Data Science and Machine Learning Bootcamp - JP/21. Neural Nets and Deep Learning/03_Tensorflow_Classification_Breast_Cancer_Wisconsin_(Diagnostic).ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "03.Tensorflow Classification - Breast Cancer Wisconsin (Diagnostic).ipynb",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "t9-AgTf41XqW"
      },
      "source": [
        "# Classification - Breast Cancer Wisconsin (Diagnostic)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "dBOz3AXi3jRB"
      },
      "source": [
        "[UCI - Breast Cancer Wisconsin (Diagnostic) Data Set](https://archive.ics.uci.edu/ml/datasets/Breast+Cancer+Wisconsin+(Diagnostic))\r\n",
        "\r\n",
        "\r\n",
        "**Data Set Characteristics:**\r\n",
        "\r\n",
        "    :Number of Instances: 569\r\n",
        "\r\n",
        "    :Number of Attributes: 30 numeric, predictive attributes and the class\r\n",
        "\r\n",
        "    :Attribute Information:\r\n",
        "        - radius (mean of distances from center to points on the perimeter)\r\n",
        "        - texture (standard deviation of gray-scale values)\r\n",
        "        - perimeter\r\n",
        "        - area\r\n",
        "        - smoothness (local variation in radius lengths)\r\n",
        "        - compactness (perimeter^2 / area - 1.0)\r\n",
        "        - concavity (severity of concave portions of the contour)\r\n",
        "        - concave points (number of concave portions of the contour)\r\n",
        "        - symmetry \r\n",
        "        - fractal dimension (\"coastline approximation\" - 1)\r\n",
        "\r\n",
        "        The mean, standard error, and \"worst\" or largest (mean of the three\r\n",
        "        largest values) of these features were computed for each image,\r\n",
        "        resulting in 30 features.  For instance, field 3 is Mean Radius, field\r\n",
        "        13 is Radius SE, field 23 is Worst Radius.\r\n",
        "\r\n",
        "        - class:\r\n",
        "                - WDBC-Malignant\r\n",
        "                - WDBC-Benign\r\n",
        "\r\n",
        "    :Summary Statistics:\r\n",
        "\r\n",
        "    ===================================== ====== ======\r\n",
        "                                           Min    Max\r\n",
        "    ===================================== ====== ======\r\n",
        "    radius (mean):                        6.981  28.11\r\n",
        "    texture (mean):                       9.71   39.28\r\n",
        "    perimeter (mean):                     43.79  188.5\r\n",
        "    area (mean):                          143.5  2501.0\r\n",
        "    smoothness (mean):                    0.053  0.163\r\n",
        "    compactness (mean):                   0.019  0.345\r\n",
        "    concavity (mean):                     0.0    0.427\r\n",
        "    concave points (mean):                0.0    0.201\r\n",
        "    symmetry (mean):                      0.106  0.304\r\n",
        "    fractal dimension (mean):             0.05   0.097\r\n",
        "    radius (standard error):              0.112  2.873\r\n",
        "    texture (standard error):             0.36   4.885\r\n",
        "    perimeter (standard error):           0.757  21.98\r\n",
        "    area (standard error):                6.802  542.2\r\n",
        "    smoothness (standard error):          0.002  0.031\r\n",
        "    compactness (standard error):         0.002  0.135\r\n",
        "    concavity (standard error):           0.0    0.396\r\n",
        "    concave points (standard error):      0.0    0.053\r\n",
        "    symmetry (standard error):            0.008  0.079\r\n",
        "    fractal dimension (standard error):   0.001  0.03\r\n",
        "    radius (worst):                       7.93   36.04\r\n",
        "    texture (worst):                      12.02  49.54\r\n",
        "    perimeter (worst):                    50.41  251.2\r\n",
        "    area (worst):                         185.2  4254.0\r\n",
        "    smoothness (worst):                   0.071  0.223\r\n",
        "    compactness (worst):                  0.027  1.058\r\n",
        "    concavity (worst):                    0.0    1.252\r\n",
        "    concave points (worst):               0.0    0.291\r\n",
        "    symmetry (worst):                     0.156  0.664\r\n",
        "    fractal dimension (worst):            0.055  0.208\r\n",
        "    ===================================== ====== ======\r\n",
        "\r\n",
        "    :Missing Attribute Values: None\r\n",
        "\r\n",
        "    :Class Distribution: 212 - Malignant, 357 - Benign\r\n",
        "\r\n",
        "    :Creator:  Dr. William H. Wolberg, W. Nick Street, Olvi L. Mangasarian\r\n",
        "\r\n",
        "    :Donor: Nick Street\r\n",
        "\r\n",
        "    :Date: November, 1995\r\n",
        "\r\n",
        "This is a copy of UCI ML Breast Cancer Wisconsin (Diagnostic) datasets.\r\n",
        "https://goo.gl/U2Uwz2\r\n",
        "\r\n",
        "Features are computed from a digitized image of a fine needle\r\n",
        "aspirate (FNA) of a breast mass.  They describe\r\n",
        "characteristics of the cell nuclei present in the image.\r\n",
        "\r\n",
        "Separating plane described above was obtained using\r\n",
        "Multisurface Method-Tree (MSM-T) [K. P. Bennett, \"Decision Tree\r\n",
        "Construction Via Linear Programming.\" Proceedings of the 4th\r\n",
        "Midwest Artificial Intelligence and Cognitive Science Society,\r\n",
        "pp. 97-101, 1992], a classification method which uses linear\r\n",
        "programming to construct a decision tree.  Relevant features\r\n",
        "were selected using an exhaustive search in the space of 1-4\r\n",
        "features and 1-3 separating planes.\r\n",
        "\r\n",
        "The actual linear program used to obtain the separating plane\r\n",
        "in the 3-dimensional space is that described in:\r\n",
        "[K. P. Bennett and O. L. Mangasarian: \"Robust Linear\r\n",
        "Programming Discrimination of Two Linearly Inseparable Sets\",\r\n",
        "Optimization Methods and Software 1, 1992, 23-34].\r\n",
        "\r\n",
        "This database is also available through the UW CS ftp server:\r\n",
        "\r\n",
        "ftp ftp.cs.wisc.edu\r\n",
        "cd math-prog/cpo-dataset/machine-learn/WDBC/\r\n",
        "\r\n",
        ".. topic:: References\r\n",
        "\r\n",
        "   - W.N. Street, W.H. Wolberg and O.L. Mangasarian. Nuclear feature extraction \r\n",
        "     for breast tumor diagnosis. IS&T/SPIE 1993 International Symposium on \r\n",
        "     Electronic Imaging: Science and Technology, volume 1905, pages 861-870,\r\n",
        "     San Jose, CA, 1993.\r\n",
        "   - O.L. Mangasarian, W.N. Street and W.H. Wolberg. Breast cancer diagnosis and \r\n",
        "     prognosis via linear programming. Operations Research, 43(4), pages 570-577, \r\n",
        "     July-August 1995.\r\n",
        "   - W.H. Wolberg, W.N. Street, and O.L. Mangasarian. Machine learning techniques\r\n",
        "     to diagnose breast cancer from fine-needle aspirates. Cancer Letters 77 (1994) \r\n",
        "     163-171.\r\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "0SCf7t8x4EOE"
      },
      "source": [
        "import numpy as np\r\n",
        "import pandas as pd\r\n",
        "import matplotlib.pyplot as plt\r\n",
        "import seaborn as sns"
      ],
      "execution_count": 1,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "VKfugrFC-X6g"
      },
      "source": [
        "folder_path = 'drive/MyDrive/TensorFlow-Data'"
      ],
      "execution_count": 2,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 258
        },
        "id": "cRQ3xsFI-vfp",
        "outputId": "14382af2-e5dd-4f24-f99c-b2983ef281e1"
      },
      "source": [
        "file_path = folder_path + '/cancer_classification.csv'\r\n",
        "\r\n",
        "df = pd.read_csv(file_path)\r\n",
        "df.head()"
      ],
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<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>mean radius</th>\n",
              "      <th>mean texture</th>\n",
              "      <th>mean perimeter</th>\n",
              "      <th>mean area</th>\n",
              "      <th>mean smoothness</th>\n",
              "      <th>mean compactness</th>\n",
              "      <th>mean concavity</th>\n",
              "      <th>mean concave points</th>\n",
              "      <th>mean symmetry</th>\n",
              "      <th>mean fractal dimension</th>\n",
              "      <th>radius error</th>\n",
              "      <th>texture error</th>\n",
              "      <th>perimeter error</th>\n",
              "      <th>area error</th>\n",
              "      <th>smoothness error</th>\n",
              "      <th>compactness error</th>\n",
              "      <th>concavity error</th>\n",
              "      <th>concave points error</th>\n",
              "      <th>symmetry error</th>\n",
              "      <th>fractal dimension error</th>\n",
              "      <th>worst radius</th>\n",
              "      <th>worst texture</th>\n",
              "      <th>worst perimeter</th>\n",
              "      <th>worst area</th>\n",
              "      <th>worst smoothness</th>\n",
              "      <th>worst compactness</th>\n",
              "      <th>worst concavity</th>\n",
              "      <th>worst concave points</th>\n",
              "      <th>worst symmetry</th>\n",
              "      <th>worst fractal dimension</th>\n",
              "      <th>benign_0__mal_1</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>17.99</td>\n",
              "      <td>10.38</td>\n",
              "      <td>122.80</td>\n",
              "      <td>1001.0</td>\n",
              "      <td>0.11840</td>\n",
              "      <td>0.27760</td>\n",
              "      <td>0.3001</td>\n",
              "      <td>0.14710</td>\n",
              "      <td>0.2419</td>\n",
              "      <td>0.07871</td>\n",
              "      <td>1.0950</td>\n",
              "      <td>0.9053</td>\n",
              "      <td>8.589</td>\n",
              "      <td>153.40</td>\n",
              "      <td>0.006399</td>\n",
              "      <td>0.04904</td>\n",
              "      <td>0.05373</td>\n",
              "      <td>0.01587</td>\n",
              "      <td>0.03003</td>\n",
              "      <td>0.006193</td>\n",
              "      <td>25.38</td>\n",
              "      <td>17.33</td>\n",
              "      <td>184.60</td>\n",
              "      <td>2019.0</td>\n",
              "      <td>0.1622</td>\n",
              "      <td>0.6656</td>\n",
              "      <td>0.7119</td>\n",
              "      <td>0.2654</td>\n",
              "      <td>0.4601</td>\n",
              "      <td>0.11890</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>20.57</td>\n",
              "      <td>17.77</td>\n",
              "      <td>132.90</td>\n",
              "      <td>1326.0</td>\n",
              "      <td>0.08474</td>\n",
              "      <td>0.07864</td>\n",
              "      <td>0.0869</td>\n",
              "      <td>0.07017</td>\n",
              "      <td>0.1812</td>\n",
              "      <td>0.05667</td>\n",
              "      <td>0.5435</td>\n",
              "      <td>0.7339</td>\n",
              "      <td>3.398</td>\n",
              "      <td>74.08</td>\n",
              "      <td>0.005225</td>\n",
              "      <td>0.01308</td>\n",
              "      <td>0.01860</td>\n",
              "      <td>0.01340</td>\n",
              "      <td>0.01389</td>\n",
              "      <td>0.003532</td>\n",
              "      <td>24.99</td>\n",
              "      <td>23.41</td>\n",
              "      <td>158.80</td>\n",
              "      <td>1956.0</td>\n",
              "      <td>0.1238</td>\n",
              "      <td>0.1866</td>\n",
              "      <td>0.2416</td>\n",
              "      <td>0.1860</td>\n",
              "      <td>0.2750</td>\n",
              "      <td>0.08902</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>19.69</td>\n",
              "      <td>21.25</td>\n",
              "      <td>130.00</td>\n",
              "      <td>1203.0</td>\n",
              "      <td>0.10960</td>\n",
              "      <td>0.15990</td>\n",
              "      <td>0.1974</td>\n",
              "      <td>0.12790</td>\n",
              "      <td>0.2069</td>\n",
              "      <td>0.05999</td>\n",
              "      <td>0.7456</td>\n",
              "      <td>0.7869</td>\n",
              "      <td>4.585</td>\n",
              "      <td>94.03</td>\n",
              "      <td>0.006150</td>\n",
              "      <td>0.04006</td>\n",
              "      <td>0.03832</td>\n",
              "      <td>0.02058</td>\n",
              "      <td>0.02250</td>\n",
              "      <td>0.004571</td>\n",
              "      <td>23.57</td>\n",
              "      <td>25.53</td>\n",
              "      <td>152.50</td>\n",
              "      <td>1709.0</td>\n",
              "      <td>0.1444</td>\n",
              "      <td>0.4245</td>\n",
              "      <td>0.4504</td>\n",
              "      <td>0.2430</td>\n",
              "      <td>0.3613</td>\n",
              "      <td>0.08758</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>11.42</td>\n",
              "      <td>20.38</td>\n",
              "      <td>77.58</td>\n",
              "      <td>386.1</td>\n",
              "      <td>0.14250</td>\n",
              "      <td>0.28390</td>\n",
              "      <td>0.2414</td>\n",
              "      <td>0.10520</td>\n",
              "      <td>0.2597</td>\n",
              "      <td>0.09744</td>\n",
              "      <td>0.4956</td>\n",
              "      <td>1.1560</td>\n",
              "      <td>3.445</td>\n",
              "      <td>27.23</td>\n",
              "      <td>0.009110</td>\n",
              "      <td>0.07458</td>\n",
              "      <td>0.05661</td>\n",
              "      <td>0.01867</td>\n",
              "      <td>0.05963</td>\n",
              "      <td>0.009208</td>\n",
              "      <td>14.91</td>\n",
              "      <td>26.50</td>\n",
              "      <td>98.87</td>\n",
              "      <td>567.7</td>\n",
              "      <td>0.2098</td>\n",
              "      <td>0.8663</td>\n",
              "      <td>0.6869</td>\n",
              "      <td>0.2575</td>\n",
              "      <td>0.6638</td>\n",
              "      <td>0.17300</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>20.29</td>\n",
              "      <td>14.34</td>\n",
              "      <td>135.10</td>\n",
              "      <td>1297.0</td>\n",
              "      <td>0.10030</td>\n",
              "      <td>0.13280</td>\n",
              "      <td>0.1980</td>\n",
              "      <td>0.10430</td>\n",
              "      <td>0.1809</td>\n",
              "      <td>0.05883</td>\n",
              "      <td>0.7572</td>\n",
              "      <td>0.7813</td>\n",
              "      <td>5.438</td>\n",
              "      <td>94.44</td>\n",
              "      <td>0.011490</td>\n",
              "      <td>0.02461</td>\n",
              "      <td>0.05688</td>\n",
