auc: 0.9995\n",
      "Epoch 250/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3485 - sel%: 0.0519 - acc: 0.9764 - auc: 0.9995\n",
      "Epoch 251/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3486 - sel%: 0.0519 - acc: 0.9760 - auc: 0.9995\n",
      "Epoch 252/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3492 - sel%: 0.0517 - acc: 0.9757 - auc: 0.9995\n",
      "Epoch 253/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3484 - sel%: 0.0522 - acc: 0.9770 - auc: 0.9996\n",
      "Epoch 254/500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3477 - sel%: 0.0522 - acc: 0.9766 - auc: 0.9996\n",
      "Epoch 255/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3480 - sel%: 0.0516 - acc: 0.9759 - auc: 0.9995\n",
      "Epoch 256/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3478 - sel%: 0.0517 - acc: 0.9763 - auc: 0.9996\n",
      "Epoch 257/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3480 - sel%: 0.0518 - acc: 0.9762 - auc: 0.9995\n",
      "Epoch 258/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3478 - sel%: 0.0515 - acc: 0.9754 - auc: 0.9995\n",
      "Epoch 259/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3468 - sel%: 0.0513 - acc: 0.9755 - auc: 0.9996\n",
      "Epoch 260/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3466 - sel%: 0.0512 - acc: 0.9756 - auc: 0.9995\n",
      "Epoch 261/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3454 - sel%: 0.0506 - acc: 0.9753 - auc: 0.9995\n",
      "Epoch 262/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3472 - sel%: 0.0510 - acc: 0.9744 - auc: 0.9995\n",
      "Epoch 263/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3465 - sel%: 0.0513 - acc: 0.9748 - auc: 0.9995\n",
      "Epoch 264/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3446 - sel%: 0.0511 - acc: 0.9764 - auc: 0.9995\n",
      "Epoch 265/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3458 - sel%: 0.0505 - acc: 0.9754 - auc: 0.9995\n",
      "Epoch 266/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3448 - sel%: 0.0511 - acc: 0.9763 - auc: 0.9996\n",
      "Epoch 267/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3448 - sel%: 0.0506 - acc: 0.9749 - auc: 0.9995\n",
      "Epoch 268/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3443 - sel%: 0.0510 - acc: 0.9764 - auc: 0.9996\n",
      "Epoch 269/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3443 - sel%: 0.0511 - acc: 0.9762 - auc: 0.9996\n",
      "Epoch 270/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3435 - sel%: 0.0507 - acc: 0.9757 - auc: 0.9996\n",
      "Epoch 271/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3431 - sel%: 0.0507 - acc: 0.9762 - auc: 0.9996\n",
      "Epoch 272/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3417 - sel%: 0.0506 - acc: 0.9766 - auc: 0.9996\n",
      "Epoch 273/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3434 - sel%: 0.0505 - acc: 0.9759 - auc: 0.9995\n",
      "Epoch 274/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3424 - sel%: 0.0502 - acc: 0.9755 - auc: 0.9995\n",
      "Epoch 275/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3413 - sel%: 0.0504 - acc: 0.9763 - auc: 0.9996\n",
      "Epoch 276/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3405 - sel%: 0.0508 - acc: 0.9766 - auc: 0.9995\n",
      "Epoch 277/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3408 - sel%: 0.0506 - acc: 0.9764 - auc: 0.9996\n",
      "Epoch 278/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3402 - sel%: 0.0508 - acc: 0.9770 - auc: 0.9996\n",
      "Epoch 279/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3406 - sel%: 0.0504 - acc: 0.9764 - auc: 0.9996\n",
      "Epoch 280/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3394 - sel%: 0.0500 - acc: 0.9757 - auc: 0.9996\n",
      "Epoch 281/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3404 - sel%: 0.0502 - acc: 0.9758 - auc: 0.9995\n",
      "Epoch 282/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3379 - sel%: 0.0504 - acc: 0.9769 - auc: 0.9996\n",
      "Epoch 283/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3383 - sel%: 0.0501 - acc: 0.9764 - auc: 0.9995\n",
      "Epoch 284/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3399 - sel%: 0.0500 - acc: 0.9756 - auc: 0.9996\n",
      "Epoch 285/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3396 - sel%: 0.0495 - acc: 0.9759 - auc: 0.9995\n",
      "Epoch 286/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3395 - sel%: 0.0501 - acc: 0.9764 - auc: 0.9996\n",
      "Epoch 287/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3395 - sel%: 0.0501 - acc: 0.9765 - auc: 0.9996\n",
      "Epoch 288/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3383 - sel%: 0.0502 - acc: 0.9764 - auc: 0.9996\n",
      "Epoch 289/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3375 - sel%: 0.0499 - acc: 0.9762 - auc: 0.9996\n",
      "Epoch 290/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3375 - sel%: 0.0500 - acc: 0.9772 - auc: 0.9996\n",
      "Epoch 291/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3377 - sel%: 0.0500 - acc: 0.9769 - auc: 0.9996\n",
      "Epoch 292/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3367 - sel%: 0.0501 - acc: 0.9770 - auc: 0.9996\n",
      "Epoch 293/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3380 - sel%: 0.0498 - acc: 0.9758 - auc: 0.9996\n",
      "Epoch 294/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3368 - sel%: 0.0499 - acc: 0.9767 - auc: 0.9996\n",
      "Epoch 295/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3364 - sel%: 0.0496 - acc: 0.9763 - auc: 0.9996\n",
      "Epoch 296/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3360 - sel%: 0.0495 - acc: 0.9765 - auc: 0.9996\n",
      "Epoch 297/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3370 - sel%: 0.0499 - acc: 0.9766 - auc: 0.9996\n",
      "Epoch 298/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3357 - sel%: 0.0498 - acc: 0.9770 - auc: 0.9996\n",
      "Epoch 299/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3367 - sel%: 0.0498 - acc: 0.9763 - auc: 0.9996\n",
      "Epoch 300/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3356 - sel%: 0.0497 - acc: 0.9767 - auc: 0.9996\n",
      "Epoch 301/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3350 - sel%: 0.0495 - acc: 0.9765 - auc: 0.9996\n",
      "Epoch 302/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3341 - sel%: 0.0496 - acc: 0.9774 - auc: 0.9996\n",
      "Epoch 303/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3352 - sel%: 0.0495 - acc: 0.9766 - auc: 0.9996\n",
      "Epoch 304/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3342 - sel%: 0.0497 - acc: 0.9776 - auc: 0.9996\n",
      "Epoch 305/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3336 - sel%: 0.0493 - acc: 0.9771 - auc: 0.9996\n",
      "Epoch 306/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3335 - sel%: 0.0487 - acc: 0.9759 - auc: 0.9995\n",
      "Epoch 307/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3315 - sel%: 0.0477 - acc: 0.9756 - auc: 0.9995\n",
      "Epoch 308/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3315 - sel%: 0.0478 - acc: 0.9752 - auc: 0.9995\n",
      "Epoch 309/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3310 - sel%: 0.0481 - acc: 0.9757 - auc: 0.9995\n",
      "Epoch 310/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3292 - sel%: 0.0477 - acc: 0.9758 - auc: 0.9996\n",
      "Epoch 311/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3293 - sel%: 0.0475 - acc: 0.9755 - auc: 0.9995\n",
      "Epoch 312/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3297 - sel%: 0.0472 - acc: 0.9747 - auc: 0.9995\n",
      "Epoch 313/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3282 - sel%: 0.0474 - acc: 0.9759 - auc: 0.9995\n",
      "Epoch 314/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3280 - sel%: 0.0473 - acc: 0.9760 - auc: 0.9995\n",
      "Epoch 315/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3281 - sel%: 0.0474 - acc: 0.9754 - auc: 0.9995\n",
      "Epoch 316/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3271 - sel%: 0.0476 - acc: 0.9762 - auc: 0.9995\n",
      "Epoch 317/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3266 - sel%: 0.0475 - acc: 0.9767 - auc: 0.9996\n",
      "Epoch 318/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3282 - sel%: 0.0479 - acc: 0.9765 - auc: 0.9995\n",
      "Epoch 319/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3264 - sel%: 0.0474 - acc: 0.9758 - auc: 0.9996\n",
      "Epoch 320/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3260 - sel%: 0.0468 - acc: 0.9755 - auc: 0.9995\n",
      "Epoch 321/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3269 - sel%: 0.0472 - acc: 0.9755 - auc: 0.9995\n",
      "Epoch 322/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3266 - sel%: 0.0473 - acc: 0.9752 - auc: 0.9996\n",
      "Epoch 323/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3247 - sel%: 0.0471 - acc: 0.9767 - auc: 0.9996\n",
      "Epoch 324/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3255 - sel%: 0.0473 - acc: 0.9765 - auc: 0.9995\n",
      "Epoch 325/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3261 - sel%: 0.0471 - acc: 0.9758 - auc: 0.9995\n",
      "Epoch 326/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3245 - sel%: 0.0469 - acc: 0.9756 - auc: 0.9996\n",
      "Epoch 327/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3246 - sel%: 0.0469 - acc: 0.9748 - auc: 0.9996\n",
      "Epoch 328/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3252 - sel%: 0.0464 - acc: 0.9746 - auc: 0.9995\n",
      "Epoch 329/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3240 - sel%: 0.0469 - acc: 0.9760 - auc: 0.9996\n",
      "Epoch 330/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3225 - sel%: 0.0469 - acc: 0.9769 - auc: 0.9995\n",
      "Epoch 331/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3229 - sel%: 0.0465 - acc: 0.9761 - auc: 0.9995\n",
      "Epoch 332/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3243 - sel%: 0.0467 - acc: 0.9763 - auc: 0.9995\n",
      "Epoch 333/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3229 - sel%: 0.0471 - acc: 0.9765 - auc: 0.9996\n",
      "Epoch 334/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3228 - sel%: 0.0469 - acc: 0.9765 - auc: 0.9995\n",
      "Epoch 335/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3216 - sel%: 0.0465 - acc: 0.9765 - auc: 0.9996\n",
      "Epoch 336/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3213 - sel%: 0.0464 - acc: 0.9765 - auc: 0.9995\n",
      "Epoch 337/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3215 - sel%: 0.0464 - acc: 0.9764 - auc: 0.9995\n",
      "Epoch 338/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3220 - sel%: 0.0466 - acc: 0.9759 - auc: 0.9996\n",
      "Epoch 339/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3230 - sel%: 0.0462 - acc: 0.9753 - auc: 0.9995\n",
      "Epoch 340/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3212 - sel%: 0.0463 - acc: 0.9761 - auc: 0.9996\n",
      "Epoch 341/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3216 - sel%: 0.0461 - acc: 0.9757 - auc: 0.9995\n",
