Covid-Classification-with-Keras.ipynb

{
  "cells": [
    {
      "attachments": {},
      "cell_type": "markdown",
      "metadata": {
        "id": "ub9_Q9k3g3xS"
      },
      "source": [
        "# Introduction\n",
        "In this project I did a classification problem, using AutoEncoder as Dimensionality reduction tool.\n",
        "\n",
        "The dataset is the clinical data of patients who tested for Covid19."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Yjot4Zz3g1Fk"
      },
      "source": [
        "# Imports"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 7,
      "metadata": {
        "id": "fbL_VNfo1CER"
      },
      "outputs": [],
      "source": [
        "import keras\n",
        "from keras import layers\n",
        "\n",
        "import numpy as np\n",
        "import pandas as pd\n",
        "import matplotlib.pyplot as plt\n",
        "import pandas as pd\n",
        "\n",
        "from sklearn.preprocessing import LabelEncoder\n",
        "from sklearn.model_selection import train_test_split\n",
        "\n",
        "from sklearn.metrics import classification_report, f1_score\n",
        "from sklearn.metrics import confusion_matrix\n",
        "import seaborn as sn"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9P_Mq1_-KSLj"
      },
      "source": [
        "# Data Preparation"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 8,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 386
        },
        "id": "7247uzx_KSLj",
        "outputId": "64cd457b-04d0-4a74-da20-e2af198ff9e6"
      },
      "outputs": [
        {
          "data": {
            "text/html": [
              "\n",
              "  <div id=\"df-f97ae724-b0c4-45c0-bdbf-5dd0273acc9b\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>Breathing Problem</th>\n",
              "      <th>Fever</th>\n",
              "      <th>Dry Cough</th>\n",
              "      <th>Sore throat</th>\n",
              "      <th>Running Nose</th>\n",
              "      <th>Asthma</th>\n",
              "      <th>Chronic Lung Disease</th>\n",
              "      <th>Headache</th>\n",
              "      <th>Heart Disease</th>\n",
              "      <th>Diabetes</th>\n",
              "      <th>...</th>\n",
              "      <th>Fatigue</th>\n",
              "      <th>Gastrointestinal</th>\n",
              "      <th>Abroad travel</th>\n",
              "      <th>Contact with COVID Patient</th>\n",
              "      <th>Attended Large Gathering</th>\n",
              "      <th>Visited Public Exposed Places</th>\n",
              "      <th>Family working in Public Exposed Places</th>\n",
              "      <th>Wearing Masks</th>\n",
              "      <th>Sanitization from Market</th>\n",
              "      <th>COVID-19</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>...</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>...</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>...</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>...</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>...</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>5 rows × 21 columns</p>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-f97ae724-b0c4-45c0-bdbf-5dd0273acc9b')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-f97ae724-b0c4-45c0-bdbf-5dd0273acc9b button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-f97ae724-b0c4-45c0-bdbf-5dd0273acc9b');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ],
            "text/plain": [
              "   Breathing Problem  Fever  Dry Cough  Sore throat  Running Nose  Asthma  \\\n",
              "0                1.0    1.0        1.0          1.0           1.0     0.0   \n",
              "1                1.0    1.0        1.0          1.0           0.0     1.0   \n",
              "2                1.0    1.0        1.0          1.0           1.0     1.0   \n",
              "3                1.0    1.0        1.0          0.0           0.0     1.0   \n",
              "4                1.0    1.0        1.0          1.0           1.0     0.0   \n",
              "\n",
              "   Chronic Lung Disease  Headache  Heart Disease  Diabetes  ...  Fatigue   \\\n",
              "0                   0.0       0.0            0.0       1.0  ...       1.0   \n",
              "1                   1.0       1.0            0.0       0.0  ...       1.0   \n",
              "2                   1.0       1.0            0.0       1.0  ...       1.0   \n",
              "3                   0.0       0.0            1.0       1.0  ...       0.0   \n",
              "4                   1.0       1.0            1.0       1.0  ...       0.0   \n",
              "\n",
              "   Gastrointestinal   Abroad travel  Contact with COVID Patient  \\\n",
              "0                1.0            0.0                         1.0   \n",
              "1                0.0            0.0                         0.0   \n",
              "2                1.0            1.0                         0.0   \n",
              "3                0.0            1.0                         0.0   \n",
              "4                1.0            0.0                         1.0   \n",
              "\n",
              "   Attended Large Gathering  Visited Public Exposed Places  \\\n",
              "0                       0.0                            1.0   \n",
              "1                       1.0                            1.0   \n",
              "2                       0.0                            0.0   \n",
              "3                       1.0                            1.0   \n",
              "4                       0.0                            1.0   \n",
              "\n",
              "   Family working in Public Exposed Places  Wearing Masks  \\\n",
              "0                                      1.0            0.0   \n",
              "1                                      0.0            0.0   \n",
              "2                                      0.0            0.0   \n",
              "3                                      0.0            0.0   \n",
              "4                                      0.0            0.0   \n",
              "\n",
              "   Sanitization from Market  COVID-19  \n",
              "0                       0.0       1.0  \n",
              "1                       0.0       1.0  \n",
              "2                       0.0       1.0  \n",
              "3                       0.0       1.0  \n",
              "4                       0.0       1.0  \n",
              "\n",
              "[5 rows x 21 columns]"
            ]
          },
          "execution_count": 8,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "covid_data = pd.read_csv('Covid Dataset.csv')\n",
        "categorical_feature_mask = covid_data.dtypes == object\n",
        "cateforical_cols = covid_data.columns[categorical_feature_mask].tolist()\n",
        "le = LabelEncoder()\n",
        "covid_data[cateforical_cols] = covid_data[cateforical_cols].apply(lambda col: le.fit_transform(col))\n",
        "covid_data = covid_data.astype(float)\n",
        "covid_data.head()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 12,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "UbcCb_G2hYwg",
        "outputId": "a927d520-05c5-41fc-b6c1-20a61dab5be2"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "X_train: (4075, 20)\n",
            "y_train: (4075,)\n",
            "X_test: (1359, 20)\n",
            "y_test: (1359,)\n"
          ]
        }
      ],
      "source": [
        "# Extracting X and Y from the dataset\n",
        "NUM_FEATURES = 20\n",
        "X_total = covid_data.iloc[:, 0:NUM_FEATURES].values\n",
        "y_total = covid_data.iloc[:,NUM_FEATURES].values\n",
        "\n",
        "\n",
        "# Spltting the data into train and test sets \n",
        "X_train, X_test, y_train, y_test = train_test_split(X_total, y_total, test_size = 0.25, random_state = 0)\n",
        "print(f'X_train: {X_train.shape}')\n",
        "print(f'y_train: {y_train.shape}')\n",
        "print(f'X_test: {X_test.shape}')\n",
        "print(f'y_test: {y_test.shape}')"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "wYM0t8DFKSLk"
      },
      "source": [
        "---------------"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "vsmMzLfvKSLl"
      },
      "source": [
        "# 1. AutoEncoder as nonlinear dimensionality reduction method\n",
        "\n",
        "Autoencoder is an unsupervised artificial neural network that compresses the data to lower dimension and then reconstructs the input back. Autoencoder finds the representation of the data in a lower dimension by focusing more on the important features getting rid of noise and redundancy. It's based on Encoder-Decoder architecture, where encoder encodes the high-dimensional data to lower-dimension and decoder takes the lower-dimensional data and tries to reconstruct the original high-dimensional data.\n",
        "\n",
        "![picture](https://drive.google.com/uc?id=1RTZwx4xL6zFV_nUENBgWlFKLKldPoyI-)\n",
        "\n",
        "In the above Diagram, X is the input data, z is the lower-dimension representation of input X and X’ is the reconstructed input data. The mapping of higher to lower dimensions can be linear or non-linear depending on the choice of the activation function.\n",
        "\n",
        "In this part we're gonna implement an autoencoder using Keras framework as dimensionality reduction module.\n",
        "\n",
        "In this notebook, we lower the dimension of our data from 20 to 2."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 15,
      "metadata": {
        "id": "nOQ6w3EjKSLm"
      },
      "outputs": [],
      "source": [
        "def AutoEncoder(input_dim, encode_dims, encode_dim, decode_dims=None):\n",
        "    \n",
        "    decode_dims = decode_dims if decode_dims else reversed(encode_dims)\n",
        "\n",
        "    # Defining input from input_dim\n",
        "    input = keras.Input(shape=(input_dim))\n",
        "\n",
        "    # The encoder. This part will be used later as reduction tool\n",
        "    encoder = keras.Sequential([\n",
        "        *[layers.Dense(dim, activation='relu') for dim in encode_dims]\n",
        "        ,layers.Dense(encode_dim, activation='linear')\n",
        "    ], name='Encoder')\n",
        "    encoded = encoder(input)\n",
        "    \n",
        "    # The Decoder\n",
        "    decoder = keras.Sequential([\n",
        "        *[layers.Dense(dim, activation='relu') for dim in decode_dims]\n",
        "        ,layers.Dense(input_dim, activation='sigmoid')\n",
        "    ], name='Decoder')\n",
        "    decoded = decoder(encoded)\n",
        "\n",
        "    return keras.Model(input, decoded)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 16,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "4TZ9xrPAUWGD",
        "outputId": "0487811f-502b-49dd-dd98-8d6375dc6208"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Model: \"model_1\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " input_2 (InputLayer)        [(None, 20)]              0         \n",
            "                                                                 \n",
            " Encoder (Sequential)        (None, 2)                 13090     \n",
            "                                                                 \n",
            " Decoder (Sequential)        (None, 20)                13108     \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 26,198\n",
            "Trainable params: 26,198\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ],