              "      <td>0.01885</td>\n",
              "      <td>0.01756</td>\n",
              "      <td>0.005115</td>\n",
              "      <td>22.54</td>\n",
              "      <td>16.67</td>\n",
              "      <td>152.20</td>\n",
              "      <td>1575.0</td>\n",
              "      <td>0.1374</td>\n",
              "      <td>0.2050</td>\n",
              "      <td>0.4000</td>\n",
              "      <td>0.1625</td>\n",
              "      <td>0.2364</td>\n",
              "      <td>0.07678</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   mean radius  mean texture  ...  worst fractal dimension  benign_0__mal_1\n",
              "0        17.99         10.38  ...                  0.11890                0\n",
              "1        20.57         17.77  ...                  0.08902                0\n",
              "2        19.69         21.25  ...                  0.08758                0\n",
              "3        11.42         20.38  ...                  0.17300                0\n",
              "4        20.29         14.34  ...                  0.07678                0\n",
              "\n",
              "[5 rows x 31 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 3
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "5OsCtmNF-8tu"
      },
      "source": [
        "# Exploratory Data Analysis"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        },
        "id": "_6h5JbeU_Hn2",
        "outputId": "c0cfddd0-5672-42f2-e088-5ef8b0da4e38"
      },
      "source": [
        "df.info()"
      ],
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "<class 'pandas.core.frame.DataFrame'>\n",
            "RangeIndex: 569 entries, 0 to 568\n",
            "Data columns (total 31 columns):\n",
            " #   Column                   Non-Null Count  Dtype  \n",
            "---  ------                   --------------  -----  \n",
            " 0   mean radius              569 non-null    float64\n",
            " 1   mean texture             569 non-null    float64\n",
            " 2   mean perimeter           569 non-null    float64\n",
            " 3   mean area                569 non-null    float64\n",
            " 4   mean smoothness          569 non-null    float64\n",
            " 5   mean compactness         569 non-null    float64\n",
            " 6   mean concavity           569 non-null    float64\n",
            " 7   mean concave points      569 non-null    float64\n",
            " 8   mean symmetry            569 non-null    float64\n",
            " 9   mean fractal dimension   569 non-null    float64\n",
            " 10  radius error             569 non-null    float64\n",
            " 11  texture error            569 non-null    float64\n",
            " 12  perimeter error          569 non-null    float64\n",
            " 13  area error               569 non-null    float64\n",
            " 14  smoothness error         569 non-null    float64\n",
            " 15  compactness error        569 non-null    float64\n",
            " 16  concavity error          569 non-null    float64\n",
            " 17  concave points error     569 non-null    float64\n",
            " 18  symmetry error           569 non-null    float64\n",
            " 19  fractal dimension error  569 non-null    float64\n",
            " 20  worst radius             569 non-null    float64\n",
            " 21  worst texture            569 non-null    float64\n",
            " 22  worst perimeter          569 non-null    float64\n",
            " 23  worst area               569 non-null    float64\n",
            " 24  worst smoothness         569 non-null    float64\n",
            " 25  worst compactness        569 non-null    float64\n",
            " 26  worst concavity          569 non-null    float64\n",
            " 27  worst concave points     569 non-null    float64\n",
            " 28  worst symmetry           569 non-null    float64\n",
            " 29  worst fractal dimension  569 non-null    float64\n",
            " 30  benign_0__mal_1          569 non-null    int64  \n",
            "dtypes: float64(30), int64(1)\n",
            "memory usage: 137.9 KB\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1010
        },
        "id": "IMb2I9Bu_IiJ",
        "outputId": "b28e9a42-b33a-47b5-b848-01aedb57e02f"
      },
      "source": [
        "df.describe().T"
      ],
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<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>count</th>\n",
              "      <th>mean</th>\n",
              "      <th>std</th>\n",
              "      <th>min</th>\n",
              "      <th>25%</th>\n",
              "      <th>50%</th>\n",
              "      <th>75%</th>\n",
              "      <th>max</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>mean radius</th>\n",
              "      <td>569.0</td>\n",
              "      <td>14.127292</td>\n",
              "      <td>3.524049</td>\n",
              "      <td>6.981000</td>\n",
              "      <td>11.700000</td>\n",
              "      <td>13.370000</td>\n",
              "      <td>15.780000</td>\n",
              "      <td>28.11000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>mean texture</th>\n",
              "      <td>569.0</td>\n",
              "      <td>19.289649</td>\n",
              "      <td>4.301036</td>\n",
              "      <td>9.710000</td>\n",
              "      <td>16.170000</td>\n",
              "      <td>18.840000</td>\n",
              "      <td>21.800000</td>\n",
              "      <td>39.28000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>mean perimeter</th>\n",
              "      <td>569.0</td>\n",
              "      <td>91.969033</td>\n",
              "      <td>24.298981</td>\n",
              "      <td>43.790000</td>\n",
              "      <td>75.170000</td>\n",
              "      <td>86.240000</td>\n",
              "      <td>104.100000</td>\n",
              "      <td>188.50000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>mean area</th>\n",
              "      <td>569.0</td>\n",
              "      <td>654.889104</td>\n",
              "      <td>351.914129</td>\n",
              "      <td>143.500000</td>\n",
              "      <td>420.300000</td>\n",
              "      <td>551.100000</td>\n",
              "      <td>782.700000</td>\n",
              "      <td>2501.00000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>mean smoothness</th>\n",
              "      <td>569.0</td>\n",
              "      <td>0.096360</td>\n",
              "      <td>0.014064</td>\n",
              "      <td>0.052630</td>\n",
              "      <td>0.086370</td>\n",
              "      <td>0.095870</td>\n",
              "      <td>0.105300</td>\n",
              "      <td>0.16340</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>mean compactness</th>\n",
              "      <td>569.0</td>\n",
              "      <td>0.104341</td>\n",
              "      <td>0.052813</td>\n",
              "      <td>0.019380</td>\n",
              "      <td>0.064920</td>\n",
              "      <td>0.092630</td>\n",
              "      <td>0.130400</td>\n",
              "      <td>0.34540</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>mean concavity</th>\n",
              "      <td>569.0</td>\n",
              "      <td>0.088799</td>\n",
              "      <td>0.079720</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.029560</td>\n",
              "      <td>0.061540</td>\n",
              "      <td>0.130700</td>\n",
              "      <td>0.42680</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>mean concave points</th>\n",
              "      <td>569.0</td>\n",
              "      <td>0.048919</td>\n",
              "      <td>0.038803</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.020310</td>\n",
              "      <td>0.033500</td>\n",
              "      <td>0.074000</td>\n",
              "      <td>0.20120</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>mean symmetry</th>\n",
              "      <td>569.0</td>\n",
              "      <td>0.181162</td>\n",
              "      <td>0.027414</td>\n",
              "      <td>0.106000</td>\n",
              "      <td>0.161900</td>\n",
              "      <td>0.179200</td>\n",
              "      <td>0.195700</td>\n",
              "      <td>0.30400</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>mean fractal dimension</th>\n",
              "      <td>569.0</td>\n",
              "      <td>0.062798</td>\n",
              "      <td>0.007060</td>\n",
              "      <td>0.049960</td>\n",
              "      <td>0.057700</td>\n",
              "      <td>0.061540</td>\n",
              "      <td>0.066120</td>\n",
              "      <td>0.09744</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>radius error</th>\n",
              "      <td>569.0</td>\n",
              "      <td>0.405172</td>\n",
              "      <td>0.277313</td>\n",
              "      <td>0.111500</td>\n",
              "      <td>0.232400</td>\n",
              "      <td>0.324200</td>\n",
              "      <td>0.478900</td>\n",
              "      <td>2.87300</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>texture error</th>\n",
              "      <td>569.0</td>\n",
              "      <td>1.216853</td>\n",
              "      <td>0.551648</td>\n",
              "      <td>0.360200</td>\n",
              "      <td>0.833900</td>\n",
              "      <td>1.108000</td>\n",
              "      <td>1.474000</td>\n",
              "      <td>4.88500</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>perimeter error</th>\n",
              "      <td>569.0</td>\n",
              "      <td>2.866059</td>\n",
              "      <td>2.021855</td>\n",
              "      <td>0.757000</td>\n",
              "      <td>1.606000</td>\n",
              "      <td>2.287000</td>\n",
              "      <td>3.357000</td>\n",
              "      <td>21.98000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>area error</th>\n",
              "      <td>569.0</td>\n",
              "      <td>40.337079</td>\n",
              "      <td>45.491006</td>\n",
              "      <td>6.802000</td>\n",
              "      <td>17.850000</td>\n",
              "      <td>24.530000</td>\n",
              "      <td>45.190000</td>\n",
              "      <td>542.20000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>smoothness error</th>\n",
              "      <td>569.0</td>\n",
              "      <td>0.007041</td>\n",
              "      <td>0.003003</td>\n",
              "      <td>0.001713</td>\n",
              "      <td>0.005169</td>\n",
              "      <td>0.006380</td>\n",
              "      <td>0.008146</td>\n",
              "      <td>0.03113</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>compactness error</th>\n",
              "      <td>569.0</td>\n",
              "      <td>0.025478</td>\n",
              "      <td>0.017908</td>\n",
              "      <td>0.002252</td>\n",
              "      <td>0.013080</td>\n",
              "      <td>0.020450</td>\n",
              "      <td>0.032450</td>\n",
              "      <td>0.13540</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>concavity error</th>\n",
              "      <td>569.0</td>\n",
              "      <td>0.031894</td>\n",
              "      <td>0.030186</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.015090</td>\n",
              "      <td>0.025890</td>\n",
              "      <td>0.042050</td>\n",
              "      <td>0.39600</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>concave points error</th>\n",
              "      <td>569.0</td>\n",
              "      <td>0.011796</td>\n",
              "      <td>0.006170</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.007638</td>\n",
              "      <td>0.010930</td>\n",
              "      <td>0.014710</td>\n",
              "      <td>0.05279</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>symmetry error</th>\n",
              "      <td>569.0</td>\n",
              "      <td>0.020542</td>\n",
              "      <td>0.008266</td>\n",
              "      <td>0.007882</td>\n",
              "      <td>0.015160</td>\n",
              "      <td>0.018730</td>\n",
              "      <td>0.023480</td>\n",
              "      <td>0.07895</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>fractal dimension error</th>\n",
              "      <td>569.0</td>\n",
              "      <td>0.003795</td>\n",
              "      <td>0.002646</td>\n",
              "      <td>0.000895</td>\n",
              "      <td>0.002248</td>\n",
              "      <td>0.003187</td>\n",
              "      <td>0.004558</td>\n",
              "      <td>0.02984</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>worst radius</th>\n",
              "      <td>569.0</td>\n",
              "      <td>16.269190</td>\n",
              "      <td>4.833242</td>\n",
              "      <td>7.930000</td>\n",
              "      <td>13.010000</td>\n",
              "      <td>14.970000</td>\n",
              "      <td>18.790000</td>\n",
              "      <td>36.04000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>worst texture</th>\n",
              "      <td>569.0</td>\n",
              "      <td>25.677223</td>\n",
              "      <td>6.146258</td>\n",
              "      <td>12.020000</td>\n",
              "      <td>21.080000</td>\n",
              "      <td>25.410000</td>\n",
              "      <td>29.720000</td>\n",
              "      <td>49.54000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>worst perimeter</th>\n",
              "      <td>569.0</td>\n",
              "      <td>107.261213</td>\n",
              "      <td>33.602542</td>\n",
              "      <td>50.410000</td>\n",
              "      <td>84.110000</td>\n",
              "      <td>97.660000</td>\n",
              "      <td>125.400000</td>\n",
              "      <td>251.20000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>worst area</th>\n",
              "      <td>569.0</td>\n",
              "      <td>880.583128</td>\n",
              "      <td>569.356993</td>\n",
              "      <td>185.200000</td>\n",
              "      <td>515.300000</td>\n",
              "      <td>686.500000</td>\n",
              "      <td>1084.000000</td>\n",
              "      <td>4254.00000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>worst smoothness</th>\n",
              "      <td>569.0</td>\n",
              "      <td>0.132369</td>\n",
              "      <td>0.022832</td>\n",
              "      <td>0.071170</td>\n",
              "      <td>0.116600</td>\n",
              "      <td>0.131300</td>\n",
              "      <td>0.146000</td>\n",
              "      <td>0.22260</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>worst compactness</th>\n",
              "      <td>569.0</td>\n",
              "      <td>0.254265</td>\n",
              "      <td>0.157336</td>\n",
              "      <td>0.027290</td>\n",
              "      <td>0.147200</td>\n",
              "      <td>0.211900</td>\n",
              "      <td>0.339100</td>\n",
              "      <td>1.05800</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>worst concavity</th>\n",
              "      <td>569.0</td>\n",