      "Epoch 342/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3205 - sel%: 0.0462 - acc: 0.9755 - auc: 0.9996\n",
      "Epoch 343/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3197 - sel%: 0.0462 - acc: 0.9767 - auc: 0.9996\n",
      "Epoch 344/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3209 - sel%: 0.0462 - acc: 0.9763 - auc: 0.9995\n",
      "Epoch 345/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3203 - sel%: 0.0460 - acc: 0.9753 - auc: 0.9996\n",
      "Epoch 346/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3195 - sel%: 0.0459 - acc: 0.9769 - auc: 0.9995\n",
      "Epoch 347/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3191 - sel%: 0.0459 - acc: 0.9770 - auc: 0.9995\n",
      "Epoch 348/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3198 - sel%: 0.0460 - acc: 0.9765 - auc: 0.9995\n",
      "Epoch 349/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3193 - sel%: 0.0463 - acc: 0.9765 - auc: 0.9995\n",
      "Epoch 350/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3187 - sel%: 0.0463 - acc: 0.9765 - auc: 0.9996\n",
      "Epoch 351/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3183 - sel%: 0.0459 - acc: 0.9763 - auc: 0.9995\n",
      "Epoch 352/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3184 - sel%: 0.0460 - acc: 0.9765 - auc: 0.9996\n",
      "Epoch 353/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3178 - sel%: 0.0459 - acc: 0.9765 - auc: 0.9996\n",
      "Epoch 354/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3178 - sel%: 0.0455 - acc: 0.9758 - auc: 0.9995\n",
      "Epoch 355/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3190 - sel%: 0.0458 - acc: 0.9763 - auc: 0.9995\n",
      "Epoch 356/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3181 - sel%: 0.0461 - acc: 0.9774 - auc: 0.9996\n",
      "Epoch 357/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3172 - sel%: 0.0454 - acc: 0.9761 - auc: 0.9996\n",
      "Epoch 358/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3170 - sel%: 0.0455 - acc: 0.9763 - auc: 0.9996\n",
      "Epoch 359/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3166 - sel%: 0.0456 - acc: 0.9763 - auc: 0.9995\n",
      "Epoch 360/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3170 - sel%: 0.0455 - acc: 0.9758 - auc: 0.9996\n",
      "Epoch 361/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3167 - sel%: 0.0453 - acc: 0.9759 - auc: 0.9995\n",
      "Epoch 362/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3177 - sel%: 0.0452 - acc: 0.9756 - auc: 0.9995\n",
      "Epoch 363/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3165 - sel%: 0.0452 - acc: 0.9761 - auc: 0.9995\n",
      "Epoch 364/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3163 - sel%: 0.0457 - acc: 0.9772 - auc: 0.9996\n",
      "Epoch 365/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3165 - sel%: 0.0455 - acc: 0.9757 - auc: 0.9995\n",
      "Epoch 366/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3149 - sel%: 0.0457 - acc: 0.9771 - auc: 0.9996\n",
      "Epoch 367/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3158 - sel%: 0.0452 - acc: 0.9767 - auc: 0.9996\n",
      "Epoch 368/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3149 - sel%: 0.0455 - acc: 0.9774 - auc: 0.9996\n",
      "Epoch 369/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3142 - sel%: 0.0456 - acc: 0.9769 - auc: 0.9996\n",
      "Epoch 370/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3143 - sel%: 0.0453 - acc: 0.9770 - auc: 0.9996\n",
      "Epoch 371/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3148 - sel%: 0.0453 - acc: 0.9768 - auc: 0.9996\n",
      "Epoch 372/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3148 - sel%: 0.0452 - acc: 0.9766 - auc: 0.9996\n",
      "Epoch 373/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3136 - sel%: 0.0456 - acc: 0.9779 - auc: 0.9996\n",
      "Epoch 374/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3140 - sel%: 0.0452 - acc: 0.9767 - auc: 0.9996\n",
      "Epoch 375/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3135 - sel%: 0.0452 - acc: 0.9767 - auc: 0.9996\n",
      "Epoch 376/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3144 - sel%: 0.0453 - acc: 0.9769 - auc: 0.9995\n",
      "Epoch 377/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3135 - sel%: 0.0454 - acc: 0.9770 - auc: 0.9995\n",
      "Epoch 378/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3132 - sel%: 0.0453 - acc: 0.9772 - auc: 0.9996\n",
      "Epoch 379/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3129 - sel%: 0.0451 - acc: 0.9770 - auc: 0.9996\n",
      "Epoch 380/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3125 - sel%: 0.0450 - acc: 0.9772 - auc: 0.9996\n",
      "Epoch 381/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3122 - sel%: 0.0449 - acc: 0.9771 - auc: 0.9996\n",
      "Epoch 382/500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3129 - sel%: 0.0454 - acc: 0.9776 - auc: 0.9996\n",
      "Epoch 383/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3117 - sel%: 0.0452 - acc: 0.9772 - auc: 0.9996\n",
      "Epoch 384/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3120 - sel%: 0.0446 - acc: 0.9768 - auc: 0.9996\n",
      "Epoch 385/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3126 - sel%: 0.0448 - acc: 0.9769 - auc: 0.9996\n",
      "Epoch 386/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3110 - sel%: 0.0448 - acc: 0.9776 - auc: 0.9996\n",
      "Epoch 387/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3107 - sel%: 0.0450 - acc: 0.9776 - auc: 0.9996\n",
      "Epoch 388/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3118 - sel%: 0.0449 - acc: 0.9775 - auc: 0.9995\n",
      "Epoch 389/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3114 - sel%: 0.0452 - acc: 0.9778 - auc: 0.9996\n",
      "Epoch 390/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3107 - sel%: 0.0449 - acc: 0.9773 - auc: 0.9996\n",
      "Epoch 391/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3107 - sel%: 0.0452 - acc: 0.9779 - auc: 0.9996\n",
      "Epoch 392/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3109 - sel%: 0.0450 - acc: 0.9777 - auc: 0.9996\n",
      "Epoch 393/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3110 - sel%: 0.0447 - acc: 0.9775 - auc: 0.9996\n",
      "Epoch 394/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3098 - sel%: 0.0451 - acc: 0.9783 - auc: 0.9996\n",
      "Epoch 395/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3098 - sel%: 0.0447 - acc: 0.9773 - auc: 0.9996\n",
      "Epoch 396/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3094 - sel%: 0.0444 - acc: 0.9775 - auc: 0.9996\n",
      "Epoch 397/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3086 - sel%: 0.0446 - acc: 0.9780 - auc: 0.9996\n",
      "Epoch 398/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3098 - sel%: 0.0444 - acc: 0.9774 - auc: 0.9995\n",
      "Epoch 399/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3087 - sel%: 0.0443 - acc: 0.9774 - auc: 0.9996\n",
      "Epoch 400/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3086 - sel%: 0.0444 - acc: 0.9775 - auc: 0.9996\n",
      "Epoch 401/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3091 - sel%: 0.0447 - acc: 0.9775 - auc: 0.9996\n",
      "Epoch 402/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3086 - sel%: 0.0443 - acc: 0.9772 - auc: 0.9996\n",
      "Epoch 403/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3082 - sel%: 0.0441 - acc: 0.9772 - auc: 0.9996\n",
      "Epoch 404/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3087 - sel%: 0.0445 - acc: 0.9779 - auc: 0.9996\n",
      "Epoch 405/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3092 - sel%: 0.0445 - acc: 0.9772 - auc: 0.9996\n",
      "Epoch 406/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3070 - sel%: 0.0442 - acc: 0.9774 - auc: 0.9996\n",
      "Epoch 407/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3077 - sel%: 0.0445 - acc: 0.9777 - auc: 0.9996\n",
      "Epoch 408/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3072 - sel%: 0.0440 - acc: 0.9774 - auc: 0.9996\n",
      "Epoch 409/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3068 - sel%: 0.0442 - acc: 0.9773 - auc: 0.9996\n",
      "Epoch 410/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3059 - sel%: 0.0442 - acc: 0.9785 - auc: 0.9997\n",
      "Epoch 411/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3051 - sel%: 0.0440 - acc: 0.9776 - auc: 0.9996\n",
      "Epoch 412/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3042 - sel%: 0.0438 - acc: 0.9779 - auc: 0.9996\n",
      "Epoch 413/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3058 - sel%: 0.0437 - acc: 0.9772 - auc: 0.9996\n",
      "Epoch 414/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3054 - sel%: 0.0437 - acc: 0.9777 - auc: 0.9996\n",
      "Epoch 415/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3053 - sel%: 0.0434 - acc: 0.9775 - auc: 0.9995\n",
      "Epoch 416/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3044 - sel%: 0.0435 - acc: 0.9769 - auc: 0.9996\n",
      "Epoch 417/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3048 - sel%: 0.0438 - acc: 0.9776 - auc: 0.9996\n",
      "Epoch 418/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3048 - sel%: 0.0437 - acc: 0.9778 - auc: 0.9996\n",
      "Epoch 419/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3037 - sel%: 0.0435 - acc: 0.9772 - auc: 0.9996\n",
      "Epoch 420/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3032 - sel%: 0.0432 - acc: 0.9781 - auc: 0.9995\n",
      "Epoch 421/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3028 - sel%: 0.0432 - acc: 0.9776 - auc: 0.9996\n",
      "Epoch 422/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3029 - sel%: 0.0434 - acc: 0.9775 - auc: 0.9996\n",
      "Epoch 423/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3027 - sel%: 0.0429 - acc: 0.9769 - auc: 0.9996\n",
      "Epoch 424/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3026 - sel%: 0.0427 - acc: 0.9773 - auc: 0.9996\n",
      "Epoch 425/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3023 - sel%: 0.0431 - acc: 0.9778 - auc: 0.9996\n",
      "Epoch 426/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3027 - sel%: 0.0431 - acc: 0.9778 - auc: 0.9996\n",
      "Epoch 427/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3011 - sel%: 0.0432 - acc: 0.9785 - auc: 0.9996\n",
      "Epoch 428/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3016 - sel%: 0.0430 - acc: 0.9780 - auc: 0.9996\n",
      "Epoch 429/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3015 - sel%: 0.0433 - acc: 0.9776 - auc: 0.9996\n",
      "Epoch 430/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3017 - sel%: 0.0432 - acc: 0.9780 - auc: 0.9996\n",
      "Epoch 431/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3012 - sel%: 0.0428 - acc: 0.9776 - auc: 0.9996\n",
      "Epoch 432/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3022 - sel%: 0.0424 - acc: 0.9766 - auc: 0.9996\n",