      "source": [
        "autoencoder = AutoEncoder(20, [128, 64, 32], 2)\n",
        "autoencoder.summary()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "MNDxvo_DlXfh"
      },
      "source": [
        "# Training the model\n",
        "Note that we are using X_train as both input and expected output of the model."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 17,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "DpKicMQOKSLn",
        "outputId": "212d317d-a9c1-46b3-de91-42c0921b9219"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 1/400\n",
            "64/64 [==============================] - 4s 7ms/step - loss: 0.5287 - val_loss: 0.4744 - lr: 0.0100\n",
            "Epoch 2/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.4480 - val_loss: 0.4318 - lr: 0.0100\n",
            "Epoch 3/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.4139 - val_loss: 0.4069 - lr: 0.0100\n",
            "Epoch 4/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.3913 - val_loss: 0.3892 - lr: 0.0100\n",
            "Epoch 5/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.3744 - val_loss: 0.3844 - lr: 0.0100\n",
            "Epoch 6/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.3665 - val_loss: 0.3518 - lr: 0.0100\n",
            "Epoch 7/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.3669 - val_loss: 0.3588 - lr: 0.0100\n",
            "Epoch 8/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.3457 - val_loss: 0.3385 - lr: 0.0100\n",
            "Epoch 9/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.3278 - val_loss: 0.3246 - lr: 0.0100\n",
            "Epoch 10/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.3180 - val_loss: 0.3289 - lr: 0.0100\n",
            "Epoch 11/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.3224 - val_loss: 0.3246 - lr: 0.0099\n",
            "Epoch 12/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.3146 - val_loss: 0.3246 - lr: 0.0099\n",
            "Epoch 13/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.3304 - val_loss: 0.3070 - lr: 0.0099\n",
            "Epoch 14/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.2873 - val_loss: 0.2971 - lr: 0.0098\n",
            "Epoch 15/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.2828 - val_loss: 0.3288 - lr: 0.0098\n",
            "Epoch 16/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.3306 - val_loss: 0.3153 - lr: 0.0097\n",
            "Epoch 17/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.2841 - val_loss: 0.2864 - lr: 0.0097\n",
            "Epoch 18/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.2707 - val_loss: 0.2806 - lr: 0.0096\n",
            "Epoch 19/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.2768 - val_loss: 0.2966 - lr: 0.0096\n",
            "Epoch 20/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.2903 - val_loss: 0.2770 - lr: 0.0095\n",
            "Epoch 21/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.2489 - val_loss: 0.2584 - lr: 0.0095\n",
            "Epoch 22/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.2449 - val_loss: 0.2516 - lr: 0.0094\n",
            "Epoch 23/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.2344 - val_loss: 0.2430 - lr: 0.0094\n",
            "Epoch 24/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.2327 - val_loss: 0.2532 - lr: 0.0093\n",
            "Epoch 25/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.2676 - val_loss: 0.2788 - lr: 0.0093\n",
            "Epoch 26/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.2453 - val_loss: 0.2420 - lr: 0.0092\n",
            "Epoch 27/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.2239 - val_loss: 0.2481 - lr: 0.0092\n",
            "Epoch 28/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.2244 - val_loss: 0.2302 - lr: 0.0091\n",
            "Epoch 29/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.2174 - val_loss: 0.2359 - lr: 0.0091\n",
            "Epoch 30/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.2227 - val_loss: 0.2335 - lr: 0.0090\n",
            "Epoch 31/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.2362 - val_loss: 0.3782 - lr: 0.0090\n",
            "Epoch 32/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.3100 - val_loss: 0.2746 - lr: 0.0090\n",
            "Epoch 33/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.2387 - val_loss: 0.2307 - lr: 0.0089\n",
            "Epoch 34/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.2219 - val_loss: 0.2369 - lr: 0.0089\n",
            "Epoch 35/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.2300 - val_loss: 0.2387 - lr: 0.0088\n",
            "Epoch 36/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.2327 - val_loss: 0.2564 - lr: 0.0088\n",
            "Epoch 37/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.2252 - val_loss: 0.2183 - lr: 0.0087\n",
            "Epoch 38/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.2010 - val_loss: 0.2005 - lr: 0.0087\n",
            "Epoch 39/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1910 - val_loss: 0.1997 - lr: 0.0086\n",
            "Epoch 40/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1913 - val_loss: 0.2142 - lr: 0.0086\n",
            "Epoch 41/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1834 - val_loss: 0.1855 - lr: 0.0086\n",
            "Epoch 42/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1770 - val_loss: 0.1899 - lr: 0.0085\n",
            "Epoch 43/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1796 - val_loss: 0.2099 - lr: 0.0085\n",
            "Epoch 44/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1984 - val_loss: 0.2289 - lr: 0.0084\n",
            "Epoch 45/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1976 - val_loss: 0.1957 - lr: 0.0084\n",
            "Epoch 46/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1841 - val_loss: 0.1865 - lr: 0.0083\n",
            "Epoch 47/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.1837 - val_loss: 0.1870 - lr: 0.0083\n",
            "Epoch 48/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1721 - val_loss: 0.1858 - lr: 0.0083\n",
            "Epoch 49/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1622 - val_loss: 0.1702 - lr: 0.0082\n",
            "Epoch 50/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1568 - val_loss: 0.1689 - lr: 0.0082\n",
            "Epoch 51/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.1599 - val_loss: 0.1670 - lr: 0.0081\n",
            "Epoch 52/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.1552 - val_loss: 0.1724 - lr: 0.0081\n",
            "Epoch 53/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1689 - val_loss: 0.1822 - lr: 0.0081\n",
            "Epoch 54/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1651 - val_loss: 0.1774 - lr: 0.0080\n",
            "Epoch 55/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1567 - val_loss: 0.1601 - lr: 0.0080\n",
            "Epoch 56/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1614 - val_loss: 0.1957 - lr: 0.0079\n",
            "Epoch 57/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1812 - val_loss: 0.1843 - lr: 0.0079\n",
            "Epoch 58/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.1672 - val_loss: 0.1657 - lr: 0.0079\n",
            "Epoch 59/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1504 - val_loss: 0.1557 - lr: 0.0078\n",
            "Epoch 60/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1371 - val_loss: 0.1488 - lr: 0.0078\n",
            "Epoch 61/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1516 - val_loss: 0.1841 - lr: 0.0077\n",
            "Epoch 62/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1640 - val_loss: 0.1888 - lr: 0.0077\n",
            "Epoch 63/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.2047 - val_loss: 0.2251 - lr: 0.0077\n",
            "Epoch 64/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1886 - val_loss: 0.1674 - lr: 0.0076\n",
            "Epoch 65/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1399 - val_loss: 0.1408 - lr: 0.0076\n",
            "Epoch 66/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1241 - val_loss: 0.1339 - lr: 0.0076\n",
            "Epoch 67/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.1220 - val_loss: 0.1343 - lr: 0.0075\n",
            "Epoch 68/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1273 - val_loss: 0.1387 - lr: 0.0075\n",
            "Epoch 69/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1295 - val_loss: 0.1405 - lr: 0.0074\n",
            "Epoch 70/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1343 - val_loss: 0.1384 - lr: 0.0074\n",
            "Epoch 71/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1312 - val_loss: 0.1448 - lr: 0.0074\n",
            "Epoch 72/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1252 - val_loss: 0.1330 - lr: 0.0073\n",
            "Epoch 73/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.1246 - val_loss: 0.1374 - lr: 0.0073\n",
            "Epoch 74/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.1326 - val_loss: 0.1369 - lr: 0.0073\n",
            "Epoch 75/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.1211 - val_loss: 0.1355 - lr: 0.0072\n",
            "Epoch 76/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.1295 - val_loss: 0.1302 - lr: 0.0072\n",
            "Epoch 77/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.1151 - val_loss: 0.1257 - lr: 0.0071\n",
            "Epoch 78/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.1126 - val_loss: 0.1199 - lr: 0.0071\n",
            "Epoch 79/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.1440 - val_loss: 0.1976 - lr: 0.0071\n",
            "Epoch 80/400\n",
            "64/64 [==============================] - 0s 7ms/step - loss: 0.2122 - val_loss: 0.2493 - lr: 0.0070\n",
            "Epoch 81/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.1626 - val_loss: 0.1368 - lr: 0.0070\n",
            "Epoch 82/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1177 - val_loss: 0.1220 - lr: 0.0070\n",
            "Epoch 83/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.1078 - val_loss: 0.1203 - lr: 0.0069\n",
            "Epoch 84/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.1137 - val_loss: 0.1458 - lr: 0.0069\n",
            "Epoch 85/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1245 - val_loss: 0.1304 - lr: 0.0069\n",
            "Epoch 86/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1134 - val_loss: 0.1182 - lr: 0.0068\n",
            "Epoch 87/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1050 - val_loss: 0.1140 - lr: 0.0068\n",
            "Epoch 88/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1055 - val_loss: 0.1138 - lr: 0.0068\n",
            "Epoch 89/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1003 - val_loss: 0.1119 - lr: 0.0067\n",
            "Epoch 90/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1025 - val_loss: 0.1278 - lr: 0.0067\n",
            "Epoch 91/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1216 - val_loss: 0.1482 - lr: 0.0067\n",
            "Epoch 92/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.1858 - val_loss: 0.1748 - lr: 0.0066\n",
            "Epoch 93/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1525 - val_loss: 0.1635 - lr: 0.0066\n",
            "Epoch 94/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1128 - val_loss: 0.1182 - lr: 0.0066\n",
            "Epoch 95/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0997 - val_loss: 0.1076 - lr: 0.0065\n",