              "      <td>0.272188</td>\n",
              "      <td>0.208624</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.114500</td>\n",
              "      <td>0.226700</td>\n",
              "      <td>0.382900</td>\n",
              "      <td>1.25200</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>worst concave points</th>\n",
              "      <td>569.0</td>\n",
              "      <td>0.114606</td>\n",
              "      <td>0.065732</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.064930</td>\n",
              "      <td>0.099930</td>\n",
              "      <td>0.161400</td>\n",
              "      <td>0.29100</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>worst symmetry</th>\n",
              "      <td>569.0</td>\n",
              "      <td>0.290076</td>\n",
              "      <td>0.061867</td>\n",
              "      <td>0.156500</td>\n",
              "      <td>0.250400</td>\n",
              "      <td>0.282200</td>\n",
              "      <td>0.317900</td>\n",
              "      <td>0.66380</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>worst fractal dimension</th>\n",
              "      <td>569.0</td>\n",
              "      <td>0.083946</td>\n",
              "      <td>0.018061</td>\n",
              "      <td>0.055040</td>\n",
              "      <td>0.071460</td>\n",
              "      <td>0.080040</td>\n",
              "      <td>0.092080</td>\n",
              "      <td>0.20750</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>benign_0__mal_1</th>\n",
              "      <td>569.0</td>\n",
              "      <td>0.627417</td>\n",
              "      <td>0.483918</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>1.000000</td>\n",
              "      <td>1.000000</td>\n",
              "      <td>1.00000</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "                         count        mean  ...          75%         max\n",
              "mean radius              569.0   14.127292  ...    15.780000    28.11000\n",
              "mean texture             569.0   19.289649  ...    21.800000    39.28000\n",
              "mean perimeter           569.0   91.969033  ...   104.100000   188.50000\n",
              "mean area                569.0  654.889104  ...   782.700000  2501.00000\n",
              "mean smoothness          569.0    0.096360  ...     0.105300     0.16340\n",
              "mean compactness         569.0    0.104341  ...     0.130400     0.34540\n",
              "mean concavity           569.0    0.088799  ...     0.130700     0.42680\n",
              "mean concave points      569.0    0.048919  ...     0.074000     0.20120\n",
              "mean symmetry            569.0    0.181162  ...     0.195700     0.30400\n",
              "mean fractal dimension   569.0    0.062798  ...     0.066120     0.09744\n",
              "radius error             569.0    0.405172  ...     0.478900     2.87300\n",
              "texture error            569.0    1.216853  ...     1.474000     4.88500\n",
              "perimeter error          569.0    2.866059  ...     3.357000    21.98000\n",
              "area error               569.0   40.337079  ...    45.190000   542.20000\n",
              "smoothness error         569.0    0.007041  ...     0.008146     0.03113\n",
              "compactness error        569.0    0.025478  ...     0.032450     0.13540\n",
              "concavity error          569.0    0.031894  ...     0.042050     0.39600\n",
              "concave points error     569.0    0.011796  ...     0.014710     0.05279\n",
              "symmetry error           569.0    0.020542  ...     0.023480     0.07895\n",
              "fractal dimension error  569.0    0.003795  ...     0.004558     0.02984\n",
              "worst radius             569.0   16.269190  ...    18.790000    36.04000\n",
              "worst texture            569.0   25.677223  ...    29.720000    49.54000\n",
              "worst perimeter          569.0  107.261213  ...   125.400000   251.20000\n",
              "worst area               569.0  880.583128  ...  1084.000000  4254.00000\n",
              "worst smoothness         569.0    0.132369  ...     0.146000     0.22260\n",
              "worst compactness        569.0    0.254265  ...     0.339100     1.05800\n",
              "worst concavity          569.0    0.272188  ...     0.382900     1.25200\n",
              "worst concave points     569.0    0.114606  ...     0.161400     0.29100\n",
              "worst symmetry           569.0    0.290076  ...     0.317900     0.66380\n",
              "worst fractal dimension  569.0    0.083946  ...     0.092080     0.20750\n",
              "benign_0__mal_1          569.0    0.627417  ...     1.000000     1.00000\n",
              "\n",
              "[31 rows x 8 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 7
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "WgJsVthg_eVn"
      },
      "source": [
        "## Check whether it is well balanced data or not\r\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        },
        "id": "9uZ35dRP_e-J",
        "outputId": "aa8aef30-cc6a-4b41-91b8-211719ab84f0"
      },
      "source": [
        "df['benign_0__mal_1'].value_counts()"
      ],
      "execution_count": 10,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "1    357\n",
              "0    212\n",
              "Name: benign_0__mal_1, dtype: int64"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 10
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        },
        "id": "e0MoZzUG_e7s",
        "outputId": "369c6073-8664-4481-c6c6-57b4c3f633ec"
      },
      "source": [
        "357/(357+212), 212/(357+212)"
      ],
      "execution_count": 13,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(0.6274165202108963, 0.37258347978910367)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 13
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 280
        },
        "id": "TIeKvgU2_e44",
        "outputId": "f4958530-ef99-4a8c-b7f9-49c99fbea867"
      },
      "source": [
        "sns.countplot(x='benign_0__mal_1', data=df);"
      ],
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "28rVlWCGAFe6"
      },
      "source": [
        "Although cases with 1 is more than 0 cases, the difference is not that extreme. "
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "NP7PB7qQAR_v"
      },
      "source": [
        "## Check correlation"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 591
        },
        "id": "797bsFDR_Mlw",
        "outputId": "57a0ee4d-bda0-4f17-dcf0-997f9177c86b"
      },
      "source": [
        "# check correlation\r\n",
        "plt.figure(figsize=(10, 8))\r\n",
        "df.corr()['benign_0__mal_1'][:-1].sort_values().plot(kind='bar');"
      ],
      "execution_count": 18,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 720x576 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ODdlsZy7_YWf"
      },
      "source": [
        "We can see that there are a lot of features which are strongly and negatively correlated wth label. So our model generally should do well picking up those patterns.\r\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 595
        },
        "id": "yKgNms2_A13a",
        "outputId": "6ea6d85c-2f45-4c12-a3d7-9af31ac0ed7f"
      },
      "source": [
        "plt.figure(figsize=(12,8))\r\n",
        "sns.heatmap(df.corr(), cmap='viridis_r');"
      ],
      "execution_count": 22,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 864x576 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "WsfX29_3BAiK"
      },
      "source": [
        "# Train Test Split\r\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "dMU9oDgTBYBp"
      },
      "source": [
        "# separate features and labels (in numpy arrray format)\r\n",
        "X = df.drop('benign_0__mal_1', axis=1).values\r\n",
        "y = df['benign_0__mal_1'].values"
      ],
      "execution_count": 25,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ZUI5FyOSBVh-"
      },
      "source": [
        "from sklearn.model_selection import train_test_split"
      ],
      "execution_count": 23,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "X4j-H22qBi-K"
      },
      "source": [
        "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.25, random_state=101)"
      ],
      "execution_count": 26,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "s4iI0u9bBz8z"
      },
      "source": [
        "# Scaling Data"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "t77-S0-eB4eF"
      },
      "source": [
        "from sklearn.preprocessing import MinMaxScaler\r\n",
        "\r\n",
        "scaler = MinMaxScaler()"
      ],
      "execution_count": 27,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "SYiJYq32B-Vt"
      },
      "source": [
        "X_train = scaler.fit_transform(X_train)\r\n",
        "X_test = scaler.transform(X_test)"
      ],
      "execution_count": 28,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "bszupC7ZCIva"
      },
      "source": [
        "# Model Building\r\n",
        "\r\n",
        "## For a binary classification problem\r\n",
        "`model.compile(optimizer='rmsprop',\r\n",
        "              loss='binary_crossentropy',\r\n",
        "              metrics=['accuracy'])`"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hWX1-CCUCR2K"
      },
      "source": [
        "from tensorflow.keras.models import Sequential\r\n",
        "from tensorflow.keras.layers import Dense"
      ],
      "execution_count": 32,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        },
        "id": "vpEZvzMOJbHk",
        "outputId": "b1bf5971-2515-431e-b3f0-635779fcd8ce"
      },
      "source": [
        "X.shape\r\n",
        "\r\n",
        "# we have 30 features"
      ],
      "execution_count": 35,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(569, 30)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 35
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "OKVYjLfxMOKa"
      },
      "source": [
        "## Example One: Choosing too many epochs and overfitting!"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bXEowCgNJLfM"
      },
      "source": [
        "model = Sequential()\r\n",
        "\r\n",
        "model.add(Dense(units=30, activation='relu'))\r\n",
        "model.add(Dense(units=15, activation='relu'))\r\n",
        "\r\n",
        "# As this is BINARY CLASSIFICATOIN problem, we want our last layer to be using Sigmoid \r\n",
        "model.add(Dense(units=1, activation='sigmoid')) # output layer with Sigmoid Activation Function\r\n",
        "\r\n",
        "\r\n",
        "model.compile(optimizer='adam', loss='binary_crossentropy')"
      ],
      "execution_count": 38,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "LB_XatYHI__-"
      },
      "source": [
        "## Training the Model\r\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "LytjG-bAKukS",
        "outputId": "bf7878ac-cfec-4067-d7a3-4e6206aea4d6"
      },
      "source": [
        "model.fit(x=X_train, y=y_train, epochs=600, validation_data=(X_test, y_test), verbose=1)"
      ],
      "execution_count": 39,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/600\n",
            "14/14 [==============================] - 1s 21ms/step - loss: 0.6788 - val_loss: 0.6332\n",
            "Epoch 2/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.6195 - val_loss: 0.5743\n",
            "Epoch 3/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.5508 - val_loss: 0.5067\n",
            "Epoch 4/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.4904 - val_loss: 0.4500\n",
            "Epoch 5/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.4279 - val_loss: 0.3969\n",
            "Epoch 6/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.3899 - val_loss: 0.3529\n",
            "Epoch 7/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.3531 - val_loss: 0.3120\n",
            "Epoch 8/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.3182 - val_loss: 0.2759\n",
            "Epoch 9/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.2785 - val_loss: 0.2475\n",
            "Epoch 10/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.2684 - val_loss: 0.2253\n",
            "Epoch 11/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.2677 - val_loss: 0.2048\n",
            "Epoch 12/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.2168 - val_loss: 0.1957\n",
            "Epoch 13/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.2345 - val_loss: 0.1762\n",
            "Epoch 14/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.2045 - val_loss: 0.1686\n",
            "Epoch 15/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1793 - val_loss: 0.1603\n",
            "Epoch 16/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1841 - val_loss: 0.1494\n",
            "Epoch 17/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1642 - val_loss: 0.1427\n",
            "Epoch 18/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.1632 - val_loss: 0.1374\n",
            "Epoch 19/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.1406 - val_loss: 0.1313\n",
            "Epoch 20/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.1454 - val_loss: 0.1258\n",
            "Epoch 21/600\n",
            "14/14 [==============================] - 0s 14ms/step - loss: 0.1547 - val_loss: 0.1216\n",
            "Epoch 22/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1076 - val_loss: 0.1225\n",
            "Epoch 23/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.1208 - val_loss: 0.1150\n",
            "Epoch 24/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1174 - val_loss: 0.1130\n",
            "Epoch 25/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1076 - val_loss: 0.1146\n",
            "Epoch 26/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1222 - val_loss: 0.1079\n",
            "Epoch 27/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.1209 - val_loss: 0.1091\n",
            "Epoch 28/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0979 - val_loss: 0.1061\n",
            "Epoch 29/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0893 - val_loss: 0.1040\n",
            "Epoch 30/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0921 - val_loss: 0.1032\n",
            "Epoch 31/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0963 - val_loss: 0.1025\n",
            "Epoch 32/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0810 - val_loss: 0.1100\n",
            "Epoch 33/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1049 - val_loss: 0.0992\n",
            "Epoch 34/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0982 - val_loss: 0.0982\n",
            "Epoch 35/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0796 - val_loss: 0.1041\n",
            "Epoch 36/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0792 - val_loss: 0.0989\n",
            "Epoch 37/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0823 - val_loss: 0.0982\n",
            "Epoch 38/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0642 - val_loss: 0.0996\n",
            "Epoch 39/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0760 - val_loss: 0.0978\n",
            "Epoch 40/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0724 - val_loss: 0.0993\n",
            "Epoch 41/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0676 - val_loss: 0.0990\n",