      "Epoch 433/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.3008 - sel%: 0.0429 - acc: 0.9775 - auc: 0.9995\n",
      "Epoch 434/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2999 - sel%: 0.0427 - acc: 0.9778 - auc: 0.9996\n",
      "Epoch 435/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2994 - sel%: 0.0429 - acc: 0.9784 - auc: 0.9996\n",
      "Epoch 436/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2996 - sel%: 0.0425 - acc: 0.9772 - auc: 0.9996\n",
      "Epoch 437/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2994 - sel%: 0.0430 - acc: 0.9786 - auc: 0.9996\n",
      "Epoch 438/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2992 - sel%: 0.0426 - acc: 0.9778 - auc: 0.9996\n",
      "Epoch 439/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2987 - sel%: 0.0427 - acc: 0.9778 - auc: 0.9996\n",
      "Epoch 440/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2980 - sel%: 0.0427 - acc: 0.9782 - auc: 0.9996\n",
      "Epoch 441/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2991 - sel%: 0.0428 - acc: 0.9781 - auc: 0.9996\n",
      "Epoch 442/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2976 - sel%: 0.0427 - acc: 0.9784 - auc: 0.9996\n",
      "Epoch 443/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2970 - sel%: 0.0425 - acc: 0.9783 - auc: 0.9996\n",
      "Epoch 444/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2985 - sel%: 0.0420 - acc: 0.9774 - auc: 0.9996\n",
      "Epoch 445/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2970 - sel%: 0.0422 - acc: 0.9782 - auc: 0.9996\n",
      "Epoch 446/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2966 - sel%: 0.0421 - acc: 0.9777 - auc: 0.9996\n",
      "Epoch 447/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2979 - sel%: 0.0421 - acc: 0.9773 - auc: 0.9996\n",
      "Epoch 448/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2977 - sel%: 0.0423 - acc: 0.9774 - auc: 0.9996\n",
      "Epoch 449/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2981 - sel%: 0.0427 - acc: 0.9781 - auc: 0.9996\n",
      "Epoch 450/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2958 - sel%: 0.0422 - acc: 0.9784 - auc: 0.9997\n",
      "Epoch 451/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2960 - sel%: 0.0422 - acc: 0.9779 - auc: 0.9996\n",
      "Epoch 452/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2961 - sel%: 0.0423 - acc: 0.9783 - auc: 0.9996\n",
      "Epoch 453/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2956 - sel%: 0.0418 - acc: 0.9779 - auc: 0.9996\n",
      "Epoch 454/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2949 - sel%: 0.0420 - acc: 0.9781 - auc: 0.9996\n",
      "Epoch 455/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2943 - sel%: 0.0421 - acc: 0.9780 - auc: 0.9997\n",
      "Epoch 456/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2951 - sel%: 0.0420 - acc: 0.9776 - auc: 0.9996\n",
      "Epoch 457/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2946 - sel%: 0.0420 - acc: 0.9783 - auc: 0.9996\n",
      "Epoch 458/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2946 - sel%: 0.0416 - acc: 0.9773 - auc: 0.9997\n",
      "Epoch 459/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2944 - sel%: 0.0419 - acc: 0.9778 - auc: 0.9996\n",
      "Epoch 460/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2936 - sel%: 0.0416 - acc: 0.9781 - auc: 0.9996\n",
      "Epoch 461/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2935 - sel%: 0.0415 - acc: 0.9775 - auc: 0.9996\n",
      "Epoch 462/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2943 - sel%: 0.0415 - acc: 0.9781 - auc: 0.9996\n",
      "Epoch 463/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2940 - sel%: 0.0417 - acc: 0.9780 - auc: 0.9996\n",
      "Epoch 464/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2926 - sel%: 0.0416 - acc: 0.9780 - auc: 0.9996\n",
      "Epoch 465/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2915 - sel%: 0.0416 - acc: 0.9782 - auc: 0.9996\n",
      "Epoch 466/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2918 - sel%: 0.0418 - acc: 0.9783 - auc: 0.9996\n",
      "Epoch 467/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2926 - sel%: 0.0416 - acc: 0.9785 - auc: 0.9996\n",
      "Epoch 468/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2913 - sel%: 0.0418 - acc: 0.9790 - auc: 0.9996\n",
      "Epoch 469/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2924 - sel%: 0.0415 - acc: 0.9786 - auc: 0.9996\n",
      "Epoch 470/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2915 - sel%: 0.0412 - acc: 0.9777 - auc: 0.9997\n",
      "Epoch 471/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2918 - sel%: 0.0412 - acc: 0.9779 - auc: 0.9996\n",
      "Epoch 472/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2905 - sel%: 0.0412 - acc: 0.9789 - auc: 0.9996\n",
      "Epoch 473/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2911 - sel%: 0.0411 - acc: 0.9780 - auc: 0.9996\n",
      "Epoch 474/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2895 - sel%: 0.0409 - acc: 0.9783 - auc: 0.9996\n",
      "Epoch 475/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2900 - sel%: 0.0405 - acc: 0.9773 - auc: 0.9996\n",
      "Epoch 476/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2895 - sel%: 0.0410 - acc: 0.9782 - auc: 0.9996\n",
      "Epoch 477/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2895 - sel%: 0.0410 - acc: 0.9780 - auc: 0.9996\n",
      "Epoch 478/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2902 - sel%: 0.0408 - acc: 0.9778 - auc: 0.9996\n",
      "Epoch 479/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2903 - sel%: 0.0410 - acc: 0.9781 - auc: 0.9996\n",
      "Epoch 480/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2904 - sel%: 0.0412 - acc: 0.9782 - auc: 0.9996\n",
      "Epoch 481/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2882 - sel%: 0.0409 - acc: 0.9779 - auc: 0.9997\n",
      "Epoch 482/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2880 - sel%: 0.0407 - acc: 0.9781 - auc: 0.9997\n",
      "Epoch 483/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2890 - sel%: 0.0405 - acc: 0.9779 - auc: 0.9996\n",
      "Epoch 484/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2889 - sel%: 0.0407 - acc: 0.9781 - auc: 0.9996\n",
      "Epoch 485/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2893 - sel%: 0.0410 - acc: 0.9785 - auc: 0.9996\n",
      "Epoch 486/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2881 - sel%: 0.0409 - acc: 0.9785 - auc: 0.9996\n",
      "Epoch 487/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2874 - sel%: 0.0403 - acc: 0.9778 - auc: 0.9996\n",
      "Epoch 488/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2883 - sel%: 0.0409 - acc: 0.9784 - auc: 0.9996\n",
      "Epoch 489/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2873 - sel%: 0.0410 - acc: 0.9793 - auc: 0.9997\n",
      "Epoch 490/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2872 - sel%: 0.0409 - acc: 0.9791 - auc: 0.9996\n",
      "Epoch 491/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2877 - sel%: 0.0409 - acc: 0.9788 - auc: 0.9996\n",
      "Epoch 492/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2874 - sel%: 0.0410 - acc: 0.9793 - auc: 0.9996\n",
      "Epoch 493/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2864 - sel%: 0.0408 - acc: 0.9788 - auc: 0.9997\n",
      "Epoch 494/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2881 - sel%: 0.0404 - acc: 0.9779 - auc: 0.9996\n",
      "Epoch 495/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2868 - sel%: 0.0406 - acc: 0.9789 - auc: 0.9996\n",
      "Epoch 496/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2878 - sel%: 0.0406 - acc: 0.9777 - auc: 0.9996\n",
      "Epoch 497/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2868 - sel%: 0.0408 - acc: 0.9786 - auc: 0.9996\n",
      "Epoch 498/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2857 - sel%: 0.0405 - acc: 0.9789 - auc: 0.9996\n",
      "Epoch 499/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2854 - sel%: 0.0403 - acc: 0.9791 - auc: 0.9996\n",
      "Epoch 500/500\n",
      "60/60 [==============================] - 1s 14ms/step - loss: -0.2855 - sel%: 0.0404 - acc: 0.9785 - auc: 0.9996\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<tensorflow.python.keras.callbacks.History at 0x7f7fe0623f10>"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Train\n",
    "realx.selector.compile(loss=None,\n",
    "                       optimizer=optimizer,\n",
    "                       metrics=metrics)\n",
    "realx.selector.fit(x_train,\n",
    "                   y_train,\n",
    "                   epochs=epochs,\n",
    "                   batch_size=batch_size,\n",
    "                   verbose=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Plot Explainations"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-03-01T22:12:09.875386Z",
     "start_time": "2021-03-01T22:12:09.566675Z"
    }
   },
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-03-01T22:12:18.428772Z",
     "start_time": "2021-03-01T22:12:16.591334Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x1440 with 20 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#1. Get Selections \n",
    "score = realx.select(x_test, batch_size, True)\n",
    "\n",
    "#2. Get Predictions\n",
    "y_score = realx.predict(x_test, batch_size)\n",
    "y_pred = y_score.argmax(1)\n",
    "y_test_num = y_test.argmax(1)\n",
    "\n",
    "#2. Plot Selections\n",
    "fig = plt.figure(figsize=(15,20))    \n",
    "j = 1\n",
    "for num in range(10):\n",
    "    for i in np.where(y_test_num==num)[0][:2]:\n",
    "        ax = fig.add_subplot(5,4, j)\n",
    "        ax.imshow(score[i,:].reshape((28,28)), cmap= 'Reds', alpha = 0.8)\n",
    "        ax.imshow(x_test[i,:].reshape((28,28)), alpha = 0.1, cmap = 'Greys')\n",
    "        ax.set_axis_off()\n",
    "        j+=1\n",
    "\n",
    "fig.subplots_adjust(wspace=0, hspace=0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Train EVAL-X"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-03-01T22:17:37.009176Z",
     "start_time": "2021-03-01T22:12:58.376018Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 1.4442 - acc: 0.6928 - auc: 0.9445\n",
      "Epoch 2/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.8886 - acc: 0.8584 - auc: 0.9860\n",
      "Epoch 3/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.7833 - acc: 0.8830 - auc: 0.9900\n",
      "Epoch 4/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.7246 - acc: 0.8972 - auc: 0.9917\n",
      "Epoch 5/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.6782 - acc: 0.9057 - auc: 0.9929\n",
      "Epoch 6/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.6427 - acc: 0.9137 - auc: 0.9938\n",
      "Epoch 7/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.6186 - acc: 0.9160 - auc: 0.9943\n",
      "Epoch 8/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.5904 - acc: 0.9226 - auc: 0.9949\n",
      "Epoch 9/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.5683 - acc: 0.9272 - auc: 0.9954\n",
      "Epoch 10/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.5476 - acc: 0.9298 - auc: 0.9958\n",