            "Epoch 96/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0982 - val_loss: 0.1117 - lr: 0.0065\n",
            "Epoch 97/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1012 - val_loss: 0.1147 - lr: 0.0065\n",
            "Epoch 98/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1008 - val_loss: 0.1067 - lr: 0.0064\n",
            "Epoch 99/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0953 - val_loss: 0.1135 - lr: 0.0064\n",
            "Epoch 100/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0986 - val_loss: 0.1043 - lr: 0.0064\n",
            "Epoch 101/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0954 - val_loss: 0.1139 - lr: 0.0063\n",
            "Epoch 102/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1030 - val_loss: 0.1402 - lr: 0.0063\n",
            "Epoch 103/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1297 - val_loss: 0.1291 - lr: 0.0063\n",
            "Epoch 104/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1097 - val_loss: 0.1222 - lr: 0.0062\n",
            "Epoch 105/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.1130 - val_loss: 0.1289 - lr: 0.0062\n",
            "Epoch 106/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.1248 - val_loss: 0.1220 - lr: 0.0062\n",
            "Epoch 107/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1244 - val_loss: 0.1409 - lr: 0.0061\n",
            "Epoch 108/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1509 - val_loss: 0.1410 - lr: 0.0061\n",
            "Epoch 109/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.1013 - val_loss: 0.1012 - lr: 0.0061\n",
            "Epoch 110/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0873 - val_loss: 0.0972 - lr: 0.0061\n",
            "Epoch 111/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0865 - val_loss: 0.1016 - lr: 0.0060\n",
            "Epoch 112/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0906 - val_loss: 0.0937 - lr: 0.0060\n",
            "Epoch 113/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0852 - val_loss: 0.0999 - lr: 0.0060\n",
            "Epoch 114/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0837 - val_loss: 0.0943 - lr: 0.0059\n",
            "Epoch 115/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0832 - val_loss: 0.1029 - lr: 0.0059\n",
            "Epoch 116/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0939 - val_loss: 0.1049 - lr: 0.0059\n",
            "Epoch 117/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0988 - val_loss: 0.1029 - lr: 0.0058\n",
            "Epoch 118/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0894 - val_loss: 0.1122 - lr: 0.0058\n",
            "Epoch 119/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0996 - val_loss: 0.1173 - lr: 0.0058\n",
            "Epoch 120/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0988 - val_loss: 0.1252 - lr: 0.0058\n",
            "Epoch 121/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0995 - val_loss: 0.0998 - lr: 0.0057\n",
            "Epoch 122/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0938 - val_loss: 0.1113 - lr: 0.0057\n",
            "Epoch 123/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0919 - val_loss: 0.1012 - lr: 0.0057\n",
            "Epoch 124/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0778 - val_loss: 0.0862 - lr: 0.0056\n",
            "Epoch 125/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0775 - val_loss: 0.0902 - lr: 0.0056\n",
            "Epoch 126/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0788 - val_loss: 0.0861 - lr: 0.0056\n",
            "Epoch 127/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0734 - val_loss: 0.0834 - lr: 0.0056\n",
            "Epoch 128/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0729 - val_loss: 0.0850 - lr: 0.0055\n",
            "Epoch 129/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0881 - val_loss: 0.1065 - lr: 0.0055\n",
            "Epoch 130/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.1047 - val_loss: 0.1383 - lr: 0.0055\n",
            "Epoch 131/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.1235 - val_loss: 0.1254 - lr: 0.0055\n",
            "Epoch 132/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0869 - val_loss: 0.0857 - lr: 0.0054\n",
            "Epoch 133/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0728 - val_loss: 0.0847 - lr: 0.0054\n",
            "Epoch 134/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0713 - val_loss: 0.0804 - lr: 0.0054\n",
            "Epoch 135/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0679 - val_loss: 0.0781 - lr: 0.0053\n",
            "Epoch 136/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0675 - val_loss: 0.0802 - lr: 0.0053\n",
            "Epoch 137/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1143 - val_loss: 0.2656 - lr: 0.0053\n",
            "Epoch 138/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1609 - val_loss: 0.1245 - lr: 0.0053\n",
            "Epoch 139/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0987 - val_loss: 0.1025 - lr: 0.0052\n",
            "Epoch 140/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0743 - val_loss: 0.0779 - lr: 0.0052\n",
            "Epoch 141/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0673 - val_loss: 0.0747 - lr: 0.0052\n",
            "Epoch 142/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0630 - val_loss: 0.0706 - lr: 0.0052\n",
            "Epoch 143/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0657 - val_loss: 0.0773 - lr: 0.0051\n",
            "Epoch 144/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0683 - val_loss: 0.0801 - lr: 0.0051\n",
            "Epoch 145/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0685 - val_loss: 0.0844 - lr: 0.0051\n",
            "Epoch 146/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0754 - val_loss: 0.0870 - lr: 0.0051\n",
            "Epoch 147/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0678 - val_loss: 0.0791 - lr: 0.0050\n",
            "Epoch 148/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0681 - val_loss: 0.0795 - lr: 0.0050\n",
            "Epoch 149/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0654 - val_loss: 0.0773 - lr: 0.0050\n",
            "Epoch 150/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0625 - val_loss: 0.0722 - lr: 0.0050\n",
            "Epoch 151/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0633 - val_loss: 0.0848 - lr: 0.0049\n",
            "Epoch 152/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0799 - val_loss: 0.1008 - lr: 0.0049\n",
            "Epoch 153/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1176 - val_loss: 0.1372 - lr: 0.0049\n",
            "Epoch 154/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1403 - val_loss: 0.1552 - lr: 0.0049\n",
            "Epoch 155/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1089 - val_loss: 0.0895 - lr: 0.0048\n",
            "Epoch 156/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0830 - val_loss: 0.0968 - lr: 0.0048\n",
            "Epoch 157/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0728 - val_loss: 0.0741 - lr: 0.0048\n",
            "Epoch 158/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0587 - val_loss: 0.0678 - lr: 0.0048\n",
            "Epoch 159/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0568 - val_loss: 0.0702 - lr: 0.0047\n",
            "Epoch 160/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0575 - val_loss: 0.0673 - lr: 0.0047\n",
            "Epoch 161/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0581 - val_loss: 0.0633 - lr: 0.0047\n",
            "Epoch 162/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0557 - val_loss: 0.0772 - lr: 0.0047\n",
            "Epoch 163/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0619 - val_loss: 0.0712 - lr: 0.0046\n",
            "Epoch 164/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0562 - val_loss: 0.0658 - lr: 0.0046\n",
            "Epoch 165/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0557 - val_loss: 0.0743 - lr: 0.0046\n",
            "Epoch 166/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0646 - val_loss: 0.0740 - lr: 0.0046\n",
            "Epoch 167/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0632 - val_loss: 0.0699 - lr: 0.0046\n",
            "Epoch 168/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0571 - val_loss: 0.0677 - lr: 0.0045\n",
            "Epoch 169/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0553 - val_loss: 0.0622 - lr: 0.0045\n",
            "Epoch 170/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0571 - val_loss: 0.0717 - lr: 0.0045\n",
            "Epoch 171/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0791 - val_loss: 0.0911 - lr: 0.0045\n",
            "Epoch 172/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0643 - val_loss: 0.0659 - lr: 0.0044\n",
            "Epoch 173/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0589 - val_loss: 0.0802 - lr: 0.0044\n",
            "Epoch 174/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.1174 - val_loss: 0.1270 - lr: 0.0044\n",
            "Epoch 175/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1427 - val_loss: 0.1252 - lr: 0.0044\n",
            "Epoch 176/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0835 - val_loss: 0.0787 - lr: 0.0044\n",
            "Epoch 177/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0605 - val_loss: 0.0674 - lr: 0.0043\n",
            "Epoch 178/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0566 - val_loss: 0.0673 - lr: 0.0043\n",
            "Epoch 179/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0548 - val_loss: 0.0647 - lr: 0.0043\n",
            "Epoch 180/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0488 - val_loss: 0.0582 - lr: 0.0043\n",
            "Epoch 181/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0485 - val_loss: 0.0592 - lr: 0.0042\n",
            "Epoch 182/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0506 - val_loss: 0.0586 - lr: 0.0042\n",
            "Epoch 183/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0479 - val_loss: 0.0593 - lr: 0.0042\n",
            "Epoch 184/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0465 - val_loss: 0.0577 - lr: 0.0042\n",
            "Epoch 185/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0475 - val_loss: 0.0608 - lr: 0.0042\n",
            "Epoch 186/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0489 - val_loss: 0.0610 - lr: 0.0041\n",
            "Epoch 187/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0693 - val_loss: 0.1923 - lr: 0.0041\n",
            "Epoch 188/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0982 - val_loss: 0.0851 - lr: 0.0041\n",
            "Epoch 189/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0708 - val_loss: 0.0702 - lr: 0.0041\n",
            "Epoch 190/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0543 - val_loss: 0.0573 - lr: 0.0041\n",
            "Epoch 191/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0466 - val_loss: 0.0545 - lr: 0.0040\n",
            "Epoch 192/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0441 - val_loss: 0.0551 - lr: 0.0040\n",
            "Epoch 193/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0488 - val_loss: 0.0604 - lr: 0.0040\n",
            "Epoch 194/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0460 - val_loss: 0.0544 - lr: 0.0040\n",
            "Epoch 195/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0465 - val_loss: 0.0606 - lr: 0.0040\n",