            "Epoch 42/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0662 - val_loss: 0.0982\n",
            "Epoch 43/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0591 - val_loss: 0.1036\n",
            "Epoch 44/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0620 - val_loss: 0.0945\n",
            "Epoch 45/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0740 - val_loss: 0.0960\n",
            "Epoch 46/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0570 - val_loss: 0.1015\n",
            "Epoch 47/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0660 - val_loss: 0.0983\n",
            "Epoch 48/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0555 - val_loss: 0.1028\n",
            "Epoch 49/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0714 - val_loss: 0.0970\n",
            "Epoch 50/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0681 - val_loss: 0.0999\n",
            "Epoch 51/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0677 - val_loss: 0.0946\n",
            "Epoch 52/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0579 - val_loss: 0.1023\n",
            "Epoch 53/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0601 - val_loss: 0.0975\n",
            "Epoch 54/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0617 - val_loss: 0.0985\n",
            "Epoch 55/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0615 - val_loss: 0.0982\n",
            "Epoch 56/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0859 - val_loss: 0.0963\n",
            "Epoch 57/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0738 - val_loss: 0.0974\n",
            "Epoch 58/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0474 - val_loss: 0.1021\n",
            "Epoch 59/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0512 - val_loss: 0.0969\n",
            "Epoch 60/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0504 - val_loss: 0.1018\n",
            "Epoch 61/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0537 - val_loss: 0.1034\n",
            "Epoch 62/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0560 - val_loss: 0.0978\n",
            "Epoch 63/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0429 - val_loss: 0.1038\n",
            "Epoch 64/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0604 - val_loss: 0.0968\n",
            "Epoch 65/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0560 - val_loss: 0.1091\n",
            "Epoch 66/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0791 - val_loss: 0.0957\n",
            "Epoch 67/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0576 - val_loss: 0.1067\n",
            "Epoch 68/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0557 - val_loss: 0.0966\n",
            "Epoch 69/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0522 - val_loss: 0.1069\n",
            "Epoch 70/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0450 - val_loss: 0.1013\n",
            "Epoch 71/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0634 - val_loss: 0.1022\n",
            "Epoch 72/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0548 - val_loss: 0.1038\n",
            "Epoch 73/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0731 - val_loss: 0.1002\n",
            "Epoch 74/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0509 - val_loss: 0.1018\n",
            "Epoch 75/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0595 - val_loss: 0.1030\n",
            "Epoch 76/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0469 - val_loss: 0.1048\n",
            "Epoch 77/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0468 - val_loss: 0.1029\n",
            "Epoch 78/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0510 - val_loss: 0.1003\n",
            "Epoch 79/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0430 - val_loss: 0.1053\n",
            "Epoch 80/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0503 - val_loss: 0.0993\n",
            "Epoch 81/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0360 - val_loss: 0.1094\n",
            "Epoch 82/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0383 - val_loss: 0.1090\n",
            "Epoch 83/600\n",
            "14/14 [==============================] - 0s 15ms/step - loss: 0.0402 - val_loss: 0.1029\n",
            "Epoch 84/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0413 - val_loss: 0.1053\n",
            "Epoch 85/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0434 - val_loss: 0.1011\n",
            "Epoch 86/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0438 - val_loss: 0.1087\n",
            "Epoch 87/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0483 - val_loss: 0.1031\n",
            "Epoch 88/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0375 - val_loss: 0.1111\n",
            "Epoch 89/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0552 - val_loss: 0.0989\n",
            "Epoch 90/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0320 - val_loss: 0.1139\n",
            "Epoch 91/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0382 - val_loss: 0.1023\n",
            "Epoch 92/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0383 - val_loss: 0.1121\n",
            "Epoch 93/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0497 - val_loss: 0.1052\n",
            "Epoch 94/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0482 - val_loss: 0.1096\n",
            "Epoch 95/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0357 - val_loss: 0.1087\n",
            "Epoch 96/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0393 - val_loss: 0.1012\n",
            "Epoch 97/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0404 - val_loss: 0.1171\n",
            "Epoch 98/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0428 - val_loss: 0.1009\n",
            "Epoch 99/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0569 - val_loss: 0.1015\n",
            "Epoch 100/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0397 - val_loss: 0.1095\n",
            "Epoch 101/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0407 - val_loss: 0.1051\n",
            "Epoch 102/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0399 - val_loss: 0.1045\n",
            "Epoch 103/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0516 - val_loss: 0.1088\n",
            "Epoch 104/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0366 - val_loss: 0.1110\n",
            "Epoch 105/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0372 - val_loss: 0.1075\n",
            "Epoch 106/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0438 - val_loss: 0.1105\n",
            "Epoch 107/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0406 - val_loss: 0.1127\n",
            "Epoch 108/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0374 - val_loss: 0.1174\n",
            "Epoch 109/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0550 - val_loss: 0.1068\n",
            "Epoch 110/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0366 - val_loss: 0.1165\n",
            "Epoch 111/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0400 - val_loss: 0.1136\n",
            "Epoch 112/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0380 - val_loss: 0.1102\n",
            "Epoch 113/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0332 - val_loss: 0.1088\n",
            "Epoch 114/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0297 - val_loss: 0.1120\n",
            "Epoch 115/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0345 - val_loss: 0.1154\n",
            "Epoch 116/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0378 - val_loss: 0.1060\n",
            "Epoch 117/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0404 - val_loss: 0.1206\n",
            "Epoch 118/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0389 - val_loss: 0.1115\n",
            "Epoch 119/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0348 - val_loss: 0.1085\n",
            "Epoch 120/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0335 - val_loss: 0.1171\n",
            "Epoch 121/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0410 - val_loss: 0.1047\n",
            "Epoch 122/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0268 - val_loss: 0.1191\n",
            "Epoch 123/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0295 - val_loss: 0.1074\n",
            "Epoch 124/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0315 - val_loss: 0.1221\n",
            "Epoch 125/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0325 - val_loss: 0.1133\n",
            "Epoch 126/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0346 - val_loss: 0.1157\n",
            "Epoch 127/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0343 - val_loss: 0.1090\n",
            "Epoch 128/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0331 - val_loss: 0.1150\n",
            "Epoch 129/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0313 - val_loss: 0.1104\n",
            "Epoch 130/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0411 - val_loss: 0.1105\n",
            "Epoch 131/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0359 - val_loss: 0.1161\n",
            "Epoch 132/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0453 - val_loss: 0.1073\n",
            "Epoch 133/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0338 - val_loss: 0.1178\n",
            "Epoch 134/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0441 - val_loss: 0.1162\n",
            "Epoch 135/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0246 - val_loss: 0.1158\n",
            "Epoch 136/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0376 - val_loss: 0.1151\n",
            "Epoch 137/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0381 - val_loss: 0.1124\n",
            "Epoch 138/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0411 - val_loss: 0.1213\n",
            "Epoch 139/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0282 - val_loss: 0.1189\n",
            "Epoch 140/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0424 - val_loss: 0.1131\n",
            "Epoch 141/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0272 - val_loss: 0.1196\n",
            "Epoch 142/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0274 - val_loss: 0.1155\n",
            "Epoch 143/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0376 - val_loss: 0.1086\n",
            "Epoch 144/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0304 - val_loss: 0.1235\n",
            "Epoch 145/600\n",
            "14/14 [==============================] - 0s 15ms/step - loss: 0.0335 - val_loss: 0.1020\n",
            "Epoch 146/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0375 - val_loss: 0.1299\n",
            "Epoch 147/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0259 - val_loss: 0.1074\n",
            "Epoch 148/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0424 - val_loss: 0.1145\n",
            "Epoch 149/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0332 - val_loss: 0.1170\n",
            "Epoch 150/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0328 - val_loss: 0.1118\n",
            "Epoch 151/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0431 - val_loss: 0.1311\n",
            "Epoch 152/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0432 - val_loss: 0.1031\n",
            "Epoch 153/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0423 - val_loss: 0.1219\n",
            "Epoch 154/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0281 - val_loss: 0.1160\n",
            "Epoch 155/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0352 - val_loss: 0.1378\n",
            "Epoch 156/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0427 - val_loss: 0.1102\n",
            "Epoch 157/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0256 - val_loss: 0.1340\n",
            "Epoch 158/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0327 - val_loss: 0.1119\n",
            "Epoch 159/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0204 - val_loss: 0.1269\n",
            "Epoch 160/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0426 - val_loss: 0.1103\n",
            "Epoch 161/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0335 - val_loss: 0.1209\n",
            "Epoch 162/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0255 - val_loss: 0.1196\n",
            "Epoch 163/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0291 - val_loss: 0.1111\n",
            "Epoch 164/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0293 - val_loss: 0.1334\n",
            "Epoch 165/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0255 - val_loss: 0.1108\n",
            "Epoch 166/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0260 - val_loss: 0.1269\n",
            "Epoch 167/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0262 - val_loss: 0.1093\n",
            "Epoch 168/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0283 - val_loss: 0.1275\n",
            "Epoch 169/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0347 - val_loss: 0.1102\n",
            "Epoch 170/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0386 - val_loss: 0.1175\n",
            "Epoch 171/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0193 - val_loss: 0.1264\n",
            "Epoch 172/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0224 - val_loss: 0.1177\n",
            "Epoch 173/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0301 - val_loss: 0.1244\n",
            "Epoch 174/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0327 - val_loss: 0.1223\n",
            "Epoch 175/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0251 - val_loss: 0.1167\n",
            "Epoch 176/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0400 - val_loss: 0.1172\n",
            "Epoch 177/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0186 - val_loss: 0.1353\n",
            "Epoch 178/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0282 - val_loss: 0.1120\n",
            "Epoch 179/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0348 - val_loss: 0.1189\n",
            "Epoch 180/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0190 - val_loss: 0.1213\n",
            "Epoch 181/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0291 - val_loss: 0.1119\n",
            "Epoch 182/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0312 - val_loss: 0.1203\n",
            "Epoch 183/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0338 - val_loss: 0.1088\n",
            "Epoch 184/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0322 - val_loss: 0.1172\n",
            "Epoch 185/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0372 - val_loss: 0.1209\n",
            "Epoch 186/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0238 - val_loss: 0.1325\n",
            "Epoch 187/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0254 - val_loss: 0.1207\n",
            "Epoch 188/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0327 - val_loss: 0.1196\n",
            "Epoch 189/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0350 - val_loss: 0.1106\n",
            "Epoch 190/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0356 - val_loss: 0.1361\n",