      "Epoch 11/500\n",
      "60/60 [==============================] - 1s 15ms/step - loss: 0.5301 - acc: 0.9323 - auc: 0.9961\n",
      "Epoch 12/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.5179 - acc: 0.9327 - auc: 0.9962\n",
      "Epoch 13/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.4993 - acc: 0.9371 - auc: 0.9965\n",
      "Epoch 14/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.4863 - acc: 0.9384 - auc: 0.9968\n",
      "Epoch 15/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.4718 - acc: 0.9412 - auc: 0.9969\n",
      "Epoch 16/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.4635 - acc: 0.9417 - auc: 0.9970\n",
      "Epoch 17/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.4508 - acc: 0.9431 - auc: 0.9973\n",
      "Epoch 18/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.4401 - acc: 0.9456 - auc: 0.9973\n",
      "Epoch 19/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.4299 - acc: 0.9467 - auc: 0.9975\n",
      "Epoch 20/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.4229 - acc: 0.9476 - auc: 0.9974\n",
      "Epoch 21/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.4142 - acc: 0.9479 - auc: 0.9975\n",
      "Epoch 22/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.4049 - acc: 0.9492 - auc: 0.9976\n",
      "Epoch 23/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.3958 - acc: 0.9503 - auc: 0.9978\n",
      "Epoch 24/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.3891 - acc: 0.9514 - auc: 0.9978\n",
      "Epoch 25/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.3812 - acc: 0.9524 - auc: 0.9979\n",
      "Epoch 26/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.3723 - acc: 0.9536 - auc: 0.9980\n",
      "Epoch 27/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.3651 - acc: 0.9544 - auc: 0.9980\n",
      "Epoch 28/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.3608 - acc: 0.9546 - auc: 0.9981\n",
      "Epoch 29/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.3536 - acc: 0.9540 - auc: 0.9981\n",
      "Epoch 30/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.3442 - acc: 0.9572 - auc: 0.9982\n",
      "Epoch 31/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.3395 - acc: 0.9570 - auc: 0.9981\n",
      "Epoch 32/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.3326 - acc: 0.9588 - auc: 0.9983\n",
      "Epoch 33/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.3297 - acc: 0.9577 - auc: 0.9983\n",
      "Epoch 34/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.3206 - acc: 0.9599 - auc: 0.9984\n",
      "Epoch 35/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.3170 - acc: 0.9598 - auc: 0.9985\n",
      "Epoch 36/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.3121 - acc: 0.9605 - auc: 0.9983\n",
      "Epoch 37/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.3058 - acc: 0.9608 - auc: 0.9985\n",
      "Epoch 38/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.3040 - acc: 0.9603 - auc: 0.9985\n",
      "Epoch 39/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.2968 - acc: 0.9617 - auc: 0.9987\n",
      "Epoch 40/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.2962 - acc: 0.9608 - auc: 0.9985\n",
      "Epoch 41/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.2862 - acc: 0.9636 - auc: 0.9987\n",
      "Epoch 42/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.2847 - acc: 0.9632 - auc: 0.9987\n",
      "Epoch 43/500\n",
      "60/60 [==============================] - 1s 11ms/step - loss: 0.2773 - acc: 0.9649 - auc: 0.9988\n",
      "Epoch 44/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.2750 - acc: 0.9644 - auc: 0.9988\n",
      "Epoch 45/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.2722 - acc: 0.9643 - auc: 0.9988\n",
      "Epoch 46/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.2686 - acc: 0.9643 - auc: 0.9988\n",
      "Epoch 47/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.2655 - acc: 0.9652 - auc: 0.9988\n",
      "Epoch 48/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.2606 - acc: 0.9656 - auc: 0.9988\n",
      "Epoch 49/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.2561 - acc: 0.9660 - auc: 0.9990\n",
      "Epoch 50/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.2534 - acc: 0.9667 - auc: 0.9989\n",
      "Epoch 51/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.2513 - acc: 0.9665 - auc: 0.9989\n",
      "Epoch 52/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.2475 - acc: 0.9674 - auc: 0.9989\n",
      "Epoch 53/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.2447 - acc: 0.9679 - auc: 0.9989\n",
      "Epoch 54/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.2411 - acc: 0.9680 - auc: 0.9989\n",
      "Epoch 55/500\n",
      "60/60 [==============================] - 1s 12ms/step - loss: 0.2366 - acc: 0.9679 - auc: 0.9991\n",
      "Epoch 56/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.2357 - acc: 0.9683 - auc: 0.9989\n",
      "Epoch 57/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.2324 - acc: 0.9684 - auc: 0.9990\n",
      "Epoch 58/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.2302 - acc: 0.9682 - auc: 0.9990\n",
      "Epoch 59/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.2259 - acc: 0.9690 - auc: 0.9991\n",
      "Epoch 60/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.2236 - acc: 0.9696 - auc: 0.9991\n",
      "Epoch 61/500\n",
      "60/60 [==============================] - 1s 12ms/step - loss: 0.2207 - acc: 0.9694 - auc: 0.9991\n",
      "Epoch 62/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.2191 - acc: 0.9706 - auc: 0.9991\n",
      "Epoch 63/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.2158 - acc: 0.9701 - auc: 0.9991\n",
      "Epoch 64/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.2126 - acc: 0.9721 - auc: 0.9991\n",
      "Epoch 65/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.2124 - acc: 0.9698 - auc: 0.9991\n",
      "Epoch 66/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.2105 - acc: 0.9700 - auc: 0.9992\n",
      "Epoch 67/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.2065 - acc: 0.9716 - auc: 0.9991\n",
      "Epoch 68/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.2039 - acc: 0.9708 - auc: 0.9993\n",
      "Epoch 69/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.2029 - acc: 0.9721 - auc: 0.9991\n",
      "Epoch 70/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.1985 - acc: 0.9725 - auc: 0.9992\n",
      "Epoch 71/500\n",
      "60/60 [==============================] - 1s 13ms/step - loss: 0.1972 - acc: 0.9726 - auc: 0.9993\n",
      "Epoch 72/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1944 - acc: 0.9723 - auc: 0.9994\n",
      "Epoch 73/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.1922 - acc: 0.9731 - auc: 0.9992\n",
      "Epoch 74/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1916 - acc: 0.9731 - auc: 0.9993\n",
      "Epoch 75/500\n",
      "60/60 [==============================] - 1s 11ms/step - loss: 0.1901 - acc: 0.9734 - auc: 0.9993\n",
      "Epoch 76/500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1867 - acc: 0.9740 - auc: 0.9993\n",
      "Epoch 77/500\n",
      "60/60 [==============================] - 1s 11ms/step - loss: 0.1836 - acc: 0.9747 - auc: 0.9993\n",
      "Epoch 78/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.1853 - acc: 0.9729 - auc: 0.9993\n",
      "Epoch 79/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1805 - acc: 0.9748 - auc: 0.9994\n",
      "Epoch 80/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1801 - acc: 0.9745 - auc: 0.9994\n",
      "Epoch 81/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.1766 - acc: 0.9758 - auc: 0.9993\n",
      "Epoch 82/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.1784 - acc: 0.9737 - auc: 0.9993\n",
      "Epoch 83/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.1743 - acc: 0.9754 - auc: 0.9994\n",
      "Epoch 84/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1733 - acc: 0.9750 - auc: 0.9994\n",
      "Epoch 85/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1709 - acc: 0.9755 - auc: 0.9994\n",
      "Epoch 86/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1693 - acc: 0.9754 - auc: 0.9994\n",
      "Epoch 87/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.1683 - acc: 0.9753 - auc: 0.9994\n",
      "Epoch 88/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1672 - acc: 0.9762 - auc: 0.9994\n",
      "Epoch 89/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1653 - acc: 0.9760 - auc: 0.9995\n",
      "Epoch 90/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1646 - acc: 0.9753 - auc: 0.9995\n",
      "Epoch 91/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1631 - acc: 0.9763 - auc: 0.9994\n",
      "Epoch 92/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.1591 - acc: 0.9772 - auc: 0.9995\n",
      "Epoch 93/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1591 - acc: 0.9768 - auc: 0.9995\n",
      "Epoch 94/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1589 - acc: 0.9766 - auc: 0.9994\n",
      "Epoch 95/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1584 - acc: 0.9771 - auc: 0.9995\n",
      "Epoch 96/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.1557 - acc: 0.9770 - auc: 0.9994\n",
      "Epoch 97/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1545 - acc: 0.9771 - auc: 0.9994\n",
      "Epoch 98/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1516 - acc: 0.9781 - auc: 0.9995\n",
      "Epoch 99/500\n",
      "60/60 [==============================] - 1s 11ms/step - loss: 0.1506 - acc: 0.9783 - auc: 0.9996\n",
      "Epoch 100/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1521 - acc: 0.9774 - auc: 0.9995\n",
      "Epoch 101/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.1491 - acc: 0.9783 - auc: 0.9995\n",
      "Epoch 102/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1473 - acc: 0.9786 - auc: 0.9995\n",
      "Epoch 103/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1473 - acc: 0.9784 - auc: 0.9995\n",
      "Epoch 104/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1440 - acc: 0.9787 - auc: 0.9996\n",
      "Epoch 105/500\n",
      "60/60 [==============================] - 1s 11ms/step - loss: 0.1436 - acc: 0.9787 - auc: 0.9996\n",
      "Epoch 106/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1448 - acc: 0.9785 - auc: 0.9995\n",
      "Epoch 107/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1431 - acc: 0.9784 - auc: 0.9995\n",
      "Epoch 108/500\n",
      "60/60 [==============================] - 1s 11ms/step - loss: 0.1411 - acc: 0.9791 - auc: 0.9996\n",
      "Epoch 109/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.1421 - acc: 0.9783 - auc: 0.9995\n",
      "Epoch 110/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1394 - acc: 0.9792 - auc: 0.9996\n",
      "Epoch 111/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1379 - acc: 0.9796 - auc: 0.9996\n",
      "Epoch 112/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1369 - acc: 0.9792 - auc: 0.9996\n",
      "Epoch 113/500\n",