            "Epoch 196/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0481 - val_loss: 0.0596 - lr: 0.0039\n",
            "Epoch 197/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0638 - val_loss: 0.0842 - lr: 0.0039\n",
            "Epoch 198/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0810 - val_loss: 0.0990 - lr: 0.0039\n",
            "Epoch 199/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0756 - val_loss: 0.0711 - lr: 0.0039\n",
            "Epoch 200/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0568 - val_loss: 0.0765 - lr: 0.0039\n",
            "Epoch 201/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0497 - val_loss: 0.0575 - lr: 0.0038\n",
            "Epoch 202/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0440 - val_loss: 0.0512 - lr: 0.0038\n",
            "Epoch 203/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0415 - val_loss: 0.0518 - lr: 0.0038\n",
            "Epoch 204/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0419 - val_loss: 0.0527 - lr: 0.0038\n",
            "Epoch 205/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0431 - val_loss: 0.0510 - lr: 0.0038\n",
            "Epoch 206/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0412 - val_loss: 0.0518 - lr: 0.0037\n",
            "Epoch 207/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0453 - val_loss: 0.0559 - lr: 0.0037\n",
            "Epoch 208/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0421 - val_loss: 0.0512 - lr: 0.0037\n",
            "Epoch 209/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0454 - val_loss: 0.0565 - lr: 0.0037\n",
            "Epoch 210/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0657 - val_loss: 0.1016 - lr: 0.0037\n",
            "Epoch 211/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.1204 - val_loss: 0.1215 - lr: 0.0037\n",
            "Epoch 212/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0757 - val_loss: 0.0623 - lr: 0.0036\n",
            "Epoch 213/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0462 - val_loss: 0.0509 - lr: 0.0036\n",
            "Epoch 214/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0388 - val_loss: 0.0470 - lr: 0.0036\n",
            "Epoch 215/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0366 - val_loss: 0.0468 - lr: 0.0036\n",
            "Epoch 216/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0361 - val_loss: 0.0462 - lr: 0.0036\n",
            "Epoch 217/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0365 - val_loss: 0.0461 - lr: 0.0035\n",
            "Epoch 218/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0368 - val_loss: 0.0500 - lr: 0.0035\n",
            "Epoch 219/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0371 - val_loss: 0.0462 - lr: 0.0035\n",
            "Epoch 220/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0370 - val_loss: 0.0482 - lr: 0.0035\n",
            "Epoch 221/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0401 - val_loss: 0.0496 - lr: 0.0035\n",
            "Epoch 222/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0391 - val_loss: 0.0482 - lr: 0.0035\n",
            "Epoch 223/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0404 - val_loss: 0.0515 - lr: 0.0034\n",
            "Epoch 224/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0563 - val_loss: 0.1367 - lr: 0.0034\n",
            "Epoch 225/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0926 - val_loss: 0.1016 - lr: 0.0034\n",
            "Epoch 226/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0624 - val_loss: 0.0641 - lr: 0.0034\n",
            "Epoch 227/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0447 - val_loss: 0.0536 - lr: 0.0034\n",
            "Epoch 228/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0363 - val_loss: 0.0462 - lr: 0.0034\n",
            "Epoch 229/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0344 - val_loss: 0.0423 - lr: 0.0033\n",
            "Epoch 230/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0331 - val_loss: 0.0424 - lr: 0.0033\n",
            "Epoch 231/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0335 - val_loss: 0.0430 - lr: 0.0033\n",
            "Epoch 232/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0343 - val_loss: 0.0433 - lr: 0.0033\n",
            "Epoch 233/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0345 - val_loss: 0.0427 - lr: 0.0033\n",
            "Epoch 234/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0330 - val_loss: 0.0440 - lr: 0.0033\n",
            "Epoch 235/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0333 - val_loss: 0.0429 - lr: 0.0032\n",
            "Epoch 236/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0330 - val_loss: 0.0429 - lr: 0.0032\n",
            "Epoch 237/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0337 - val_loss: 0.0423 - lr: 0.0032\n",
            "Epoch 238/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0466 - val_loss: 0.0923 - lr: 0.0032\n",
            "Epoch 239/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0679 - val_loss: 0.0635 - lr: 0.0032\n",
            "Epoch 240/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0430 - val_loss: 0.0455 - lr: 0.0032\n",
            "Epoch 241/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0347 - val_loss: 0.0433 - lr: 0.0031\n",
            "Epoch 242/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0370 - val_loss: 0.0465 - lr: 0.0031\n",
            "Epoch 243/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0363 - val_loss: 0.0509 - lr: 0.0031\n",
            "Epoch 244/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0362 - val_loss: 0.0428 - lr: 0.0031\n",
            "Epoch 245/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0341 - val_loss: 0.0477 - lr: 0.0031\n",
            "Epoch 246/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0418 - val_loss: 0.0521 - lr: 0.0031\n",
            "Epoch 247/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0379 - val_loss: 0.0456 - lr: 0.0030\n",
            "Epoch 248/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0353 - val_loss: 0.0431 - lr: 0.0030\n",
            "Epoch 249/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0330 - val_loss: 0.0419 - lr: 0.0030\n",
            "Epoch 250/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0375 - val_loss: 0.0449 - lr: 0.0030\n",
            "Epoch 251/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0366 - val_loss: 0.0472 - lr: 0.0030\n",
            "Epoch 252/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0348 - val_loss: 0.0442 - lr: 0.0030\n",
            "Epoch 253/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0322 - val_loss: 0.0469 - lr: 0.0030\n",
            "Epoch 254/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0333 - val_loss: 0.0458 - lr: 0.0029\n",
            "Epoch 255/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0366 - val_loss: 0.0486 - lr: 0.0029\n",
            "Epoch 256/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0397 - val_loss: 0.0507 - lr: 0.0029\n",
            "Epoch 257/400\n",
            "64/64 [==============================] - 0s 7ms/step - loss: 0.0659 - val_loss: 0.0865 - lr: 0.0029\n",
            "Epoch 258/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0563 - val_loss: 0.0494 - lr: 0.0029\n",
            "Epoch 259/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0425 - val_loss: 0.0450 - lr: 0.0029\n",
            "Epoch 260/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0344 - val_loss: 0.0416 - lr: 0.0029\n",
            "Epoch 261/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0305 - val_loss: 0.0379 - lr: 0.0028\n",
            "Epoch 262/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0285 - val_loss: 0.0402 - lr: 0.0028\n",
            "Epoch 263/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0288 - val_loss: 0.0378 - lr: 0.0028\n",
            "Epoch 264/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0272 - val_loss: 0.0349 - lr: 0.0028\n",
            "Epoch 265/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0273 - val_loss: 0.0368 - lr: 0.0028\n",
            "Epoch 266/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0305 - val_loss: 0.0408 - lr: 0.0028\n",
            "Epoch 267/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0328 - val_loss: 0.0416 - lr: 0.0028\n",
            "Epoch 268/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0365 - val_loss: 0.0436 - lr: 0.0027\n",
            "Epoch 269/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0329 - val_loss: 0.0399 - lr: 0.0027\n",
            "Epoch 270/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0294 - val_loss: 0.0401 - lr: 0.0027\n",
            "Epoch 271/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0321 - val_loss: 0.0415 - lr: 0.0027\n",
            "Epoch 272/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0291 - val_loss: 0.0358 - lr: 0.0027\n",
            "Epoch 273/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0250 - val_loss: 0.0341 - lr: 0.0027\n",
            "Epoch 274/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0246 - val_loss: 0.0329 - lr: 0.0027\n",
            "Epoch 275/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0262 - val_loss: 0.0408 - lr: 0.0026\n",
            "Epoch 276/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0275 - val_loss: 0.0390 - lr: 0.0026\n",
            "Epoch 277/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0324 - val_loss: 0.0479 - lr: 0.0026\n",
            "Epoch 278/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0350 - val_loss: 0.0538 - lr: 0.0026\n",
            "Epoch 279/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0822 - val_loss: 0.1168 - lr: 0.0026\n",
            "Epoch 280/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0741 - val_loss: 0.0739 - lr: 0.0026\n",
            "Epoch 281/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0636 - val_loss: 0.0511 - lr: 0.0026\n",
            "Epoch 282/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0311 - val_loss: 0.0345 - lr: 0.0026\n",
            "Epoch 283/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0253 - val_loss: 0.0331 - lr: 0.0025\n",
            "Epoch 284/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0243 - val_loss: 0.0321 - lr: 0.0025\n",
            "Epoch 285/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0236 - val_loss: 0.0318 - lr: 0.0025\n",
            "Epoch 286/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0232 - val_loss: 0.0304 - lr: 0.0025\n",
            "Epoch 287/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0222 - val_loss: 0.0298 - lr: 0.0025\n",
            "Epoch 288/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0226 - val_loss: 0.0318 - lr: 0.0025\n",
            "Epoch 289/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0235 - val_loss: 0.0315 - lr: 0.0025\n",
            "Epoch 290/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0256 - val_loss: 0.0332 - lr: 0.0025\n",
            "Epoch 291/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0237 - val_loss: 0.0323 - lr: 0.0024\n",
            "Epoch 292/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0230 - val_loss: 0.0296 - lr: 0.0024\n",
            "Epoch 293/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0222 - val_loss: 0.0321 - lr: 0.0024\n",
            "Epoch 294/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0227 - val_loss: 0.0310 - lr: 0.0024\n",
            "Epoch 295/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0240 - val_loss: 0.0326 - lr: 0.0024\n",