            "Epoch 191/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0262 - val_loss: 0.1093\n",
            "Epoch 192/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0236 - val_loss: 0.1418\n",
            "Epoch 193/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0339 - val_loss: 0.1132\n",
            "Epoch 194/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0251 - val_loss: 0.1310\n",
            "Epoch 195/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0298 - val_loss: 0.1165\n",
            "Epoch 196/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0236 - val_loss: 0.1239\n",
            "Epoch 197/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0227 - val_loss: 0.1154\n",
            "Epoch 198/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0333 - val_loss: 0.1164\n",
            "Epoch 199/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0247 - val_loss: 0.1356\n",
            "Epoch 200/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0269 - val_loss: 0.1087\n",
            "Epoch 201/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0264 - val_loss: 0.1497\n",
            "Epoch 202/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0224 - val_loss: 0.1087\n",
            "Epoch 203/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0317 - val_loss: 0.1432\n",
            "Epoch 204/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0302 - val_loss: 0.1154\n",
            "Epoch 205/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0202 - val_loss: 0.1399\n",
            "Epoch 206/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0314 - val_loss: 0.1154\n",
            "Epoch 207/600\n",
            "14/14 [==============================] - 0s 15ms/step - loss: 0.0254 - val_loss: 0.1299\n",
            "Epoch 208/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0230 - val_loss: 0.1206\n",
            "Epoch 209/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0185 - val_loss: 0.1281\n",
            "Epoch 210/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0223 - val_loss: 0.1253\n",
            "Epoch 211/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0325 - val_loss: 0.1239\n",
            "Epoch 212/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0240 - val_loss: 0.1344\n",
            "Epoch 213/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0268 - val_loss: 0.1245\n",
            "Epoch 214/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0165 - val_loss: 0.1227\n",
            "Epoch 215/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0196 - val_loss: 0.1325\n",
            "Epoch 216/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0201 - val_loss: 0.1314\n",
            "Epoch 217/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0226 - val_loss: 0.1128\n",
            "Epoch 218/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0182 - val_loss: 0.1509\n",
            "Epoch 219/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0264 - val_loss: 0.1184\n",
            "Epoch 220/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0205 - val_loss: 0.1530\n",
            "Epoch 221/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0259 - val_loss: 0.1228\n",
            "Epoch 222/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0191 - val_loss: 0.1369\n",
            "Epoch 223/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0314 - val_loss: 0.1120\n",
            "Epoch 224/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0227 - val_loss: 0.1405\n",
            "Epoch 225/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0214 - val_loss: 0.1263\n",
            "Epoch 226/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0257 - val_loss: 0.1159\n",
            "Epoch 227/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0201 - val_loss: 0.1275\n",
            "Epoch 228/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0204 - val_loss: 0.1211\n",
            "Epoch 229/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0163 - val_loss: 0.1175\n",
            "Epoch 230/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0216 - val_loss: 0.1308\n",
            "Epoch 231/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0241 - val_loss: 0.1217\n",
            "Epoch 232/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0188 - val_loss: 0.1317\n",
            "Epoch 233/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0254 - val_loss: 0.1196\n",
            "Epoch 234/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0259 - val_loss: 0.1541\n",
            "Epoch 235/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0224 - val_loss: 0.1096\n",
            "Epoch 236/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0187 - val_loss: 0.1440\n",
            "Epoch 237/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0232 - val_loss: 0.1241\n",
            "Epoch 238/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0272 - val_loss: 0.1231\n",
            "Epoch 239/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0173 - val_loss: 0.1515\n",
            "Epoch 240/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0292 - val_loss: 0.1110\n",
            "Epoch 241/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0243 - val_loss: 0.1379\n",
            "Epoch 242/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0179 - val_loss: 0.1213\n",
            "Epoch 243/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0301 - val_loss: 0.1245\n",
            "Epoch 244/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0233 - val_loss: 0.1282\n",
            "Epoch 245/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0176 - val_loss: 0.1356\n",
            "Epoch 246/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0216 - val_loss: 0.1134\n",
            "Epoch 247/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0210 - val_loss: 0.1697\n",
            "Epoch 248/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0272 - val_loss: 0.1254\n",
            "Epoch 249/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0170 - val_loss: 0.1292\n",
            "Epoch 250/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0209 - val_loss: 0.1340\n",
            "Epoch 251/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0163 - val_loss: 0.1289\n",
            "Epoch 252/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0158 - val_loss: 0.1294\n",
            "Epoch 253/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0197 - val_loss: 0.1246\n",
            "Epoch 254/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0137 - val_loss: 0.1443\n",
            "Epoch 255/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0186 - val_loss: 0.1320\n",
            "Epoch 256/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0177 - val_loss: 0.1155\n",
            "Epoch 257/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0171 - val_loss: 0.1417\n",
            "Epoch 258/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0183 - val_loss: 0.1150\n",
            "Epoch 259/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0151 - val_loss: 0.1643\n",
            "Epoch 260/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0193 - val_loss: 0.1155\n",
            "Epoch 261/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0238 - val_loss: 0.1354\n",
            "Epoch 262/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0148 - val_loss: 0.1198\n",
            "Epoch 263/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0192 - val_loss: 0.1342\n",
            "Epoch 264/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0211 - val_loss: 0.1224\n",
            "Epoch 265/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0145 - val_loss: 0.1429\n",
            "Epoch 266/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0223 - val_loss: 0.1171\n",
            "Epoch 267/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0187 - val_loss: 0.1501\n",
            "Epoch 268/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0181 - val_loss: 0.1371\n",
            "Epoch 269/600\n",
            "14/14 [==============================] - 0s 16ms/step - loss: 0.0183 - val_loss: 0.1196\n",
            "Epoch 270/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0282 - val_loss: 0.1332\n",
            "Epoch 271/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0191 - val_loss: 0.1316\n",
            "Epoch 272/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0156 - val_loss: 0.1386\n",
            "Epoch 273/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0166 - val_loss: 0.1265\n",
            "Epoch 274/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0192 - val_loss: 0.1285\n",
            "Epoch 275/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0136 - val_loss: 0.1382\n",
            "Epoch 276/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0170 - val_loss: 0.1347\n",
            "Epoch 277/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0184 - val_loss: 0.1294\n",
            "Epoch 278/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0141 - val_loss: 0.1362\n",
            "Epoch 279/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0127 - val_loss: 0.1250\n",
            "Epoch 280/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0146 - val_loss: 0.1327\n",
            "Epoch 281/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0142 - val_loss: 0.1239\n",
            "Epoch 282/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0141 - val_loss: 0.1630\n",
            "Epoch 283/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0169 - val_loss: 0.1251\n",
            "Epoch 284/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0100 - val_loss: 0.1429\n",
            "Epoch 285/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0141 - val_loss: 0.1330\n",
            "Epoch 286/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0161 - val_loss: 0.1333\n",
            "Epoch 287/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0145 - val_loss: 0.1281\n",
            "Epoch 288/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0151 - val_loss: 0.1292\n",
            "Epoch 289/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0135 - val_loss: 0.1539\n",
            "Epoch 290/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0151 - val_loss: 0.1271\n",
            "Epoch 291/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0177 - val_loss: 0.1296\n",
            "Epoch 292/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0149 - val_loss: 0.1567\n",
            "Epoch 293/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0160 - val_loss: 0.1255\n",
            "Epoch 294/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0180 - val_loss: 0.1189\n",
            "Epoch 295/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0119 - val_loss: 0.1708\n",
            "Epoch 296/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0144 - val_loss: 0.1074\n",
            "Epoch 297/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0234 - val_loss: 0.1635\n",
            "Epoch 298/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0137 - val_loss: 0.1118\n",
            "Epoch 299/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0123 - val_loss: 0.1403\n",
            "Epoch 300/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0089 - val_loss: 0.1339\n",
            "Epoch 301/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0122 - val_loss: 0.1314\n",
            "Epoch 302/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0126 - val_loss: 0.1248\n",
            "Epoch 303/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0132 - val_loss: 0.1240\n",
            "Epoch 304/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0124 - val_loss: 0.1511\n",
            "Epoch 305/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0148 - val_loss: 0.1174\n",
            "Epoch 306/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0102 - val_loss: 0.1529\n",
            "Epoch 307/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0147 - val_loss: 0.1217\n",
            "Epoch 308/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0118 - val_loss: 0.1369\n",
            "Epoch 309/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0144 - val_loss: 0.1550\n",
            "Epoch 310/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0133 - val_loss: 0.1170\n",
            "Epoch 311/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0142 - val_loss: 0.1562\n",
            "Epoch 312/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0113 - val_loss: 0.1296\n",
            "Epoch 313/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0187 - val_loss: 0.1301\n",
            "Epoch 314/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0123 - val_loss: 0.1497\n",
            "Epoch 315/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0147 - val_loss: 0.1170\n",
            "Epoch 316/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0153 - val_loss: 0.1774\n",
            "Epoch 317/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0166 - val_loss: 0.1337\n",
            "Epoch 318/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0114 - val_loss: 0.1471\n",
            "Epoch 319/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0094 - val_loss: 0.1352\n",
            "Epoch 320/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0164 - val_loss: 0.1330\n",
            "Epoch 321/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0136 - val_loss: 0.1410\n",
            "Epoch 322/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0132 - val_loss: 0.1276\n",
            "Epoch 323/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0146 - val_loss: 0.1593\n",
            "Epoch 324/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0109 - val_loss: 0.1308\n",
            "Epoch 325/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0140 - val_loss: 0.1573\n",
            "Epoch 326/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0082 - val_loss: 0.1385\n",
            "Epoch 327/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0128 - val_loss: 0.1804\n",
            "Epoch 328/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0166 - val_loss: 0.1131\n",
            "Epoch 329/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0125 - val_loss: 0.1595\n",
            "Epoch 330/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0129 - val_loss: 0.1315\n",
            "Epoch 331/600\n",
            "14/14 [==============================] - 0s 16ms/step - loss: 0.0156 - val_loss: 0.1091\n",
            "Epoch 332/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0206 - val_loss: 0.1680\n",
            "Epoch 333/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0131 - val_loss: 0.1252\n",
            "Epoch 334/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0119 - val_loss: 0.1222\n",
            "Epoch 335/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0153 - val_loss: 0.1527\n",
            "Epoch 336/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0128 - val_loss: 0.1253\n",
            "Epoch 337/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0102 - val_loss: 0.1589\n",
            "Epoch 338/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0146 - val_loss: 0.1281\n",