      "60/60 [==============================] - 1s 11ms/step - loss: 0.1364 - acc: 0.9794 - auc: 0.9995\n",
      "Epoch 114/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1330 - acc: 0.9801 - auc: 0.9996\n",
      "Epoch 115/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1347 - acc: 0.9796 - auc: 0.9996\n",
      "Epoch 116/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.1317 - acc: 0.9807 - auc: 0.9996\n",
      "Epoch 117/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.1319 - acc: 0.9805 - auc: 0.9996\n",
      "Epoch 118/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1321 - acc: 0.9801 - auc: 0.9995\n",
      "Epoch 119/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1316 - acc: 0.9801 - auc: 0.9996\n",
      "Epoch 120/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1295 - acc: 0.9801 - auc: 0.9997\n",
      "Epoch 121/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1276 - acc: 0.9803 - auc: 0.9997\n",
      "Epoch 122/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1292 - acc: 0.9799 - auc: 0.9996\n",
      "Epoch 123/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1270 - acc: 0.9811 - auc: 0.9996\n",
      "Epoch 124/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1257 - acc: 0.9807 - auc: 0.9996\n",
      "Epoch 125/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1238 - acc: 0.9817 - auc: 0.9996\n",
      "Epoch 126/500\n",
      "60/60 [==============================] - 1s 11ms/step - loss: 0.1243 - acc: 0.9807 - auc: 0.9997\n",
      "Epoch 127/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1236 - acc: 0.9815 - auc: 0.9996\n",
      "Epoch 128/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.1228 - acc: 0.9812 - auc: 0.9997\n",
      "Epoch 129/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1227 - acc: 0.9810 - auc: 0.9997\n",
      "Epoch 130/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1217 - acc: 0.9812 - auc: 0.9996\n",
      "Epoch 131/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1193 - acc: 0.9813 - auc: 0.9998\n",
      "Epoch 132/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1184 - acc: 0.9822 - auc: 0.9997\n",
      "Epoch 133/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1187 - acc: 0.9810 - auc: 0.9997\n",
      "Epoch 134/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1183 - acc: 0.9815 - auc: 0.9997\n",
      "Epoch 135/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.1163 - acc: 0.9820 - auc: 0.9997\n",
      "Epoch 136/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.1164 - acc: 0.9817 - auc: 0.9996\n",
      "Epoch 137/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.1163 - acc: 0.9814 - auc: 0.9996\n",
      "Epoch 138/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.1157 - acc: 0.9819 - auc: 0.9997\n",
      "Epoch 139/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1136 - acc: 0.9826 - auc: 0.9997\n",
      "Epoch 140/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.1135 - acc: 0.9821 - auc: 0.9997\n",
      "Epoch 141/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.1132 - acc: 0.9823 - auc: 0.9997\n",
      "Epoch 142/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1128 - acc: 0.9826 - auc: 0.9996\n",
      "Epoch 143/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1095 - acc: 0.9838 - auc: 0.9998\n",
      "Epoch 144/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1110 - acc: 0.9822 - auc: 0.9997\n",
      "Epoch 145/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.1100 - acc: 0.9831 - auc: 0.9997\n",
      "Epoch 146/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1088 - acc: 0.9835 - auc: 0.9997\n",
      "Epoch 147/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1075 - acc: 0.9833 - auc: 0.9997\n",
      "Epoch 148/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.1099 - acc: 0.9828 - auc: 0.9997\n",
      "Epoch 149/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.1070 - acc: 0.9836 - auc: 0.9997\n",
      "Epoch 150/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.1070 - acc: 0.9833 - auc: 0.9997\n",
      "Epoch 151/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1068 - acc: 0.9830 - auc: 0.9997\n",
      "Epoch 152/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1052 - acc: 0.9834 - auc: 0.9997\n",
      "Epoch 153/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.1058 - acc: 0.9833 - auc: 0.9997\n",
      "Epoch 154/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.1061 - acc: 0.9831 - auc: 0.9997\n",
      "Epoch 155/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.1018 - acc: 0.9842 - auc: 0.9998\n",
      "Epoch 156/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1029 - acc: 0.9837 - auc: 0.9997\n",
      "Epoch 157/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.1019 - acc: 0.9843 - auc: 0.9997\n",
      "Epoch 158/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.1024 - acc: 0.9836 - auc: 0.9998\n",
      "Epoch 159/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.1039 - acc: 0.9835 - auc: 0.9997\n",
      "Epoch 160/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0998 - acc: 0.9849 - auc: 0.9997\n",
      "Epoch 161/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0999 - acc: 0.9847 - auc: 0.9997\n",
      "Epoch 162/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0994 - acc: 0.9848 - auc: 0.9997\n",
      "Epoch 163/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0992 - acc: 0.9843 - auc: 0.9998\n",
      "Epoch 164/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0996 - acc: 0.9836 - auc: 0.9998\n",
      "Epoch 165/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0975 - acc: 0.9845 - auc: 0.9998\n",
      "Epoch 166/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0955 - acc: 0.9855 - auc: 0.9997\n",
      "Epoch 167/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0971 - acc: 0.9848 - auc: 0.9998\n",
      "Epoch 168/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0972 - acc: 0.9851 - auc: 0.9998\n",
      "Epoch 169/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0974 - acc: 0.9847 - auc: 0.9997\n",
      "Epoch 170/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0963 - acc: 0.9848 - auc: 0.9998\n",
      "Epoch 171/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0963 - acc: 0.9847 - auc: 0.9997\n",
      "Epoch 172/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0953 - acc: 0.9847 - auc: 0.9997\n",
      "Epoch 173/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0931 - acc: 0.9856 - auc: 0.9998\n",
      "Epoch 174/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0938 - acc: 0.9852 - auc: 0.9998\n",
      "Epoch 175/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0944 - acc: 0.9845 - auc: 0.9997\n",
      "Epoch 176/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0938 - acc: 0.9851 - auc: 0.9997\n",
      "Epoch 177/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0936 - acc: 0.9854 - auc: 0.9997\n",
      "Epoch 178/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0922 - acc: 0.9849 - auc: 0.9998\n",
      "Epoch 179/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0919 - acc: 0.9852 - auc: 0.9998\n",
      "Epoch 180/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0931 - acc: 0.9845 - auc: 0.9998\n",
      "Epoch 181/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0919 - acc: 0.9854 - auc: 0.9997\n",
      "Epoch 182/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0912 - acc: 0.9852 - auc: 0.9997\n",
      "Epoch 183/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0913 - acc: 0.9851 - auc: 0.9997\n",
      "Epoch 184/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0904 - acc: 0.9850 - auc: 0.9998\n",
      "Epoch 185/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0886 - acc: 0.9862 - auc: 0.9998\n",
      "Epoch 186/500\n",
      "60/60 [==============================] - 1s 11ms/step - loss: 0.0908 - acc: 0.9853 - auc: 0.9998\n",
      "Epoch 187/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0879 - acc: 0.9861 - auc: 0.9998\n",
      "Epoch 188/500\n",
      "60/60 [==============================] - 1s 11ms/step - loss: 0.0873 - acc: 0.9863 - auc: 0.9998\n",
      "Epoch 189/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0875 - acc: 0.9859 - auc: 0.9998\n",
      "Epoch 190/500\n",
      "60/60 [==============================] - 1s 11ms/step - loss: 0.0890 - acc: 0.9855 - auc: 0.9998\n",
      "Epoch 191/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0873 - acc: 0.9856 - auc: 0.9998\n",
      "Epoch 192/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0887 - acc: 0.9855 - auc: 0.9998\n",
      "Epoch 193/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0862 - acc: 0.9861 - auc: 0.9998\n",
      "Epoch 194/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0869 - acc: 0.9858 - auc: 0.9998\n",
      "Epoch 195/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0863 - acc: 0.9861 - auc: 0.9998\n",
      "Epoch 196/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0839 - acc: 0.9864 - auc: 0.9999\n",
      "Epoch 197/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0849 - acc: 0.9864 - auc: 0.9998\n",
      "Epoch 198/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0842 - acc: 0.9864 - auc: 0.9998\n",
      "Epoch 199/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0851 - acc: 0.9863 - auc: 0.9998\n",
      "Epoch 200/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0841 - acc: 0.9863 - auc: 0.9998\n",
      "Epoch 201/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0838 - acc: 0.9863 - auc: 0.9998\n",
      "Epoch 202/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0830 - acc: 0.9872 - auc: 0.9998\n",
      "Epoch 203/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0822 - acc: 0.9868 - auc: 0.9998\n",
      "Epoch 204/500\n",
      "60/60 [==============================] - 1s 12ms/step - loss: 0.0824 - acc: 0.9864 - auc: 0.9998\n",
      "Epoch 205/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0824 - acc: 0.9869 - auc: 0.9998\n",
      "Epoch 206/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0829 - acc: 0.9865 - auc: 0.9998\n",
      "Epoch 207/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0820 - acc: 0.9868 - auc: 0.9998\n",
      "Epoch 208/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0823 - acc: 0.9865 - auc: 0.9998\n",
      "Epoch 209/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0820 - acc: 0.9865 - auc: 0.9998\n",
      "Epoch 210/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0809 - acc: 0.9868 - auc: 0.9998\n",
      "Epoch 211/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0802 - acc: 0.9872 - auc: 0.9998\n",
      "Epoch 212/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0805 - acc: 0.9868 - auc: 0.9998\n",
      "Epoch 213/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0808 - acc: 0.9867 - auc: 0.9998\n",
      "Epoch 214/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0782 - acc: 0.9875 - auc: 0.9998\n",
      "Epoch 215/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0803 - acc: 0.9865 - auc: 0.9998\n",