            "Epoch 296/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0245 - val_loss: 0.0314 - lr: 0.0024\n",
            "Epoch 297/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0224 - val_loss: 0.0299 - lr: 0.0024\n",
            "Epoch 298/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0234 - val_loss: 0.0345 - lr: 0.0024\n",
            "Epoch 299/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0277 - val_loss: 0.0484 - lr: 0.0023\n",
            "Epoch 300/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0380 - val_loss: 0.0691 - lr: 0.0023\n",
            "Epoch 301/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0783 - val_loss: 0.1040 - lr: 0.0023\n",
            "Epoch 302/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0560 - val_loss: 0.0425 - lr: 0.0023\n",
            "Epoch 303/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0276 - val_loss: 0.0345 - lr: 0.0023\n",
            "Epoch 304/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0235 - val_loss: 0.0310 - lr: 0.0023\n",
            "Epoch 305/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0214 - val_loss: 0.0283 - lr: 0.0023\n",
            "Epoch 306/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0205 - val_loss: 0.0281 - lr: 0.0023\n",
            "Epoch 307/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0208 - val_loss: 0.0284 - lr: 0.0023\n",
            "Epoch 308/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0203 - val_loss: 0.0267 - lr: 0.0022\n",
            "Epoch 309/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0205 - val_loss: 0.0287 - lr: 0.0022\n",
            "Epoch 310/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0210 - val_loss: 0.0286 - lr: 0.0022\n",
            "Epoch 311/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0212 - val_loss: 0.0298 - lr: 0.0022\n",
            "Epoch 312/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0208 - val_loss: 0.0290 - lr: 0.0022\n",
            "Epoch 313/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0224 - val_loss: 0.0313 - lr: 0.0022\n",
            "Epoch 314/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0229 - val_loss: 0.0320 - lr: 0.0022\n",
            "Epoch 315/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0249 - val_loss: 0.0354 - lr: 0.0022\n",
            "Epoch 316/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0276 - val_loss: 0.0351 - lr: 0.0022\n",
            "Epoch 317/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0258 - val_loss: 0.0310 - lr: 0.0021\n",
            "Epoch 318/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0222 - val_loss: 0.0314 - lr: 0.0021\n",
            "Epoch 319/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0210 - val_loss: 0.0274 - lr: 0.0021\n",
            "Epoch 320/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0202 - val_loss: 0.0283 - lr: 0.0021\n",
            "Epoch 321/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0205 - val_loss: 0.0298 - lr: 0.0021\n",
            "Epoch 322/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0234 - val_loss: 0.0324 - lr: 0.0021\n",
            "Epoch 323/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0261 - val_loss: 0.0413 - lr: 0.0021\n",
            "Epoch 324/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0429 - val_loss: 0.0514 - lr: 0.0021\n",
            "Epoch 325/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0278 - val_loss: 0.0325 - lr: 0.0021\n",
            "Epoch 326/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0222 - val_loss: 0.0308 - lr: 0.0021\n",
            "Epoch 327/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0217 - val_loss: 0.0286 - lr: 0.0020\n",
            "Epoch 328/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0221 - val_loss: 0.0302 - lr: 0.0020\n",
            "Epoch 329/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0196 - val_loss: 0.0282 - lr: 0.0020\n",
            "Epoch 330/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0202 - val_loss: 0.0280 - lr: 0.0020\n",
            "Epoch 331/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0199 - val_loss: 0.0330 - lr: 0.0020\n",
            "Epoch 332/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0283 - val_loss: 0.0332 - lr: 0.0020\n",
            "Epoch 333/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0245 - val_loss: 0.0296 - lr: 0.0020\n",
            "Epoch 334/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0210 - val_loss: 0.0304 - lr: 0.0020\n",
            "Epoch 335/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0244 - val_loss: 0.0340 - lr: 0.0020\n",
            "Epoch 336/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0271 - val_loss: 0.0354 - lr: 0.0020\n",
            "Epoch 337/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0292 - val_loss: 0.0341 - lr: 0.0019\n",
            "Epoch 338/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0225 - val_loss: 0.0281 - lr: 0.0019\n",
            "Epoch 339/400\n",
            "64/64 [==============================] - 0s 8ms/step - loss: 0.0202 - val_loss: 0.0292 - lr: 0.0019\n",
            "Epoch 340/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0204 - val_loss: 0.0288 - lr: 0.0019\n",
            "Epoch 341/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0777 - val_loss: 0.1224 - lr: 0.0019\n",
            "Epoch 342/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0590 - val_loss: 0.0435 - lr: 0.0019\n",
            "Epoch 343/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0238 - val_loss: 0.0265 - lr: 0.0019\n",
            "Epoch 344/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0185 - val_loss: 0.0261 - lr: 0.0019\n",
            "Epoch 345/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0179 - val_loss: 0.0245 - lr: 0.0019\n",
            "Epoch 346/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0170 - val_loss: 0.0246 - lr: 0.0019\n",
            "Epoch 347/400\n",
            "64/64 [==============================] - 0s 7ms/step - loss: 0.0173 - val_loss: 0.0257 - lr: 0.0018\n",
            "Epoch 348/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0171 - val_loss: 0.0246 - lr: 0.0018\n",
            "Epoch 349/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0168 - val_loss: 0.0245 - lr: 0.0018\n",
            "Epoch 350/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0164 - val_loss: 0.0248 - lr: 0.0018\n",
            "Epoch 351/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0171 - val_loss: 0.0252 - lr: 0.0018\n",
            "Epoch 352/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0169 - val_loss: 0.0240 - lr: 0.0018\n",
            "Epoch 353/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0163 - val_loss: 0.0236 - lr: 0.0018\n",
            "Epoch 354/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0164 - val_loss: 0.0242 - lr: 0.0018\n",
            "Epoch 355/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0179 - val_loss: 0.0244 - lr: 0.0018\n",
            "Epoch 356/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0172 - val_loss: 0.0244 - lr: 0.0018\n",
            "Epoch 357/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0167 - val_loss: 0.0244 - lr: 0.0018\n",
            "Epoch 358/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0161 - val_loss: 0.0237 - lr: 0.0017\n",
            "Epoch 359/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0166 - val_loss: 0.0237 - lr: 0.0017\n",
            "Epoch 360/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0162 - val_loss: 0.0257 - lr: 0.0017\n",
            "Epoch 361/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0169 - val_loss: 0.0244 - lr: 0.0017\n",
            "Epoch 362/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0173 - val_loss: 0.0256 - lr: 0.0017\n",
            "Epoch 363/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0177 - val_loss: 0.0269 - lr: 0.0017\n",
            "Epoch 364/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0183 - val_loss: 0.0272 - lr: 0.0017\n",
            "Epoch 365/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0177 - val_loss: 0.0236 - lr: 0.0017\n",
            "Epoch 366/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0181 - val_loss: 0.0249 - lr: 0.0017\n",
            "Epoch 367/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0165 - val_loss: 0.0230 - lr: 0.0017\n",
            "Epoch 368/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0169 - val_loss: 0.0271 - lr: 0.0017\n",
            "Epoch 369/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0212 - val_loss: 0.0300 - lr: 0.0017\n",
            "Epoch 370/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0251 - val_loss: 0.0874 - lr: 0.0016\n",
            "Epoch 371/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0280 - val_loss: 0.0292 - lr: 0.0016\n",
            "Epoch 372/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0192 - val_loss: 0.0251 - lr: 0.0016\n",
            "Epoch 373/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0169 - val_loss: 0.0246 - lr: 0.0016\n",
            "Epoch 374/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0164 - val_loss: 0.0242 - lr: 0.0016\n",
            "Epoch 375/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0193 - val_loss: 0.0302 - lr: 0.0016\n",
            "Epoch 376/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0215 - val_loss: 0.0298 - lr: 0.0016\n",
            "Epoch 377/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0192 - val_loss: 0.0267 - lr: 0.0016\n",
            "Epoch 378/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0187 - val_loss: 0.0234 - lr: 0.0016\n",
            "Epoch 379/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0163 - val_loss: 0.0226 - lr: 0.0016\n",
            "Epoch 380/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0147 - val_loss: 0.0208 - lr: 0.0016\n",
            "Epoch 381/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0145 - val_loss: 0.0220 - lr: 0.0016\n",
            "Epoch 382/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0152 - val_loss: 0.0227 - lr: 0.0015\n",
            "Epoch 383/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0156 - val_loss: 0.0214 - lr: 0.0015\n",
            "Epoch 384/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0148 - val_loss: 0.0238 - lr: 0.0015\n",
            "Epoch 385/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0208 - val_loss: 0.0297 - lr: 0.0015\n",
            "Epoch 386/400\n",
            "64/64 [==============================] - 0s 7ms/step - loss: 0.0194 - val_loss: 0.0237 - lr: 0.0015\n",
            "Epoch 387/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0233 - val_loss: 0.0359 - lr: 0.0015\n",
            "Epoch 388/400\n",
            "64/64 [==============================] - 0s 7ms/step - loss: 0.0249 - val_loss: 0.0255 - lr: 0.0015\n",
            "Epoch 389/400\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0175 - val_loss: 0.0260 - lr: 0.0015\n",
            "Epoch 390/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0178 - val_loss: 0.0237 - lr: 0.0015\n",
            "Epoch 391/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0147 - val_loss: 0.0214 - lr: 0.0015\n",
            "Epoch 392/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0145 - val_loss: 0.0214 - lr: 0.0015\n",
            "Epoch 393/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0139 - val_loss: 0.0227 - lr: 0.0015\n",
            "Epoch 394/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0151 - val_loss: 0.0221 - lr: 0.0015\n",