            "Epoch 339/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0148 - val_loss: 0.1425\n",
            "Epoch 340/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0118 - val_loss: 0.1347\n",
            "Epoch 341/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0081 - val_loss: 0.1659\n",
            "Epoch 342/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0115 - val_loss: 0.1321\n",
            "Epoch 343/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0139 - val_loss: 0.1366\n",
            "Epoch 344/600\n",
            "14/14 [==============================] - 0s 7ms/step - loss: 0.0089 - val_loss: 0.1380\n",
            "Epoch 345/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0080 - val_loss: 0.1266\n",
            "Epoch 346/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0137 - val_loss: 0.1421\n",
            "Epoch 347/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0140 - val_loss: 0.1233\n",
            "Epoch 348/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0084 - val_loss: 0.1596\n",
            "Epoch 349/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0103 - val_loss: 0.1226\n",
            "Epoch 350/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0088 - val_loss: 0.1575\n",
            "Epoch 351/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0095 - val_loss: 0.1289\n",
            "Epoch 352/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0125 - val_loss: 0.1407\n",
            "Epoch 353/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0081 - val_loss: 0.1525\n",
            "Epoch 354/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0086 - val_loss: 0.1415\n",
            "Epoch 355/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0095 - val_loss: 0.1426\n",
            "Epoch 356/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0096 - val_loss: 0.1397\n",
            "Epoch 357/600\n",
            "14/14 [==============================] - 0s 7ms/step - loss: 0.0117 - val_loss: 0.1347\n",
            "Epoch 358/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0130 - val_loss: 0.1373\n",
            "Epoch 359/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0118 - val_loss: 0.1434\n",
            "Epoch 360/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0078 - val_loss: 0.1606\n",
            "Epoch 361/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0101 - val_loss: 0.1320\n",
            "Epoch 362/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0146 - val_loss: 0.1304\n",
            "Epoch 363/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0093 - val_loss: 0.1479\n",
            "Epoch 364/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0087 - val_loss: 0.1477\n",
            "Epoch 365/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0083 - val_loss: 0.1592\n",
            "Epoch 366/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0067 - val_loss: 0.1457\n",
            "Epoch 367/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0092 - val_loss: 0.1510\n",
            "Epoch 368/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0112 - val_loss: 0.1259\n",
            "Epoch 369/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0090 - val_loss: 0.1858\n",
            "Epoch 370/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0085 - val_loss: 0.1364\n",
            "Epoch 371/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0103 - val_loss: 0.1520\n",
            "Epoch 372/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0094 - val_loss: 0.1579\n",
            "Epoch 373/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0073 - val_loss: 0.1399\n",
            "Epoch 374/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0085 - val_loss: 0.1527\n",
            "Epoch 375/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0081 - val_loss: 0.1327\n",
            "Epoch 376/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0085 - val_loss: 0.1781\n",
            "Epoch 377/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0062 - val_loss: 0.1384\n",
            "Epoch 378/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0099 - val_loss: 0.1843\n",
            "Epoch 379/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0086 - val_loss: 0.1274\n",
            "Epoch 380/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0088 - val_loss: 0.1553\n",
            "Epoch 381/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0097 - val_loss: 0.1501\n",
            "Epoch 382/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0081 - val_loss: 0.1526\n",
            "Epoch 383/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0202 - val_loss: 0.1297\n",
            "Epoch 384/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0355 - val_loss: 0.1493\n",
            "Epoch 385/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0080 - val_loss: 0.1695\n",
            "Epoch 386/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0070 - val_loss: 0.1452\n",
            "Epoch 387/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0052 - val_loss: 0.1627\n",
            "Epoch 388/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0073 - val_loss: 0.1306\n",
            "Epoch 389/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0086 - val_loss: 0.1794\n",
            "Epoch 390/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0087 - val_loss: 0.1328\n",
            "Epoch 391/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0070 - val_loss: 0.1697\n",
            "Epoch 392/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0060 - val_loss: 0.1568\n",
            "Epoch 393/600\n",
            "14/14 [==============================] - 0s 15ms/step - loss: 0.0067 - val_loss: 0.1687\n",
            "Epoch 394/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0099 - val_loss: 0.1381\n",
            "Epoch 395/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0107 - val_loss: 0.1550\n",
            "Epoch 396/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0090 - val_loss: 0.1631\n",
            "Epoch 397/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0108 - val_loss: 0.1612\n",
            "Epoch 398/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0079 - val_loss: 0.1609\n",
            "Epoch 399/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0057 - val_loss: 0.1565\n",
            "Epoch 400/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0087 - val_loss: 0.1468\n",
            "Epoch 401/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0111 - val_loss: 0.1385\n",
            "Epoch 402/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0097 - val_loss: 0.1808\n",
            "Epoch 403/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0086 - val_loss: 0.1358\n",
            "Epoch 404/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0094 - val_loss: 0.2052\n",
            "Epoch 405/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0110 - val_loss: 0.1637\n",
            "Epoch 406/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0228 - val_loss: 0.1380\n",
            "Epoch 407/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0499 - val_loss: 0.1533\n",
            "Epoch 408/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0136 - val_loss: 0.1888\n",
            "Epoch 409/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0173 - val_loss: 0.1289\n",
            "Epoch 410/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0163 - val_loss: 0.1642\n",
            "Epoch 411/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0094 - val_loss: 0.1316\n",
            "Epoch 412/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0082 - val_loss: 0.1711\n",
            "Epoch 413/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0060 - val_loss: 0.1438\n",
            "Epoch 414/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0073 - val_loss: 0.1411\n",
            "Epoch 415/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0077 - val_loss: 0.1643\n",
            "Epoch 416/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0059 - val_loss: 0.1532\n",
            "Epoch 417/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0096 - val_loss: 0.1600\n",
            "Epoch 418/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0086 - val_loss: 0.1567\n",
            "Epoch 419/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0060 - val_loss: 0.1662\n",
            "Epoch 420/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0043 - val_loss: 0.1687\n",
            "Epoch 421/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0063 - val_loss: 0.1554\n",
            "Epoch 422/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0076 - val_loss: 0.1699\n",
            "Epoch 423/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0094 - val_loss: 0.1511\n",
            "Epoch 424/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0084 - val_loss: 0.2188\n",
            "Epoch 425/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0088 - val_loss: 0.1433\n",
            "Epoch 426/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0050 - val_loss: 0.1950\n",
            "Epoch 427/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0112 - val_loss: 0.1395\n",
            "Epoch 428/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0057 - val_loss: 0.1779\n",
            "Epoch 429/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0082 - val_loss: 0.1506\n",
            "Epoch 430/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0083 - val_loss: 0.1755\n",
            "Epoch 431/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0045 - val_loss: 0.1626\n",
            "Epoch 432/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0083 - val_loss: 0.1582\n",
            "Epoch 433/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0072 - val_loss: 0.1617\n",
            "Epoch 434/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0051 - val_loss: 0.1715\n",
            "Epoch 435/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0054 - val_loss: 0.1518\n",
            "Epoch 436/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0064 - val_loss: 0.1803\n",
            "Epoch 437/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0091 - val_loss: 0.1476\n",
            "Epoch 438/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0044 - val_loss: 0.1832\n",
            "Epoch 439/600\n",
            "14/14 [==============================] - 0s 7ms/step - loss: 0.0086 - val_loss: 0.1465\n",
            "Epoch 440/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0133 - val_loss: 0.1575\n",
            "Epoch 441/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0086 - val_loss: 0.1574\n",
            "Epoch 442/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0070 - val_loss: 0.1649\n",
            "Epoch 443/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0055 - val_loss: 0.1673\n",
            "Epoch 444/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0052 - val_loss: 0.1941\n",
            "Epoch 445/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0072 - val_loss: 0.1503\n",
            "Epoch 446/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0050 - val_loss: 0.2150\n",
            "Epoch 447/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0092 - val_loss: 0.1682\n",
            "Epoch 448/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0060 - val_loss: 0.1730\n",
            "Epoch 449/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0065 - val_loss: 0.1592\n",
            "Epoch 450/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0074 - val_loss: 0.1665\n",
            "Epoch 451/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0046 - val_loss: 0.1683\n",
            "Epoch 452/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0064 - val_loss: 0.1680\n",
            "Epoch 453/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0069 - val_loss: 0.1498\n",
            "Epoch 454/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0086 - val_loss: 0.2196\n",
            "Epoch 455/600\n",
            "14/14 [==============================] - 0s 15ms/step - loss: 0.0058 - val_loss: 0.1601\n",
            "Epoch 456/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0058 - val_loss: 0.2211\n",
            "Epoch 457/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0067 - val_loss: 0.1540\n",
            "Epoch 458/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0056 - val_loss: 0.1857\n",
            "Epoch 459/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0073 - val_loss: 0.1501\n",
            "Epoch 460/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0099 - val_loss: 0.1724\n",
            "Epoch 461/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0043 - val_loss: 0.1813\n",
            "Epoch 462/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0037 - val_loss: 0.1782\n",
            "Epoch 463/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0048 - val_loss: 0.1780\n",
            "Epoch 464/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0047 - val_loss: 0.1744\n",
            "Epoch 465/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0030 - val_loss: 0.1722\n",
            "Epoch 466/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0037 - val_loss: 0.1573\n",
            "Epoch 467/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0085 - val_loss: 0.1803\n",
            "Epoch 468/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0034 - val_loss: 0.1803\n",
            "Epoch 469/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0068 - val_loss: 0.1568\n",
            "Epoch 470/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0078 - val_loss: 0.2186\n",
            "Epoch 471/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0061 - val_loss: 0.1619\n",
            "Epoch 472/600\n",
            "14/14 [==============================] - 0s 7ms/step - loss: 0.0030 - val_loss: 0.1977\n",
            "Epoch 473/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0054 - val_loss: 0.1559\n",
            "Epoch 474/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0081 - val_loss: 0.2100\n",
            "Epoch 475/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0063 - val_loss: 0.1955\n",
            "Epoch 476/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0049 - val_loss: 0.1913\n",
            "Epoch 477/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0076 - val_loss: 0.1624\n",
            "Epoch 478/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0091 - val_loss: 0.2024\n",
            "Epoch 479/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0066 - val_loss: 0.2054\n",
            "Epoch 480/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0101 - val_loss: 0.1631\n",
            "Epoch 481/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0058 - val_loss: 0.2109\n",
            "Epoch 482/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0058 - val_loss: 0.1787\n",
            "Epoch 483/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0058 - val_loss: 0.1878\n",
            "Epoch 484/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0030 - val_loss: 0.1898\n",
            "Epoch 485/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0049 - val_loss: 0.1881\n",
            "Epoch 486/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0041 - val_loss: 0.1842\n",