      "Epoch 216/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0791 - acc: 0.9871 - auc: 0.9998\n",
      "Epoch 217/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0788 - acc: 0.9875 - auc: 0.9998\n",
      "Epoch 218/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0791 - acc: 0.9873 - auc: 0.9998\n",
      "Epoch 219/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0789 - acc: 0.9872 - auc: 0.9998\n",
      "Epoch 220/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0773 - acc: 0.9873 - auc: 0.9998\n",
      "Epoch 221/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0784 - acc: 0.9869 - auc: 0.9998\n",
      "Epoch 222/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0774 - acc: 0.9874 - auc: 0.9998\n",
      "Epoch 223/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0775 - acc: 0.9873 - auc: 0.9998\n",
      "Epoch 224/500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0787 - acc: 0.9867 - auc: 0.9998\n",
      "Epoch 225/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0765 - acc: 0.9879 - auc: 0.9999\n",
      "Epoch 226/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0764 - acc: 0.9872 - auc: 0.9998\n",
      "Epoch 227/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0767 - acc: 0.9875 - auc: 0.9998\n",
      "Epoch 228/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0751 - acc: 0.9877 - auc: 0.9998\n",
      "Epoch 229/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0748 - acc: 0.9874 - auc: 0.9998\n",
      "Epoch 230/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0761 - acc: 0.9873 - auc: 0.9998\n",
      "Epoch 231/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0735 - acc: 0.9886 - auc: 0.9999\n",
      "Epoch 232/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0727 - acc: 0.9886 - auc: 0.9999\n",
      "Epoch 233/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0767 - acc: 0.9872 - auc: 0.9998\n",
      "Epoch 234/500\n",
      "60/60 [==============================] - 1s 12ms/step - loss: 0.0731 - acc: 0.9888 - auc: 0.9999\n",
      "Epoch 235/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0739 - acc: 0.9881 - auc: 0.9998\n",
      "Epoch 236/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0738 - acc: 0.9882 - auc: 0.9998\n",
      "Epoch 237/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0737 - acc: 0.9880 - auc: 0.9999\n",
      "Epoch 238/500\n",
      "60/60 [==============================] - 1s 11ms/step - loss: 0.0725 - acc: 0.9881 - auc: 0.9999\n",
      "Epoch 239/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0738 - acc: 0.9876 - auc: 0.9998\n",
      "Epoch 240/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0730 - acc: 0.9874 - auc: 0.9998\n",
      "Epoch 241/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0717 - acc: 0.9886 - auc: 0.9999\n",
      "Epoch 242/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0716 - acc: 0.9883 - auc: 0.9999\n",
      "Epoch 243/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0722 - acc: 0.9881 - auc: 0.9998\n",
      "Epoch 244/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0725 - acc: 0.9877 - auc: 0.9998\n",
      "Epoch 245/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0730 - acc: 0.9874 - auc: 0.9999\n",
      "Epoch 246/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0715 - acc: 0.9884 - auc: 0.9999\n",
      "Epoch 247/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0726 - acc: 0.9873 - auc: 0.9999\n",
      "Epoch 248/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0705 - acc: 0.9886 - auc: 0.9998\n",
      "Epoch 249/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0713 - acc: 0.9883 - auc: 0.9998\n",
      "Epoch 250/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0708 - acc: 0.9881 - auc: 0.9999\n",
      "Epoch 251/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0701 - acc: 0.9882 - auc: 0.9999\n",
      "Epoch 252/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0708 - acc: 0.9884 - auc: 0.9998\n",
      "Epoch 253/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0721 - acc: 0.9879 - auc: 0.9999\n",
      "Epoch 254/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0703 - acc: 0.9887 - auc: 0.9999\n",
      "Epoch 255/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0692 - acc: 0.9889 - auc: 0.9999\n",
      "Epoch 256/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0707 - acc: 0.9885 - auc: 0.9998\n",
      "Epoch 257/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0709 - acc: 0.9882 - auc: 0.9998\n",
      "Epoch 258/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0699 - acc: 0.9882 - auc: 0.9998\n",
      "Epoch 259/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0689 - acc: 0.9887 - auc: 0.9999\n",
      "Epoch 260/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0690 - acc: 0.9883 - auc: 0.9998\n",
      "Epoch 261/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0690 - acc: 0.9887 - auc: 0.9998\n",
      "Epoch 262/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0701 - acc: 0.9884 - auc: 0.9999\n",
      "Epoch 263/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0692 - acc: 0.9884 - auc: 0.9998\n",
      "Epoch 264/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0677 - acc: 0.9889 - auc: 0.9999\n",
      "Epoch 265/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0685 - acc: 0.9883 - auc: 0.9999\n",
      "Epoch 266/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0684 - acc: 0.9887 - auc: 0.9998\n",
      "Epoch 267/500\n",
      "60/60 [==============================] - 1s 11ms/step - loss: 0.0666 - acc: 0.9894 - auc: 0.9999\n",
      "Epoch 268/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0661 - acc: 0.9894 - auc: 0.9999\n",
      "Epoch 269/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0681 - acc: 0.9887 - auc: 0.9998\n",
      "Epoch 270/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0675 - acc: 0.9887 - auc: 0.9999\n",
      "Epoch 271/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0665 - acc: 0.9890 - auc: 0.9999\n",
      "Epoch 272/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0683 - acc: 0.9884 - auc: 0.9998\n",
      "Epoch 273/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0670 - acc: 0.9889 - auc: 0.9999\n",
      "Epoch 274/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0660 - acc: 0.9897 - auc: 0.9999\n",
      "Epoch 275/500\n",
      "60/60 [==============================] - 1s 11ms/step - loss: 0.0661 - acc: 0.9887 - auc: 0.9999\n",
      "Epoch 276/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0678 - acc: 0.9884 - auc: 0.9999\n",
      "Epoch 277/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0654 - acc: 0.9893 - auc: 0.9999\n",
      "Epoch 278/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0676 - acc: 0.9883 - auc: 0.9999\n",
      "Epoch 279/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0663 - acc: 0.9891 - auc: 0.9998\n",
      "Epoch 280/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0663 - acc: 0.9888 - auc: 0.9999\n",
      "Epoch 281/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0663 - acc: 0.9887 - auc: 0.9999\n",
      "Epoch 282/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0652 - acc: 0.9892 - auc: 0.9998\n",
      "Epoch 283/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0651 - acc: 0.9895 - auc: 0.9998\n",
      "Epoch 284/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0666 - acc: 0.9885 - auc: 0.9998\n",
      "Epoch 285/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0649 - acc: 0.9892 - auc: 0.9999\n",
      "Epoch 286/500\n",
      "60/60 [==============================] - 1s 11ms/step - loss: 0.0635 - acc: 0.9892 - auc: 0.9999\n",
      "Epoch 287/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0648 - acc: 0.9885 - auc: 0.9999\n",
      "Epoch 288/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0660 - acc: 0.9892 - auc: 0.9998\n",
      "Epoch 289/500\n",
      "60/60 [==============================] - 1s 11ms/step - loss: 0.0663 - acc: 0.9887 - auc: 0.9998\n",
      "Epoch 290/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0632 - acc: 0.9900 - auc: 0.9999\n",
      "Epoch 291/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0649 - acc: 0.9890 - auc: 0.9999\n",
      "Epoch 292/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0652 - acc: 0.9886 - auc: 0.9999\n",
      "Epoch 293/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0630 - acc: 0.9898 - auc: 0.9999\n",
      "Epoch 294/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0636 - acc: 0.9892 - auc: 0.9999\n",
      "Epoch 295/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0649 - acc: 0.9888 - auc: 0.9999\n",
      "Epoch 296/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0630 - acc: 0.9894 - auc: 0.9999\n",
      "Epoch 297/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0634 - acc: 0.9896 - auc: 0.9999\n",
      "Epoch 298/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0640 - acc: 0.9891 - auc: 0.9999\n",
      "Epoch 299/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0638 - acc: 0.9891 - auc: 0.9998\n",
      "Epoch 300/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0636 - acc: 0.9889 - auc: 0.9999\n",
      "Epoch 301/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0641 - acc: 0.9889 - auc: 0.9998\n",
      "Epoch 302/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0636 - acc: 0.9890 - auc: 0.9999\n",
      "Epoch 303/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0625 - acc: 0.9899 - auc: 0.9999\n",
      "Epoch 304/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0625 - acc: 0.9896 - auc: 0.9999\n",
      "Epoch 305/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0629 - acc: 0.9892 - auc: 0.9999\n",
      "Epoch 306/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0625 - acc: 0.9895 - auc: 0.9999\n",
      "Epoch 307/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0615 - acc: 0.9897 - auc: 0.9999\n",
      "Epoch 308/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0620 - acc: 0.9898 - auc: 0.9999\n",
      "Epoch 309/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0610 - acc: 0.9899 - auc: 0.9999\n",
      "Epoch 310/500\n",
      "60/60 [==============================] - 1s 11ms/step - loss: 0.0622 - acc: 0.9896 - auc: 0.9999\n",
      "Epoch 311/500\n",
      "60/60 [==============================] - 1s 11ms/step - loss: 0.0617 - acc: 0.9899 - auc: 0.9999\n",
      "Epoch 312/500\n",
      "60/60 [==============================] - 1s 11ms/step - loss: 0.0615 - acc: 0.9898 - auc: 0.9998\n",
      "Epoch 313/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0617 - acc: 0.9899 - auc: 0.9999\n",
      "Epoch 314/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0631 - acc: 0.9890 - auc: 0.9998\n",
      "Epoch 315/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0612 - acc: 0.9896 - auc: 0.9999\n",
      "Epoch 316/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0622 - acc: 0.9894 - auc: 0.9999\n",
      "Epoch 317/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0618 - acc: 0.9895 - auc: 0.9999\n",
      "Epoch 318/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0601 - acc: 0.9902 - auc: 0.9999\n",