            "Epoch 395/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0148 - val_loss: 0.0210 - lr: 0.0015\n",
            "Epoch 396/400\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0146 - val_loss: 0.0194 - lr: 0.0014\n",
            "Epoch 397/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0144 - val_loss: 0.0216 - lr: 0.0014\n",
            "Epoch 398/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0146 - val_loss: 0.0216 - lr: 0.0014\n",
            "Epoch 399/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0149 - val_loss: 0.0227 - lr: 0.0014\n",
            "Epoch 400/400\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0172 - val_loss: 0.0273 - lr: 0.0014\n"
          ]
        }
      ],
      "source": [
        "def scheduler(epoch, lr):\n",
        "    return lr if epoch < 10 else lr * 0.995\n",
        "\n",
        "sch = keras.callbacks.LearningRateScheduler(scheduler)\n",
        "opt = keras.optimizers.Adam(learning_rate=0.01)\n",
        "\n",
        "autoencoder.compile(optimizer=opt, loss='binary_crossentropy')\n",
        "history = autoencoder.fit(X_train, X_train,\n",
        "                epochs=400, callbacks=[sch],\n",
        "                batch_size=64,\n",
        "                shuffle=True,\n",
        "                validation_data=(X_test, X_test))"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "p--802Vqlo93"
      },
      "source": [
        "# AutoEncoder Plots"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 28,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "id": "wYZfUbhcoEKA",
        "outputId": "816782bc-ef3b-4bf2-c015-81423e61f568"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        }
      ],
      "source": [
        "# Plotting Loss vs Epoch for train and test sets\n",
        "x = [i + 1 for i in range(400)]\n",
        "plt.plot(x, history.history['loss'], color='r', label='train')\n",
        "plt.plot(x, history.history['val_loss'], color='b', label='test')\n",
        "plt.xlabel(\"Epoch\")\n",
        "plt.ylabel(\"Loss\")\n",
        "plt.title(\"Loss vs Epoch\")\n",
        "plt.legend()\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 29,
      "metadata": {
        "id": "ZMHo7DOOdqCw"
      },
      "outputs": [],
      "source": [
        "def show_scatter(encoder, X, y, title):\n",
        "    colors = ['blue', 'red']\n",
        "    for layer in encoder:\n",
        "        X = layer(X)\n",
        "    fig, ax = plt.subplots()\n",
        "    for i in (0, 1):\n",
        "        mask = (y == i)\n",
        "        x = X[mask]\n",
        "        label = 'YES' if i else 'NO'\n",
        "        ax.scatter(x[:, 0], x[:, 1], c=colors[i],s=2, label=label, edgecolors='none')\n",
        "    \n",
        "    ax.legend()\n",
        "    ax.grid(True)\n",
        "    plt.title(title)\n",
        "    plt.show()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 30,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 545
        },
        "id": "jiNIZEwNKSLn",
        "outputId": "01f23b53-5db7-41d9-ff98-6f24565d12e2"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        }
      ],
      "source": [
        "# Plotting the encoded data\n",
        "encoder = autoencoder.layers[:2]\n",
        "\n",
        "show_scatter(encoder, X_train, y_train, 'Train samples')\n",
        "show_scatter(encoder, X_test, y_test, 'Test samples')"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "nul40bhIqcpE"
      },
      "source": [
        "# The Classifier\n",
        "Now we get rid of the decoder part of our AutoEncder and replace it with a new MLP model ```head```."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 31,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "1TANc6sCY2ZO",
        "outputId": "d14ec10f-9b48-4fb2-e295-b7ab2ecb7412"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Model: \"sequential_1\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " Encoder (Sequential)        (None, 2)                 13090     \n",
            "                                                                 \n",
            " sequential (Sequential)     (None, 1)                 2769      \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 15,859\n",
            "Trainable params: 2,769\n",
            "Non-trainable params: 13,090\n",
            "_________________________________________________________________\n"
          ]
        }
      ],
      "source": [
        "head = keras.Sequential([\n",
        "    *[layers.Dense(dim, activation='relu') for dim in [16, 32, 64]],\n",
        "    layers.Dense(1, activation='sigmoid')\n",
        "])\n",
        "\n",
        "classifier = keras.Sequential([\n",
        "    *autoencoder.layers[:2],\n",
        "    head\n",
        "])\n",
        "classifier.layers[0].trainable = False\n",
        "classifier.summary()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 32,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "bbwO4YjqDg2R",
        "outputId": "3c0a7424-f270-4459-e749-9ab6fc9553ff"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 1/200\n",
            "64/64 [==============================] - 1s 7ms/step - loss: 0.2014 - val_loss: 0.1768 - lr: 0.0100\n",
            "Epoch 2/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1504 - val_loss: 0.1619 - lr: 0.0100\n",
            "Epoch 3/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1470 - val_loss: 0.1702 - lr: 0.0100\n",
            "Epoch 4/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1369 - val_loss: 0.1556 - lr: 0.0100\n",
            "Epoch 5/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1281 - val_loss: 0.1423 - lr: 0.0100\n",
            "Epoch 6/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1241 - val_loss: 0.1727 - lr: 0.0100\n",
            "Epoch 7/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1292 - val_loss: 0.1398 - lr: 0.0100\n",
            "Epoch 8/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1180 - val_loss: 0.1302 - lr: 0.0100\n",
            "Epoch 9/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1072 - val_loss: 0.1274 - lr: 0.0100\n",
            "Epoch 10/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1095 - val_loss: 0.1744 - lr: 0.0100\n",
            "Epoch 11/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1079 - val_loss: 0.1346 - lr: 0.0099\n",
            "Epoch 12/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.1055 - val_loss: 0.1248 - lr: 0.0099\n",
            "Epoch 13/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0942 - val_loss: 0.1193 - lr: 0.0099\n",
            "Epoch 14/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0929 - val_loss: 0.1077 - lr: 0.0098\n",
            "Epoch 15/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0895 - val_loss: 0.1086 - lr: 0.0098\n",
            "Epoch 16/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0859 - val_loss: 0.1033 - lr: 0.0097\n",
            "Epoch 17/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0904 - val_loss: 0.1154 - lr: 0.0097\n",
            "Epoch 18/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0876 - val_loss: 0.1175 - lr: 0.0096\n",
            "Epoch 19/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0828 - val_loss: 0.0969 - lr: 0.0096\n",
            "Epoch 20/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0809 - val_loss: 0.1134 - lr: 0.0095\n",
            "Epoch 21/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0817 - val_loss: 0.1173 - lr: 0.0095\n",
            "Epoch 22/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0853 - val_loss: 0.1236 - lr: 0.0094\n",
            "Epoch 23/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0757 - val_loss: 0.1154 - lr: 0.0094\n",
            "Epoch 24/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0745 - val_loss: 0.1023 - lr: 0.0093\n",
            "Epoch 25/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0729 - val_loss: 0.0989 - lr: 0.0093\n",
            "Epoch 26/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0817 - val_loss: 0.1205 - lr: 0.0092\n",
            "Epoch 27/200\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0754 - val_loss: 0.0946 - lr: 0.0092\n",
            "Epoch 28/200\n",
            "64/64 [==============================] - 0s 8ms/step - loss: 0.0644 - val_loss: 0.0911 - lr: 0.0091\n",
            "Epoch 29/200\n",
            "64/64 [==============================] - 0s 7ms/step - loss: 0.0707 - val_loss: 0.1060 - lr: 0.0091\n",
            "Epoch 30/200\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0696 - val_loss: 0.0932 - lr: 0.0090\n",
            "Epoch 31/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0655 - val_loss: 0.0929 - lr: 0.0090\n",
            "Epoch 32/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0677 - val_loss: 0.0926 - lr: 0.0090\n",
            "Epoch 33/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0649 - val_loss: 0.0897 - lr: 0.0089\n",
            "Epoch 34/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0677 - val_loss: 0.1015 - lr: 0.0089\n",
            "Epoch 35/200\n",
            "64/64 [==============================] - 1s 9ms/step - loss: 0.0616 - val_loss: 0.0920 - lr: 0.0088\n",
            "Epoch 36/200\n",
            "64/64 [==============================] - 1s 10ms/step - loss: 0.0602 - val_loss: 0.0996 - lr: 0.0088\n",
            "Epoch 37/200\n",
            "64/64 [==============================] - 1s 11ms/step - loss: 0.0687 - val_loss: 0.1090 - lr: 0.0087\n",
            "Epoch 38/200\n",
            "64/64 [==============================] - 1s 9ms/step - loss: 0.0613 - val_loss: 0.0812 - lr: 0.0087\n",
            "Epoch 39/200\n",
            "64/64 [==============================] - 1s 8ms/step - loss: 0.0606 - val_loss: 0.1001 - lr: 0.0086\n",
            "Epoch 40/200\n",
            "64/64 [==============================] - 1s 9ms/step - loss: 0.1047 - val_loss: 0.1129 - lr: 0.0086\n",
            "Epoch 41/200\n",
            "64/64 [==============================] - 0s 7ms/step - loss: 0.0782 - val_loss: 0.0926 - lr: 0.0086\n",
            "Epoch 42/200\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0637 - val_loss: 0.0932 - lr: 0.0085\n",
            "Epoch 43/200\n",
            "64/64 [==============================] - 1s 8ms/step - loss: 0.0619 - val_loss: 0.0979 - lr: 0.0085\n",
            "Epoch 44/200\n",
            "64/64 [==============================] - 0s 7ms/step - loss: 0.0678 - val_loss: 0.0929 - lr: 0.0084\n",
            "Epoch 45/200\n",
            "64/64 [==============================] - 0s 7ms/step - loss: 0.0622 - val_loss: 0.0878 - lr: 0.0084\n",
            "Epoch 46/200\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0606 - val_loss: 0.0798 - lr: 0.0083\n",
            "Epoch 47/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0643 - val_loss: 0.0831 - lr: 0.0083\n",
            "Epoch 48/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0564 - val_loss: 0.0831 - lr: 0.0083\n",
            "Epoch 49/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0567 - val_loss: 0.0790 - lr: 0.0082\n",
            "Epoch 50/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0533 - val_loss: 0.0860 - lr: 0.0082\n",
            "Epoch 51/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0533 - val_loss: 0.1004 - lr: 0.0081\n",
            "Epoch 52/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0570 - val_loss: 0.0792 - lr: 0.0081\n",
            "Epoch 53/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0566 - val_loss: 0.1090 - lr: 0.0081\n",
            "Epoch 54/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0662 - val_loss: 0.0807 - lr: 0.0080\n",