            "Epoch 487/600\n",
            "14/14 [==============================] - 0s 7ms/step - loss: 0.0044 - val_loss: 0.1829\n",
            "Epoch 488/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0051 - val_loss: 0.2084\n",
            "Epoch 489/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0044 - val_loss: 0.1828\n",
            "Epoch 490/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0044 - val_loss: 0.2183\n",
            "Epoch 491/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0058 - val_loss: 0.1751\n",
            "Epoch 492/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0049 - val_loss: 0.2080\n",
            "Epoch 493/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0053 - val_loss: 0.1767\n",
            "Epoch 494/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0022 - val_loss: 0.1884\n",
            "Epoch 495/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0038 - val_loss: 0.1868\n",
            "Epoch 496/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0046 - val_loss: 0.1882\n",
            "Epoch 497/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0048 - val_loss: 0.1918\n",
            "Epoch 498/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0040 - val_loss: 0.1846\n",
            "Epoch 499/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0043 - val_loss: 0.1953\n",
            "Epoch 500/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0032 - val_loss: 0.1983\n",
            "Epoch 501/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0029 - val_loss: 0.1962\n",
            "Epoch 502/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0036 - val_loss: 0.2012\n",
            "Epoch 503/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0036 - val_loss: 0.1958\n",
            "Epoch 504/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0023 - val_loss: 0.1978\n",
            "Epoch 505/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0032 - val_loss: 0.1884\n",
            "Epoch 506/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0044 - val_loss: 0.1978\n",
            "Epoch 507/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0032 - val_loss: 0.1948\n",
            "Epoch 508/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0037 - val_loss: 0.2080\n",
            "Epoch 509/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0027 - val_loss: 0.1902\n",
            "Epoch 510/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0036 - val_loss: 0.2031\n",
            "Epoch 511/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0036 - val_loss: 0.1854\n",
            "Epoch 512/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0025 - val_loss: 0.2100\n",
            "Epoch 513/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0033 - val_loss: 0.1981\n",
            "Epoch 514/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0033 - val_loss: 0.2252\n",
            "Epoch 515/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0044 - val_loss: 0.1892\n",
            "Epoch 516/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0037 - val_loss: 0.2134\n",
            "Epoch 517/600\n",
            "14/14 [==============================] - 0s 15ms/step - loss: 0.0034 - val_loss: 0.1894\n",
            "Epoch 518/600\n",
            "14/14 [==============================] - 0s 7ms/step - loss: 0.0056 - val_loss: 0.1986\n",
            "Epoch 519/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0057 - val_loss: 0.1953\n",
            "Epoch 520/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0030 - val_loss: 0.2187\n",
            "Epoch 521/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0031 - val_loss: 0.1935\n",
            "Epoch 522/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0039 - val_loss: 0.2314\n",
            "Epoch 523/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0037 - val_loss: 0.1910\n",
            "Epoch 524/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0054 - val_loss: 0.2178\n",
            "Epoch 525/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0045 - val_loss: 0.1899\n",
            "Epoch 526/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0023 - val_loss: 0.2227\n",
            "Epoch 527/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0045 - val_loss: 0.2111\n",
            "Epoch 528/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0037 - val_loss: 0.2052\n",
            "Epoch 529/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0040 - val_loss: 0.2307\n",
            "Epoch 530/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0042 - val_loss: 0.1966\n",
            "Epoch 531/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0050 - val_loss: 0.2111\n",
            "Epoch 532/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0029 - val_loss: 0.2251\n",
            "Epoch 533/600\n",
            "14/14 [==============================] - 0s 7ms/step - loss: 0.0038 - val_loss: 0.1952\n",
            "Epoch 534/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0039 - val_loss: 0.2277\n",
            "Epoch 535/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0033 - val_loss: 0.2046\n",
            "Epoch 536/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0040 - val_loss: 0.2023\n",
            "Epoch 537/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0028 - val_loss: 0.2383\n",
            "Epoch 538/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0043 - val_loss: 0.2135\n",
            "Epoch 539/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0025 - val_loss: 0.2208\n",
            "Epoch 540/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0028 - val_loss: 0.2066\n",
            "Epoch 541/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0024 - val_loss: 0.2164\n",
            "Epoch 542/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0030 - val_loss: 0.2407\n",
            "Epoch 543/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0026 - val_loss: 0.2152\n",
            "Epoch 544/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0026 - val_loss: 0.2133\n",
            "Epoch 545/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0030 - val_loss: 0.2156\n",
            "Epoch 546/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0038 - val_loss: 0.2140\n",
            "Epoch 547/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0033 - val_loss: 0.2264\n",
            "Epoch 548/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0022 - val_loss: 0.2278\n",
            "Epoch 549/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0032 - val_loss: 0.2353\n",
            "Epoch 550/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0046 - val_loss: 0.2184\n",
            "Epoch 551/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0020 - val_loss: 0.2311\n",
            "Epoch 552/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0023 - val_loss: 0.2348\n",
            "Epoch 553/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0023 - val_loss: 0.2295\n",
            "Epoch 554/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0044 - val_loss: 0.2189\n",
            "Epoch 555/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0026 - val_loss: 0.2565\n",
            "Epoch 556/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0033 - val_loss: 0.2523\n",
            "Epoch 557/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0036 - val_loss: 0.2210\n",
            "Epoch 558/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0033 - val_loss: 0.2488\n",
            "Epoch 559/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0023 - val_loss: 0.2203\n",
            "Epoch 560/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0034 - val_loss: 0.2441\n",
            "Epoch 561/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0026 - val_loss: 0.2466\n",
            "Epoch 562/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0028 - val_loss: 0.2248\n",
            "Epoch 563/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0027 - val_loss: 0.2433\n",
            "Epoch 564/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0025 - val_loss: 0.2283\n",
            "Epoch 565/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0016 - val_loss: 0.2468\n",
            "Epoch 566/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0023 - val_loss: 0.2297\n",
            "Epoch 567/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0026 - val_loss: 0.2344\n",
            "Epoch 568/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0021 - val_loss: 0.2551\n",
            "Epoch 569/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0034 - val_loss: 0.2157\n",
            "Epoch 570/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0018 - val_loss: 0.2460\n",
            "Epoch 571/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0025 - val_loss: 0.2257\n",
            "Epoch 572/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0036 - val_loss: 0.2470\n",
            "Epoch 573/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0019 - val_loss: 0.2468\n",
            "Epoch 574/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0025 - val_loss: 0.2328\n",
            "Epoch 575/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0031 - val_loss: 0.2601\n",
            "Epoch 576/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0023 - val_loss: 0.2515\n",
            "Epoch 577/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0036 - val_loss: 0.2220\n",
            "Epoch 578/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0051 - val_loss: 0.3052\n",
            "Epoch 579/600\n",
            "14/14 [==============================] - 0s 15ms/step - loss: 0.0036 - val_loss: 0.2243\n",
            "Epoch 580/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0037 - val_loss: 0.2605\n",
            "Epoch 581/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0023 - val_loss: 0.2515\n",
            "Epoch 582/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0027 - val_loss: 0.2386\n",
            "Epoch 583/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0043 - val_loss: 0.3107\n",
            "Epoch 584/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0049 - val_loss: 0.2423\n",
            "Epoch 585/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0017 - val_loss: 0.2372\n",
            "Epoch 586/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0025 - val_loss: 0.2522\n",
            "Epoch 587/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0023 - val_loss: 0.2654\n",
            "Epoch 588/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0017 - val_loss: 0.2585\n",
            "Epoch 589/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0029 - val_loss: 0.2394\n",
            "Epoch 590/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0033 - val_loss: 0.2629\n",
            "Epoch 591/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0023 - val_loss: 0.2370\n",
            "Epoch 592/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0033 - val_loss: 0.2696\n",
            "Epoch 593/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0018 - val_loss: 0.2632\n",
            "Epoch 594/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0027 - val_loss: 0.2448\n",
            "Epoch 595/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0015 - val_loss: 0.2757\n",
            "Epoch 596/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0018 - val_loss: 0.2390\n",
            "Epoch 597/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0034 - val_loss: 0.2886\n",
            "Epoch 598/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0026 - val_loss: 0.2850\n",
            "Epoch 599/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0023 - val_loss: 0.2603\n",
            "Epoch 600/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0022 - val_loss: 0.2681\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tensorflow.python.keras.callbacks.History at 0x7f5899d91a20>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 39
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "xxnHXybbK6pW"
      },
      "source": [
        "## Check history of Training"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 204
        },
        "id": "4USqfP8qLO0_",
        "outputId": "55351a5d-fee8-4a81-c4ec-d0ccf1d00ba8"
      },
      "source": [
        "losses = pd.DataFrame(model.history.history)\r\n",
        "losses.head()"
      ],
      "execution_count": 42,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<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>loss</th>\n",
              "      <th>val_loss</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>0.662070</td>\n",
              "      <td>0.633214</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>0.606209</td>\n",
              "      <td>0.574316</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>0.541479</td>\n",
              "      <td>0.506659</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>0.478809</td>\n",
              "      <td>0.450043</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>0.427443</td>\n",
              "      <td>0.396876</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "       loss  val_loss\n",
              "0  0.662070  0.633214\n",
              "1  0.606209  0.574316\n",
              "2  0.541479  0.506659\n",
              "3  0.478809  0.450043\n",
              "4  0.427443  0.396876"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 42
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 265
        },
        "id": "QnK_UE9JLV8A",
        "outputId": "9db00de5-fd63-4208-95b5-b908bc679a8b"
      },
      "source": [
        "losses.plot();"
      ],
      "execution_count": 43,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "SGLuAWV9Lbi0"
      },
      "source": [
        "We can see from the chart that our model is definitely OVERFITTING because `val_loss` (orange line) suddenly spikes up after a certain epochs.\r\n",
        "\r\n",
        "val_loss is getting wrose and wrose after a certain point in time.\r\n",
        "\r\n",
        "To avoid this, we can put Early Stopping."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "j9lpiWNxMTgQ"
      },
      "source": [
        "# Model Building (with Early Stopping)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "e3r5FcmLMbpX"
      },
      "source": [
        "\r\n",
        "## Example Two: Early Stopping"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ZwBX_vewLvt2"
      },
      "source": [
        "model = Sequential()\r\n",
        "\r\n",
        "model.add(Dense(units=30, activation='relu'))\r\n",
        "model.add(Dense(units=15, activation='relu'))\r\n",
        "\r\n",
        "# As this is BINARY CLASSIFICATOIN problem, we want our last layer to be using Sigmoid \r\n",
        "model.add(Dense(units=1, activation='sigmoid')) # output layer with Sigmoid Activation Function\r\n",
        "\r\n",
        "model.compile(optimizer='adam', loss='binary_crossentropy')"
      ],
      "execution_count": 44,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "jaQCvrwXNYZO"
      },
      "source": [
        "## Early stopping"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0ldQPGIqNNXx"
      },
      "source": [
        "Stop training when a monitored quantity has stopped improving.\r\n",
        "\r\n",
        "Parameters:\r\n",