      "Epoch 319/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0606 - acc: 0.9899 - auc: 0.9999\n",
      "Epoch 320/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0611 - acc: 0.9899 - auc: 0.9998\n",
      "Epoch 321/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0603 - acc: 0.9902 - auc: 0.9998\n",
      "Epoch 322/500\n",
      "60/60 [==============================] - 1s 11ms/step - loss: 0.0604 - acc: 0.9902 - auc: 0.9999\n",
      "Epoch 323/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0605 - acc: 0.9895 - auc: 0.9999\n",
      "Epoch 324/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0599 - acc: 0.9899 - auc: 0.9999\n",
      "Epoch 325/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0607 - acc: 0.9900 - auc: 0.9999\n",
      "Epoch 326/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0611 - acc: 0.9897 - auc: 0.9999\n",
      "Epoch 327/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0592 - acc: 0.9903 - auc: 0.9999\n",
      "Epoch 328/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0587 - acc: 0.9902 - auc: 0.9999\n",
      "Epoch 329/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0620 - acc: 0.9893 - auc: 0.9998\n",
      "Epoch 330/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0607 - acc: 0.9894 - auc: 0.9999\n",
      "Epoch 331/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0592 - acc: 0.9901 - auc: 0.9999\n",
      "Epoch 332/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0580 - acc: 0.9906 - auc: 0.9999\n",
      "Epoch 333/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0591 - acc: 0.9904 - auc: 0.9999\n",
      "Epoch 334/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0593 - acc: 0.9904 - auc: 0.9998\n",
      "Epoch 335/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0594 - acc: 0.9900 - auc: 0.9999\n",
      "Epoch 336/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0588 - acc: 0.9903 - auc: 0.9999\n",
      "Epoch 337/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0588 - acc: 0.9904 - auc: 0.9999\n",
      "Epoch 338/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0594 - acc: 0.9901 - auc: 0.9999\n",
      "Epoch 339/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0588 - acc: 0.9895 - auc: 0.9999\n",
      "Epoch 340/500\n",
      "60/60 [==============================] - 1s 11ms/step - loss: 0.0593 - acc: 0.9899 - auc: 0.9999\n",
      "Epoch 341/500\n",
      "60/60 [==============================] - 1s 11ms/step - loss: 0.0607 - acc: 0.9895 - auc: 0.9998\n",
      "Epoch 342/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0595 - acc: 0.9900 - auc: 0.9999\n",
      "Epoch 343/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0581 - acc: 0.9902 - auc: 0.9999\n",
      "Epoch 344/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0601 - acc: 0.9895 - auc: 0.9999\n",
      "Epoch 345/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0585 - acc: 0.9899 - auc: 0.9998\n",
      "Epoch 346/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0585 - acc: 0.9901 - auc: 0.9999\n",
      "Epoch 347/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0593 - acc: 0.9902 - auc: 0.9998\n",
      "Epoch 348/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0596 - acc: 0.9898 - auc: 0.9999\n",
      "Epoch 349/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0581 - acc: 0.9903 - auc: 0.9999\n",
      "Epoch 350/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0588 - acc: 0.9897 - auc: 0.9999\n",
      "Epoch 351/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0575 - acc: 0.9906 - auc: 0.9999\n",
      "Epoch 352/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0578 - acc: 0.9900 - auc: 0.9999\n",
      "Epoch 353/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0581 - acc: 0.9903 - auc: 0.9999\n",
      "Epoch 354/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0586 - acc: 0.9895 - auc: 0.9999\n",
      "Epoch 355/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0583 - acc: 0.9898 - auc: 0.9999\n",
      "Epoch 356/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0575 - acc: 0.9903 - auc: 0.9999\n",
      "Epoch 357/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0586 - acc: 0.9899 - auc: 0.9998\n",
      "Epoch 358/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0581 - acc: 0.9901 - auc: 0.9999\n",
      "Epoch 359/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0575 - acc: 0.9903 - auc: 0.9999\n",
      "Epoch 360/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0585 - acc: 0.9905 - auc: 0.9998\n",
      "Epoch 361/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0578 - acc: 0.9901 - auc: 0.9999\n",
      "Epoch 362/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0579 - acc: 0.9896 - auc: 0.9999\n",
      "Epoch 363/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0572 - acc: 0.9904 - auc: 0.9999\n",
      "Epoch 364/500\n",
      "60/60 [==============================] - 1s 11ms/step - loss: 0.0572 - acc: 0.9900 - auc: 0.9999\n",
      "Epoch 365/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0571 - acc: 0.9905 - auc: 0.9999\n",
      "Epoch 366/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0566 - acc: 0.9906 - auc: 0.9999\n",
      "Epoch 367/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0575 - acc: 0.9901 - auc: 0.9998\n",
      "Epoch 368/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0579 - acc: 0.9903 - auc: 0.9999\n",
      "Epoch 369/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0558 - acc: 0.9905 - auc: 0.9999\n",
      "Epoch 370/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0560 - acc: 0.9907 - auc: 0.9999\n",
      "Epoch 371/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0559 - acc: 0.9906 - auc: 0.9999\n",
      "Epoch 372/500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0562 - acc: 0.9909 - auc: 0.9999\n",
      "Epoch 373/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0559 - acc: 0.9905 - auc: 0.9999\n",
      "Epoch 374/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0575 - acc: 0.9900 - auc: 0.9999\n",
      "Epoch 375/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0572 - acc: 0.9904 - auc: 0.9999\n",
      "Epoch 376/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0534 - acc: 0.9916 - auc: 0.9999\n",
      "Epoch 377/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0555 - acc: 0.9905 - auc: 0.9999\n",
      "Epoch 378/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0575 - acc: 0.9899 - auc: 0.9999\n",
      "Epoch 379/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0566 - acc: 0.9904 - auc: 0.9999\n",
      "Epoch 380/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0566 - acc: 0.9904 - auc: 0.9999\n",
      "Epoch 381/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0567 - acc: 0.9906 - auc: 0.9999\n",
      "Epoch 382/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0568 - acc: 0.9901 - auc: 0.9999\n",
      "Epoch 383/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0541 - acc: 0.9913 - auc: 0.9999\n",
      "Epoch 384/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0551 - acc: 0.9910 - auc: 0.9999\n",
      "Epoch 385/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0561 - acc: 0.9903 - auc: 0.9999\n",
      "Epoch 386/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0555 - acc: 0.9903 - auc: 0.9999\n",
      "Epoch 387/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0567 - acc: 0.9900 - auc: 0.9999\n",
      "Epoch 388/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0565 - acc: 0.9899 - auc: 0.9999\n",
      "Epoch 389/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0552 - acc: 0.9908 - auc: 0.9999\n",
      "Epoch 390/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0542 - acc: 0.9909 - auc: 0.9999\n",
      "Epoch 391/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0547 - acc: 0.9911 - auc: 0.9999\n",
      "Epoch 392/500\n",
      "60/60 [==============================] - 0s 8ms/step - loss: 0.0552 - acc: 0.9906 - auc: 0.9999\n",
      "Epoch 393/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0555 - acc: 0.9905 - auc: 0.9999\n",
      "Epoch 394/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0546 - acc: 0.9911 - auc: 0.9999\n",
      "Epoch 395/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0558 - acc: 0.9902 - auc: 0.9999\n",
      "Epoch 396/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0557 - acc: 0.9907 - auc: 0.9999\n",
      "Epoch 397/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0554 - acc: 0.9908 - auc: 0.9999\n",
      "Epoch 398/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0548 - acc: 0.9908 - auc: 0.9999\n",
      "Epoch 399/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0548 - acc: 0.9904 - auc: 0.9999\n",
      "Epoch 400/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0550 - acc: 0.9907 - auc: 0.9999\n",
      "Epoch 401/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0549 - acc: 0.9907 - auc: 0.9999\n",
      "Epoch 402/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0558 - acc: 0.9904 - auc: 0.9999\n",
      "Epoch 403/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0531 - acc: 0.9917 - auc: 0.9999\n",
      "Epoch 404/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0553 - acc: 0.9904 - auc: 0.9998\n",
      "Epoch 405/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0557 - acc: 0.9899 - auc: 0.9999\n",
      "Epoch 406/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0539 - acc: 0.9912 - auc: 0.9999\n",
      "Epoch 407/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0533 - acc: 0.9911 - auc: 0.9999\n",
      "Epoch 408/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0544 - acc: 0.9909 - auc: 0.9999\n",
      "Epoch 409/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0540 - acc: 0.9912 - auc: 0.9999\n",
      "Epoch 410/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0544 - acc: 0.9910 - auc: 0.9999\n",
      "Epoch 411/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0542 - acc: 0.9903 - auc: 0.9999\n",
      "Epoch 412/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0533 - acc: 0.9913 - auc: 0.9999\n",
      "Epoch 413/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0527 - acc: 0.9912 - auc: 0.9999\n",
      "Epoch 414/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0551 - acc: 0.9901 - auc: 0.9999\n",
      "Epoch 415/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0544 - acc: 0.9906 - auc: 0.9998\n",
      "Epoch 416/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0541 - acc: 0.9913 - auc: 0.9999\n",
      "Epoch 417/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0534 - acc: 0.9914 - auc: 0.9999\n",
      "Epoch 418/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0541 - acc: 0.9902 - auc: 0.9999\n",
      "Epoch 419/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0520 - acc: 0.9915 - auc: 0.9999\n",
      "Epoch 420/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0536 - acc: 0.9909 - auc: 0.9999\n",
      "Epoch 421/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0540 - acc: 0.9906 - auc: 0.9999\n",