            "Epoch 55/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0552 - val_loss: 0.0797 - lr: 0.0080\n",
            "Epoch 56/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0498 - val_loss: 0.0756 - lr: 0.0079\n",
            "Epoch 57/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0545 - val_loss: 0.0924 - lr: 0.0079\n",
            "Epoch 58/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0517 - val_loss: 0.0789 - lr: 0.0079\n",
            "Epoch 59/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0533 - val_loss: 0.0826 - lr: 0.0078\n",
            "Epoch 60/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0571 - val_loss: 0.0856 - lr: 0.0078\n",
            "Epoch 61/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0577 - val_loss: 0.0814 - lr: 0.0077\n",
            "Epoch 62/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0603 - val_loss: 0.0931 - lr: 0.0077\n",
            "Epoch 63/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0585 - val_loss: 0.0793 - lr: 0.0077\n",
            "Epoch 64/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0549 - val_loss: 0.0820 - lr: 0.0076\n",
            "Epoch 65/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0509 - val_loss: 0.0910 - lr: 0.0076\n",
            "Epoch 66/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0510 - val_loss: 0.0711 - lr: 0.0076\n",
            "Epoch 67/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0506 - val_loss: 0.0768 - lr: 0.0075\n",
            "Epoch 68/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0490 - val_loss: 0.0818 - lr: 0.0075\n",
            "Epoch 69/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0515 - val_loss: 0.0765 - lr: 0.0074\n",
            "Epoch 70/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0503 - val_loss: 0.0755 - lr: 0.0074\n",
            "Epoch 71/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0658 - val_loss: 0.0820 - lr: 0.0074\n",
            "Epoch 72/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0527 - val_loss: 0.0805 - lr: 0.0073\n",
            "Epoch 73/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0518 - val_loss: 0.0885 - lr: 0.0073\n",
            "Epoch 74/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0560 - val_loss: 0.0960 - lr: 0.0073\n",
            "Epoch 75/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0505 - val_loss: 0.0714 - lr: 0.0072\n",
            "Epoch 76/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0464 - val_loss: 0.0767 - lr: 0.0072\n",
            "Epoch 77/200\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0541 - val_loss: 0.0891 - lr: 0.0071\n",
            "Epoch 78/200\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0649 - val_loss: 0.0809 - lr: 0.0071\n",
            "Epoch 79/200\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0496 - val_loss: 0.0816 - lr: 0.0071\n",
            "Epoch 80/200\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0484 - val_loss: 0.0736 - lr: 0.0070\n",
            "Epoch 81/200\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0463 - val_loss: 0.0803 - lr: 0.0070\n",
            "Epoch 82/200\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0447 - val_loss: 0.0702 - lr: 0.0070\n",
            "Epoch 83/200\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0456 - val_loss: 0.0732 - lr: 0.0069\n",
            "Epoch 84/200\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0470 - val_loss: 0.0717 - lr: 0.0069\n",
            "Epoch 85/200\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0464 - val_loss: 0.0678 - lr: 0.0069\n",
            "Epoch 86/200\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0461 - val_loss: 0.0746 - lr: 0.0068\n",
            "Epoch 87/200\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0464 - val_loss: 0.0724 - lr: 0.0068\n",
            "Epoch 88/200\n",
            "64/64 [==============================] - 0s 7ms/step - loss: 0.0490 - val_loss: 0.0664 - lr: 0.0068\n",
            "Epoch 89/200\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0468 - val_loss: 0.0669 - lr: 0.0067\n",
            "Epoch 90/200\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0449 - val_loss: 0.0682 - lr: 0.0067\n",
            "Epoch 91/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0438 - val_loss: 0.0758 - lr: 0.0067\n",
            "Epoch 92/200\n",
            "64/64 [==============================] - 0s 7ms/step - loss: 0.0422 - val_loss: 0.0778 - lr: 0.0066\n",
            "Epoch 93/200\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0517 - val_loss: 0.0808 - lr: 0.0066\n",
            "Epoch 94/200\n",
            "64/64 [==============================] - 1s 8ms/step - loss: 0.0497 - val_loss: 0.0737 - lr: 0.0066\n",
            "Epoch 95/200\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0463 - val_loss: 0.0713 - lr: 0.0065\n",
            "Epoch 96/200\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0453 - val_loss: 0.0718 - lr: 0.0065\n",
            "Epoch 97/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0445 - val_loss: 0.0677 - lr: 0.0065\n",
            "Epoch 98/200\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0445 - val_loss: 0.0746 - lr: 0.0064\n",
            "Epoch 99/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0422 - val_loss: 0.0905 - lr: 0.0064\n",
            "Epoch 100/200\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0511 - val_loss: 0.0778 - lr: 0.0064\n",
            "Epoch 101/200\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0873 - val_loss: 0.1015 - lr: 0.0063\n",
            "Epoch 102/200\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0577 - val_loss: 0.0820 - lr: 0.0063\n",
            "Epoch 103/200\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0471 - val_loss: 0.0684 - lr: 0.0063\n",
            "Epoch 104/200\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0433 - val_loss: 0.0690 - lr: 0.0062\n",
            "Epoch 105/200\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0438 - val_loss: 0.0836 - lr: 0.0062\n",
            "Epoch 106/200\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0562 - val_loss: 0.0814 - lr: 0.0062\n",
            "Epoch 107/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0444 - val_loss: 0.0700 - lr: 0.0061\n",
            "Epoch 108/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0440 - val_loss: 0.0692 - lr: 0.0061\n",
            "Epoch 109/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0447 - val_loss: 0.0723 - lr: 0.0061\n",
            "Epoch 110/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0434 - val_loss: 0.0693 - lr: 0.0061\n",
            "Epoch 111/200\n",
            "64/64 [==============================] - 0s 7ms/step - loss: 0.0425 - val_loss: 0.0665 - lr: 0.0060\n",
            "Epoch 112/200\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0419 - val_loss: 0.0788 - lr: 0.0060\n",
            "Epoch 113/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0467 - val_loss: 0.0707 - lr: 0.0060\n",
            "Epoch 114/200\n",
            "64/64 [==============================] - 0s 7ms/step - loss: 0.0441 - val_loss: 0.0716 - lr: 0.0059\n",
            "Epoch 115/200\n",
            "64/64 [==============================] - 0s 7ms/step - loss: 0.0450 - val_loss: 0.0666 - lr: 0.0059\n",
            "Epoch 116/200\n",
            "64/64 [==============================] - 1s 12ms/step - loss: 0.0413 - val_loss: 0.0821 - lr: 0.0059\n",
            "Epoch 117/200\n",
            "64/64 [==============================] - 1s 18ms/step - loss: 0.0476 - val_loss: 0.0659 - lr: 0.0058\n",
            "Epoch 118/200\n",
            "64/64 [==============================] - 1s 10ms/step - loss: 0.0446 - val_loss: 0.0716 - lr: 0.0058\n",
            "Epoch 119/200\n",
            "64/64 [==============================] - 1s 9ms/step - loss: 0.0407 - val_loss: 0.0752 - lr: 0.0058\n",
            "Epoch 120/200\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0455 - val_loss: 0.0769 - lr: 0.0058\n",
            "Epoch 121/200\n",
            "64/64 [==============================] - 0s 8ms/step - loss: 0.0490 - val_loss: 0.0726 - lr: 0.0057\n",
            "Epoch 122/200\n",
            "64/64 [==============================] - 1s 9ms/step - loss: 0.0463 - val_loss: 0.0702 - lr: 0.0057\n",
            "Epoch 123/200\n",
            "64/64 [==============================] - 1s 8ms/step - loss: 0.0432 - val_loss: 0.0664 - lr: 0.0057\n",
            "Epoch 124/200\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0414 - val_loss: 0.0676 - lr: 0.0056\n",
            "Epoch 125/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0465 - val_loss: 0.0687 - lr: 0.0056\n",
            "Epoch 126/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0430 - val_loss: 0.0728 - lr: 0.0056\n",
            "Epoch 127/200\n",
            "64/64 [==============================] - 0s 8ms/step - loss: 0.0398 - val_loss: 0.0757 - lr: 0.0056\n",
            "Epoch 128/200\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0425 - val_loss: 0.0730 - lr: 0.0055\n",
            "Epoch 129/200\n",
            "64/64 [==============================] - 0s 8ms/step - loss: 0.0397 - val_loss: 0.0670 - lr: 0.0055\n",
            "Epoch 130/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0388 - val_loss: 0.0637 - lr: 0.0055\n",
            "Epoch 131/200\n",
            "64/64 [==============================] - 0s 7ms/step - loss: 0.0393 - val_loss: 0.0651 - lr: 0.0055\n",
            "Epoch 132/200\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0380 - val_loss: 0.0739 - lr: 0.0054\n",
            "Epoch 133/200\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0450 - val_loss: 0.0866 - lr: 0.0054\n",
            "Epoch 134/200\n",
            "64/64 [==============================] - 0s 7ms/step - loss: 0.0455 - val_loss: 0.0664 - lr: 0.0054\n",
            "Epoch 135/200\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0412 - val_loss: 0.0644 - lr: 0.0053\n",
            "Epoch 136/200\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0417 - val_loss: 0.0661 - lr: 0.0053\n",
            "Epoch 137/200\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0383 - val_loss: 0.0639 - lr: 0.0053\n",
            "Epoch 138/200\n",
            "64/64 [==============================] - 0s 7ms/step - loss: 0.0530 - val_loss: 0.0850 - lr: 0.0053\n",
            "Epoch 139/200\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0436 - val_loss: 0.0683 - lr: 0.0052\n",
            "Epoch 140/200\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0394 - val_loss: 0.0694 - lr: 0.0052\n",
            "Epoch 141/200\n",
            "64/64 [==============================] - 0s 7ms/step - loss: 0.0382 - val_loss: 0.0669 - lr: 0.0052\n",
            "Epoch 142/200\n",
            "64/64 [==============================] - 0s 7ms/step - loss: 0.0378 - val_loss: 0.0676 - lr: 0.0052\n",
            "Epoch 143/200\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0354 - val_loss: 0.0664 - lr: 0.0051\n",
            "Epoch 144/200\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0358 - val_loss: 0.0667 - lr: 0.0051\n",
            "Epoch 145/200\n",
            "64/64 [==============================] - 1s 9ms/step - loss: 0.0356 - val_loss: 0.0651 - lr: 0.0051\n",
            "Epoch 146/200\n",
            "64/64 [==============================] - 1s 9ms/step - loss: 0.0357 - val_loss: 0.0695 - lr: 0.0051\n",
            "Epoch 147/200\n",
            "64/64 [==============================] - 0s 8ms/step - loss: 0.0378 - val_loss: 0.0654 - lr: 0.0050\n",
            "Epoch 148/200\n",
            "64/64 [==============================] - 1s 8ms/step - loss: 0.0366 - val_loss: 0.0680 - lr: 0.0050\n",
            "Epoch 149/200\n",
            "64/64 [==============================] - 1s 8ms/step - loss: 0.0352 - val_loss: 0.0915 - lr: 0.0050\n",
            "Epoch 150/200\n",
            "64/64 [==============================] - 0s 7ms/step - loss: 0.0399 - val_loss: 0.0694 - lr: 0.0050\n",
            "Epoch 151/200\n",
            "64/64 [==============================] - 0s 7ms/step - loss: 0.0382 - val_loss: 0.0688 - lr: 0.0049\n",
            "Epoch 152/200\n",
            "64/64 [==============================] - 1s 9ms/step - loss: 0.0367 - val_loss: 0.0783 - lr: 0.0049\n",
            "Epoch 153/200\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0397 - val_loss: 0.0710 - lr: 0.0049\n",
            "Epoch 154/200\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0368 - val_loss: 0.0725 - lr: 0.0049\n",
            "Epoch 155/200\n",
            "64/64 [==============================] - 0s 7ms/step - loss: 0.0375 - val_loss: 0.0644 - lr: 0.0048\n",
            "Epoch 156/200\n",
            "64/64 [==============================] - 0s 8ms/step - loss: 0.0360 - val_loss: 0.0642 - lr: 0.0048\n",
            "Epoch 157/200\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0350 - val_loss: 0.0669 - lr: 0.0048\n",
            "Epoch 158/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0360 - val_loss: 0.0615 - lr: 0.0048\n",
            "Epoch 159/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0368 - val_loss: 0.0855 - lr: 0.0047\n",
            "Epoch 160/200\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0437 - val_loss: 0.0661 - lr: 0.0047\n",
            "Epoch 161/200\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0388 - val_loss: 0.0705 - lr: 0.0047\n",
            "Epoch 162/200\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0353 - val_loss: 0.0669 - lr: 0.0047\n",
            "Epoch 163/200\n",
            "64/64 [==============================] - 0s 4ms/step - loss: 0.0366 - val_loss: 0.0674 - lr: 0.0046\n",
            "Epoch 164/200\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0357 - val_loss: 0.0665 - lr: 0.0046\n",
            "Epoch 165/200\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0363 - val_loss: 0.0742 - lr: 0.0046\n",
            "Epoch 166/200\n",
            "64/64 [==============================] - 0s 7ms/step - loss: 0.0390 - val_loss: 0.0790 - lr: 0.0046\n",
            "Epoch 167/200\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0373 - val_loss: 0.0641 - lr: 0.0046\n",
            "Epoch 168/200\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0358 - val_loss: 0.0666 - lr: 0.0045\n",
            "Epoch 169/200\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0361 - val_loss: 0.0651 - lr: 0.0045\n",
            "Epoch 170/200\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0353 - val_loss: 0.0780 - lr: 0.0045\n",
            "Epoch 171/200\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0361 - val_loss: 0.0659 - lr: 0.0045\n",
            "Epoch 172/200\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0352 - val_loss: 0.0650 - lr: 0.0044\n",
            "Epoch 173/200\n",
            "64/64 [==============================] - 1s 8ms/step - loss: 0.0356 - val_loss: 0.0670 - lr: 0.0044\n",
            "Epoch 174/200\n",
            "64/64 [==============================] - 0s 8ms/step - loss: 0.0358 - val_loss: 0.0674 - lr: 0.0044\n",
            "Epoch 175/200\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0356 - val_loss: 0.0668 - lr: 0.0044\n",
            "Epoch 176/200\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0369 - val_loss: 0.0674 - lr: 0.0044\n",
            "Epoch 177/200\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0357 - val_loss: 0.0676 - lr: 0.0043\n",
            "Epoch 178/200\n",
            "64/64 [==============================] - 1s 9ms/step - loss: 0.0352 - val_loss: 0.0646 - lr: 0.0043\n",
            "Epoch 179/200\n",
            "64/64 [==============================] - 0s 7ms/step - loss: 0.0346 - val_loss: 0.0640 - lr: 0.0043\n",
            "Epoch 180/200\n",
            "64/64 [==============================] - 0s 8ms/step - loss: 0.0351 - val_loss: 0.0638 - lr: 0.0043\n",
            "Epoch 181/200\n",
            "64/64 [==============================] - 0s 8ms/step - loss: 0.0393 - val_loss: 0.0690 - lr: 0.0042\n",
            "Epoch 182/200\n",
            "64/64 [==============================] - 1s 9ms/step - loss: 0.0363 - val_loss: 0.0726 - lr: 0.0042\n",
            "Epoch 183/200\n",
            "64/64 [==============================] - 0s 7ms/step - loss: 0.0352 - val_loss: 0.0727 - lr: 0.0042\n",
            "Epoch 184/200\n",
            "64/64 [==============================] - 1s 9ms/step - loss: 0.0359 - val_loss: 0.0690 - lr: 0.0042\n",
            "Epoch 185/200\n",
            "64/64 [==============================] - 0s 7ms/step - loss: 0.0363 - val_loss: 0.0643 - lr: 0.0042\n",
            "Epoch 186/200\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0353 - val_loss: 0.0642 - lr: 0.0041\n",
            "Epoch 187/200\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0341 - val_loss: 0.0633 - lr: 0.0041\n",
            "Epoch 188/200\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0361 - val_loss: 0.0741 - lr: 0.0041\n",
            "Epoch 189/200\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0355 - val_loss: 0.0680 - lr: 0.0041\n",
            "Epoch 190/200\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0339 - val_loss: 0.0666 - lr: 0.0041\n",
            "Epoch 191/200\n",
            "64/64 [==============================] - 0s 7ms/step - loss: 0.0342 - val_loss: 0.0700 - lr: 0.0040\n",
            "Epoch 192/200\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0348 - val_loss: 0.0729 - lr: 0.0040\n",
            "Epoch 193/200\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0378 - val_loss: 0.0685 - lr: 0.0040\n",
            "Epoch 194/200\n",
            "64/64 [==============================] - 0s 7ms/step - loss: 0.0351 - val_loss: 0.0669 - lr: 0.0040\n",
            "Epoch 195/200\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0348 - val_loss: 0.0660 - lr: 0.0040\n",
            "Epoch 196/200\n",
            "64/64 [==============================] - 1s 8ms/step - loss: 0.0342 - val_loss: 0.0655 - lr: 0.0039\n",
            "Epoch 197/200\n",
            "64/64 [==============================] - 0s 5ms/step - loss: 0.0360 - val_loss: 0.0650 - lr: 0.0039\n",
            "Epoch 198/200\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0343 - val_loss: 0.0664 - lr: 0.0039\n",
            "Epoch 199/200\n",
            "64/64 [==============================] - 0s 6ms/step - loss: 0.0341 - val_loss: 0.0650 - lr: 0.0039\n",
            "Epoch 200/200\n",
            "64/64 [==============================] - 1s 8ms/step - loss: 0.0347 - val_loss: 0.0658 - lr: 0.0039\n"
          ]
        }
      ],
      "source": [
        "sch = keras.callbacks.LearningRateScheduler(scheduler)\n",
        "opt = keras.optimizers.Adam(learning_rate=0.01)\n",
        "\n",
        "classifier.compile(optimizer=opt, loss='binary_crossentropy')\n",
        "classification_history = classifier.fit(X_train, y_train,\n",
        "                epochs=200, callbacks=[sch],\n",
        "                batch_size=64,\n",
        "                shuffle=True,\n",
        "                validation_data=(X_test, y_test))"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "qqMlpVXUu56I"
      },
      "source": [
        "# Plots (again)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 34,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "id": "vJAT4BtsutKk",
        "outputId": "411abe33-55c1-49fa-b9f4-2b4e395d3d48"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        }
      ],
      "source": [
        "# Plotting Loss vs Epoch for train and test sets\n",
        "x = [i + 1 for i in range(200)]\n",
        "plt.plot(x, classification_history.history['loss'], color='r', label='train')\n",
        "plt.plot(x, classification_history.history['val_loss'], color='b', label='test')\n",
        "plt.xlabel(\"Epoch\")\n",
        "plt.ylabel(\"Loss\")\n",
        "plt.title(\"Loss vs Epoch\")\n",
        "plt.legend()\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "EI-imcz4tU9a"
      },
      "source": [
        "# Choosing threshold\n",
        "Right now we have a model which outputs a number between 0 and 1. In order to make decision about weather or not the patient has the disease, we need to choose a threshold between 0 and 1. This threshold will then be used to classify samples as having or not having the disease."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 36,
      "metadata": {
        "id": "n3q5wQzgEfcm"
      },
      "outputs": [],
      "source": [
        "y_pred = np.array(classifier(X_test)).squeeze()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 42,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "y8bM7-VgG9_L",
        "outputId": "ad4b502e-42e5-400b-dc06-77d907c5373c"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "best threshold: 0.696\n"
          ]
        }
      ],
      "source": [
        "best_threshold = None\n",
        "best_f1 = 0\n",
        "for threshold in [0.3 + i/1000 for i in range(700)]:\n",
        "    pred = (y_pred > threshold).astype(int)\n",
        "    f1 = f1_score(pred, y_test)\n",
        "    if f1 > best_f1:\n",
        "        best_f1 = f1\n",
        "        best_threshold = threshold\n",
        "print(f'best threshold: {best_threshold}')"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "degEfv5-voWr"
      },
      "source": [
        "# Classification Report and Confusion Matrix"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 45,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 592
        },
        "id": "xkBQsfVVHtJD",
        "outputId": "7455d765-c29e-4192-e257-883fd17139e1"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "         0.0       0.91      1.00      0.95       261\n",
            "         1.0       1.00      0.98      0.99      1098\n",
            "\n",
            "    accuracy                           0.98      1359\n",
            "   macro avg       0.96      0.99      0.97      1359\n",
            "weighted avg       0.98      0.98      0.98      1359\n",
            "\n"
          ]
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 720x504 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        }
      ],
      "source": [
        "best_pred = (y_pred > best_threshold).astype(int)\n",
        "print(classification_report(y_test, best_pred))\n",
        "df_cm = pd.DataFrame(confusion_matrix(y_test, best_pred))\n",
        "plt.figure(figsize = (10,7))\n",
        "sn.heatmap(df_cm, annot=True)\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "MWZj_bchv4oh"
      },
      "source": [
        "# Conclusion\n",
        "In this notebook we achieved 98% accuracy and f1-score on the test set which is great."
      ]
    }
  ],
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "provenance": []
    },
    "gpuClass": "standard",
    "kernelspec": {
      "display_name": "Python 3 (ipykernel)",
      "language": "python",
      "name": "python3"
    },
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.9.7"
    },
    "vscode": {
      "interpreter": {
        "hash": "81794d4967e6c3204c66dcd87b604927b115b27c00565d3d43f05ba2f3a2cb0d"
      }
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}