      "Epoch 422/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0533 - acc: 0.9914 - auc: 0.9999\n",
      "Epoch 423/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0546 - acc: 0.9904 - auc: 0.9999\n",
      "Epoch 424/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0541 - acc: 0.9904 - auc: 0.9999\n",
      "Epoch 425/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0535 - acc: 0.9908 - auc: 0.9999\n",
      "Epoch 426/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0528 - acc: 0.9910 - auc: 0.9999\n",
      "Epoch 427/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0530 - acc: 0.9911 - auc: 0.9999\n",
      "Epoch 428/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0543 - acc: 0.9908 - auc: 0.9998\n",
      "Epoch 429/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0535 - acc: 0.9909 - auc: 0.9999\n",
      "Epoch 430/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0529 - acc: 0.9909 - auc: 0.9999\n",
      "Epoch 431/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0529 - acc: 0.9910 - auc: 0.9999\n",
      "Epoch 432/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0528 - acc: 0.9905 - auc: 0.9999\n",
      "Epoch 433/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0537 - acc: 0.9905 - auc: 0.9999\n",
      "Epoch 434/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0551 - acc: 0.9902 - auc: 0.9998\n",
      "Epoch 435/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0543 - acc: 0.9905 - auc: 0.9999\n",
      "Epoch 436/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0536 - acc: 0.9908 - auc: 0.9999\n",
      "Epoch 437/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0538 - acc: 0.9909 - auc: 0.9998\n",
      "Epoch 438/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0527 - acc: 0.9911 - auc: 0.9999\n",
      "Epoch 439/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0526 - acc: 0.9909 - auc: 0.9999\n",
      "Epoch 440/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0520 - acc: 0.9913 - auc: 0.9999\n",
      "Epoch 441/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0530 - acc: 0.9909 - auc: 0.9999\n",
      "Epoch 442/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0522 - acc: 0.9909 - auc: 0.9999\n",
      "Epoch 443/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0520 - acc: 0.9909 - auc: 0.9999\n",
      "Epoch 444/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0521 - acc: 0.9913 - auc: 0.9999\n",
      "Epoch 445/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0527 - acc: 0.9911 - auc: 0.9999\n",
      "Epoch 446/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0527 - acc: 0.9913 - auc: 0.9999\n",
      "Epoch 447/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0524 - acc: 0.9909 - auc: 0.9999\n",
      "Epoch 448/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0530 - acc: 0.9910 - auc: 0.9999\n",
      "Epoch 449/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0522 - acc: 0.9915 - auc: 0.9999\n",
      "Epoch 450/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0526 - acc: 0.9911 - auc: 0.9999\n",
      "Epoch 451/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0499 - acc: 0.9921 - auc: 0.9999\n",
      "Epoch 452/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0516 - acc: 0.9912 - auc: 0.9999\n",
      "Epoch 453/500\n",
      "60/60 [==============================] - 1s 12ms/step - loss: 0.0527 - acc: 0.9914 - auc: 0.9999\n",
      "Epoch 454/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0523 - acc: 0.9914 - auc: 0.9999\n",
      "Epoch 455/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0520 - acc: 0.9912 - auc: 0.9999\n",
      "Epoch 456/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0507 - acc: 0.9916 - auc: 0.9999\n",
      "Epoch 457/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0517 - acc: 0.9911 - auc: 0.9999\n",
      "Epoch 458/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0505 - acc: 0.9918 - auc: 0.9999\n",
      "Epoch 459/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0509 - acc: 0.9912 - auc: 0.9999\n",
      "Epoch 460/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0513 - acc: 0.9915 - auc: 0.9999\n",
      "Epoch 461/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0517 - acc: 0.9908 - auc: 0.9999\n",
      "Epoch 462/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0512 - acc: 0.9912 - auc: 0.9999\n",
      "Epoch 463/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0527 - acc: 0.9906 - auc: 0.9999\n",
      "Epoch 464/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0497 - acc: 0.9921 - auc: 0.9999\n",
      "Epoch 465/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0516 - acc: 0.9912 - auc: 0.9999\n",
      "Epoch 466/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0504 - acc: 0.9915 - auc: 0.9999\n",
      "Epoch 467/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0517 - acc: 0.9912 - auc: 0.9999\n",
      "Epoch 468/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0507 - acc: 0.9916 - auc: 0.9999\n",
      "Epoch 469/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0515 - acc: 0.9912 - auc: 0.9999\n",
      "Epoch 470/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0509 - acc: 0.9912 - auc: 0.9999\n",
      "Epoch 471/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0509 - acc: 0.9915 - auc: 0.9999\n",
      "Epoch 472/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0498 - acc: 0.9917 - auc: 0.9999\n",
      "Epoch 473/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0504 - acc: 0.9916 - auc: 0.9999\n",
      "Epoch 474/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0517 - acc: 0.9910 - auc: 0.9999\n",
      "Epoch 475/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0529 - acc: 0.9906 - auc: 0.9999\n",
      "Epoch 476/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0505 - acc: 0.9918 - auc: 0.9999\n",
      "Epoch 477/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0518 - acc: 0.9911 - auc: 0.9999\n",
      "Epoch 478/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0517 - acc: 0.9911 - auc: 0.9999\n",
      "Epoch 479/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0507 - acc: 0.9913 - auc: 0.9999\n",
      "Epoch 480/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0501 - acc: 0.9918 - auc: 0.9999\n",
      "Epoch 481/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0526 - acc: 0.9908 - auc: 0.9999\n",
      "Epoch 482/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0516 - acc: 0.9908 - auc: 0.9999\n",
      "Epoch 483/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0504 - acc: 0.9914 - auc: 0.9999\n",
      "Epoch 484/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0503 - acc: 0.9918 - auc: 0.9998\n",
      "Epoch 485/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0494 - acc: 0.9918 - auc: 1.0000\n",
      "Epoch 486/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0510 - acc: 0.9912 - auc: 0.9999\n",
      "Epoch 487/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0500 - acc: 0.9922 - auc: 0.9999\n",
      "Epoch 488/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0501 - acc: 0.9917 - auc: 0.9999\n",
      "Epoch 489/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0516 - acc: 0.9912 - auc: 0.9999\n",
      "Epoch 490/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0503 - acc: 0.9914 - auc: 0.9999\n",
      "Epoch 491/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0511 - acc: 0.9909 - auc: 0.9999\n",
      "Epoch 492/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0517 - acc: 0.9910 - auc: 0.9999\n",
      "Epoch 493/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0493 - acc: 0.9917 - auc: 0.9999\n",
      "Epoch 494/500\n",
      "60/60 [==============================] - 1s 8ms/step - loss: 0.0505 - acc: 0.9911 - auc: 0.9999\n",
      "Epoch 495/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0509 - acc: 0.9911 - auc: 0.9999\n",
      "Epoch 496/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0495 - acc: 0.9917 - auc: 0.9999\n",
      "Epoch 497/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0513 - acc: 0.9910 - auc: 0.9999\n",
      "Epoch 498/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0520 - acc: 0.9910 - auc: 0.9999\n",
      "Epoch 499/500\n",
      "60/60 [==============================] - 1s 10ms/step - loss: 0.0495 - acc: 0.9918 - auc: 0.9999\n",
      "Epoch 500/500\n",
      "60/60 [==============================] - 1s 9ms/step - loss: 0.0506 - acc: 0.9915 - auc: 0.9999\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<tensorflow.python.keras.callbacks.History at 0x7f7fe056df10>"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "realx.evalx.compile(loss=loss,\n",
    "                    optimizer=optimizer,\n",
    "                    metrics=metrics)\n",
    "realx.evalx.fit(x_train,\n",
    "                y_train,\n",
    "                epochs=epochs,\n",
    "                batch_size=batch_size,\n",
    "                verbose=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# EVAL-X Evaluation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-03-01T22:17:43.366619Z",
     "start_time": "2021-03-01T22:17:42.518608Z"
    }
   },
   "outputs": [],
   "source": [
    "from sklearn.metrics import roc_auc_score, accuracy_score, average_precision_score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-03-01T22:17:45.239704Z",
     "start_time": "2021-03-01T22:17:44.283453Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "REAL-X: ACC:0.9713, AUROC:0.9990713405555556, AUPRC:0.9940232888389957\n",
      "EVAL-X: eACC:0.9404, eAUROC:0.9974457355555557, eAUPRC:0.9835009266679966\n"
     ]
    }
   ],
   "source": [
    "ey_score = realx.evaluate(x_test, batch_size)\n",
    "y_score = realx.predict(x_test, batch_size)\n",
    "\n",
    "#Get Results\n",
    "### ROC\n",
    "AUROC = roc_auc_score(y_test, y_score, 'micro')\n",
    "eAUROC = roc_auc_score(y_test, ey_score, 'micro')\n",
    "### PRC\n",
    "AUPRC = average_precision_score(y_test, y_score, 'micro')\n",
    "eAUPRC = average_precision_score(y_test, ey_score, 'micro')\n",
    "\n",
    "## b) Compute Accuracy\n",
    "ACC = accuracy_score(y_test.argmax(1), y_score.argmax(1))\n",
    "eACC = accuracy_score(y_test.argmax(1), ey_score.argmax(1))\n",
    "\n",
    "print('REAL-X: ACC:{}, AUROC:{}, AUPRC:{}'.format(ACC, AUROC, AUPRC))\n",
    "print('EVAL-X: eACC:{}, eAUROC:{}, eAUPRC:{}'.format(eACC, eAUROC, eAUPRC))"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.6"
  },
  "varInspector": {
   "cols": {
    "lenName": 16,
    "lenType": 16,
    "lenVar": 40
   },
   "kernels_config": {
    "python": {
     "delete_cmd_postfix": "",
     "delete_cmd_prefix": "del ",
     "library": "var_list.py",
     "varRefreshCmd": "print(var_dic_list())"
    },
    "r": {
     "delete_cmd_postfix": ") ",
     "delete_cmd_prefix": "rm(",
     "library": "var_list.r",
     "varRefreshCmd": "cat(var_dic_list()) "
    }
   },
   "types_to_exclude": [
    "module",
    "function",
    "builtin_function_or_method",
    "instance",
    "_Feature"
   ],
   "window_display": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}