        "\r\n",
        "* `monitor`: Quantity to be monitored.\r\n",
        "* `min_delta`: Minimum change in the monitored quantity to qualify as an improvement, i.e. an absolute change of less than min_delta, will count as no improvement.\r\n",
        "* `patience`: Number of epochs with no improvement after which training will be stopped.\r\n",
        "* `verbose`: verbosity mode.\r\n",
        "* `mode`: One of `{\"auto\", \"min\", \"max\"}`. \r\n",
        "  * In `min` mode, training will stop when the quantity monitored has stopped decreasing; \r\n",
        "  * in `max` mode it will stop when the quantitymonitored has stopped increasing; \r\n",
        "  * in `auto`mode, the direction is automatically inferred from the name of the monitored quantity."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YPgqquXOMxXJ"
      },
      "source": [
        "from tensorflow.keras.callbacks import EarlyStopping"
      ],
      "execution_count": 45,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "d5BPYV_xNNVE"
      },
      "source": [
        "# In our case, we want to check orange line 'val_loss' whether the `loss` is getting smaller and smaller or not\r\n",
        "# As we are watching `loss`, the smaller the value => the better the model. So we will put mode as 'min'.\r\n",
        "# For cases like watching `accuracy` which we want to maximize , we want to use 'max'.\r\n",
        "\r\n",
        "early_stop = EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=25)"
      ],
      "execution_count": 47,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "lwTyqd7mMqdX"
      },
      "source": [
        "## Model Training with Early Stopping"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "G1TIh-34NNSc",
        "outputId": "ebb813df-af2b-42d2-f998-3abc3591fcf7"
      },
      "source": [
        "model.fit(x=X_train, \r\n",
        "                y=y_train, \r\n",
        "                epochs=600, \r\n",
        "                validation_data=(X_test, y_test), \r\n",
        "                callbacks=[early_stop], \r\n",
        "                verbose=1)"
      ],
      "execution_count": 48,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/600\n",
            "14/14 [==============================] - 1s 15ms/step - loss: 0.6647 - val_loss: 0.6276\n",
            "Epoch 2/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.6296 - val_loss: 0.5839\n",
            "Epoch 3/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.5838 - val_loss: 0.5387\n",
            "Epoch 4/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.5364 - val_loss: 0.4895\n",
            "Epoch 5/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.4966 - val_loss: 0.4377\n",
            "Epoch 6/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.4515 - val_loss: 0.3896\n",
            "Epoch 7/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.3905 - val_loss: 0.3470\n",
            "Epoch 8/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.3597 - val_loss: 0.3100\n",
            "Epoch 9/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.3271 - val_loss: 0.2845\n",
            "Epoch 10/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.3000 - val_loss: 0.2583\n",
            "Epoch 11/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.2899 - val_loss: 0.2405\n",
            "Epoch 12/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.2381 - val_loss: 0.2227\n",
            "Epoch 13/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.2281 - val_loss: 0.2105\n",
            "Epoch 14/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.2198 - val_loss: 0.1981\n",
            "Epoch 15/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.2362 - val_loss: 0.1913\n",
            "Epoch 16/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.2030 - val_loss: 0.1806\n",
            "Epoch 17/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1909 - val_loss: 0.1746\n",
            "Epoch 18/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1759 - val_loss: 0.1660\n",
            "Epoch 19/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1889 - val_loss: 0.1605\n",
            "Epoch 20/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.1629 - val_loss: 0.1546\n",
            "Epoch 21/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1572 - val_loss: 0.1513\n",
            "Epoch 22/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.1689 - val_loss: 0.1487\n",
            "Epoch 23/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1601 - val_loss: 0.1416\n",
            "Epoch 24/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.1506 - val_loss: 0.1405\n",
            "Epoch 25/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.1306 - val_loss: 0.1358\n",
            "Epoch 26/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1276 - val_loss: 0.1346\n",
            "Epoch 27/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.1263 - val_loss: 0.1310\n",
            "Epoch 28/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.1139 - val_loss: 0.1299\n",
            "Epoch 29/600\n",
            "14/14 [==============================] - 0s 15ms/step - loss: 0.1336 - val_loss: 0.1261\n",
            "Epoch 30/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.1191 - val_loss: 0.1260\n",
            "Epoch 31/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.1112 - val_loss: 0.1235\n",
            "Epoch 32/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.1046 - val_loss: 0.1211\n",
            "Epoch 33/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.1038 - val_loss: 0.1209\n",
            "Epoch 34/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0965 - val_loss: 0.1192\n",
            "Epoch 35/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0939 - val_loss: 0.1209\n",
            "Epoch 36/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0880 - val_loss: 0.1162\n",
            "Epoch 37/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0934 - val_loss: 0.1187\n",
            "Epoch 38/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0896 - val_loss: 0.1175\n",
            "Epoch 39/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1059 - val_loss: 0.1144\n",
            "Epoch 40/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0740 - val_loss: 0.1158\n",
            "Epoch 41/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0801 - val_loss: 0.1157\n",
            "Epoch 42/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0804 - val_loss: 0.1129\n",
            "Epoch 43/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0810 - val_loss: 0.1174\n",
            "Epoch 44/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0713 - val_loss: 0.1108\n",
            "Epoch 45/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0821 - val_loss: 0.1130\n",
            "Epoch 46/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0851 - val_loss: 0.1139\n",
            "Epoch 47/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0732 - val_loss: 0.1116\n",
            "Epoch 48/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0686 - val_loss: 0.1104\n",
            "Epoch 49/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0695 - val_loss: 0.1103\n",
            "Epoch 50/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0658 - val_loss: 0.1100\n",
            "Epoch 51/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0593 - val_loss: 0.1126\n",
            "Epoch 52/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0519 - val_loss: 0.1115\n",
            "Epoch 53/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0661 - val_loss: 0.1109\n",
            "Epoch 54/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0606 - val_loss: 0.1106\n",
            "Epoch 55/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0649 - val_loss: 0.1103\n",
            "Epoch 56/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0685 - val_loss: 0.1109\n",
            "Epoch 57/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0590 - val_loss: 0.1145\n",
            "Epoch 58/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0609 - val_loss: 0.1092\n",
            "Epoch 59/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0682 - val_loss: 0.1077\n",
            "Epoch 60/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0536 - val_loss: 0.1106\n",
            "Epoch 61/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0645 - val_loss: 0.1101\n",
            "Epoch 62/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0731 - val_loss: 0.1062\n",
            "Epoch 63/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0571 - val_loss: 0.1090\n",
            "Epoch 64/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0517 - val_loss: 0.1106\n",
            "Epoch 65/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0737 - val_loss: 0.1096\n",
            "Epoch 66/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0674 - val_loss: 0.1073\n",
            "Epoch 67/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0529 - val_loss: 0.1110\n",
            "Epoch 68/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0621 - val_loss: 0.1098\n",
            "Epoch 69/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0567 - val_loss: 0.1114\n",
            "Epoch 70/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0539 - val_loss: 0.1097\n",
            "Epoch 71/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0437 - val_loss: 0.1172\n",
            "Epoch 72/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0509 - val_loss: 0.1094\n",
            "Epoch 73/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0470 - val_loss: 0.1071\n",
            "Epoch 74/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0554 - val_loss: 0.1165\n",
            "Epoch 75/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0551 - val_loss: 0.1068\n",
            "Epoch 76/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0592 - val_loss: 0.1078\n",
            "Epoch 77/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0576 - val_loss: 0.1122\n",
            "Epoch 78/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0507 - val_loss: 0.1109\n",
            "Epoch 79/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0435 - val_loss: 0.1115\n",
            "Epoch 80/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0536 - val_loss: 0.1091\n",
            "Epoch 81/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0427 - val_loss: 0.1153\n",
            "Epoch 82/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0670 - val_loss: 0.1077\n",
            "Epoch 83/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0611 - val_loss: 0.1113\n",
            "Epoch 84/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0476 - val_loss: 0.1111\n",
            "Epoch 85/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0515 - val_loss: 0.1098\n",
            "Epoch 86/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0609 - val_loss: 0.1097\n",
            "Epoch 87/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0578 - val_loss: 0.1119\n",
            "Epoch 00087: early stopping\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tensorflow.python.keras.callbacks.History at 0x7f5896b63208>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 48
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Ab4UO7zLNNPw"
      },
      "source": [
        "Now we can see that because of Early Stopping, the model traning got stopped once there is no improvement in training (after 87 epoch).\r\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "4AsjHfk8PxJN"
      },
      "source": [
        "## Check history of Training (with Early Stopping)"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "QAvfqW68P47Z"
      },
      "source": [
        "model_loss = pd.DataFrame(model.history.history)"
      ],
      "execution_count": 53,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 265
        },
        "id": "nmYRGF6FQACN",
        "outputId": "e7494172-d5ee-4d63-8e6e-72f00e82eb6d"
      },
      "source": [
        "model_loss.plot();"
      ],
      "execution_count": 54,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "T3974ys6QBSQ"
      },
      "source": [
        "# Model Building (with Adding in DropOut Layers)\r\n",
        "* Turn off % of neurons randomly\r\n",
        "* `rate`: 0 - 1 (percentage of neurons to be turned off randomly while training on each epoch)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "6fAxxJrNQiB8"
      },
      "source": [
        "## Example Three: Adding in DropOut Layers"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "HhEOcyEMQvK6"
      },
      "source": [
        "from tensorflow.keras.layers import Dropout"
      ],
      "execution_count": 55,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "AETyrGv5Qke-"
      },
      "source": [
        "model = Sequential()\r\n",
        "\r\n",
        "model.add(Dense(units=30, activation='relu'))\r\n",
        "model.add(Dropout(rate=0.5)) # meaning that 50% of neurons will be RANDOMLY turned off per each epoch training\r\n",
        "\r\n",
        "model.add(Dense(units=15, activation='relu'))\r\n",
        "model.add(Dropout(rate=0.5))\r\n",
        "\r\n",
        "# BINARY CLASSIFICATOIN\r\n",
        "model.add(Dense(units=1, activation='sigmoid'))\r\n",
        "\r\n",
        "model.compile(optimizer='adam', loss='binary_crossentropy')"
      ],
      "execution_count": 56,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "xvAIvaBUSHuG",
        "outputId": "8a21c702-f8a8-4e93-aed2-73b9a741f718"
      },
      "source": [
        "# We will use dropout in combination with earlystop callback. This will help us the model from overfitting\r\n",
        "model.fit(x=X_train, \r\n",
        "                y=y_train, \r\n",
        "                epochs=600, \r\n",
        "                validation_data=(X_test, y_test), \r\n",
        "                callbacks=[early_stop], \r\n",
        "                verbose=1)"
      ],
      "execution_count": 57,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/600\n",
            "14/14 [==============================] - 1s 15ms/step - loss: 0.7288 - val_loss: 0.6840\n",
            "Epoch 2/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.6861 - val_loss: 0.6673\n",
            "Epoch 3/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.6613 - val_loss: 0.6492\n",
            "Epoch 4/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.6427 - val_loss: 0.6340\n",
            "Epoch 5/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.6301 - val_loss: 0.6183\n",
            "Epoch 6/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.6176 - val_loss: 0.6012\n",
            "Epoch 7/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.6108 - val_loss: 0.5838\n",
            "Epoch 8/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.6022 - val_loss: 0.5580\n",
            "Epoch 9/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.5814 - val_loss: 0.5265\n",
            "Epoch 10/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.5538 - val_loss: 0.5016\n",
            "Epoch 11/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.5437 - val_loss: 0.4743\n",
            "Epoch 12/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.5187 - val_loss: 0.4449\n",
            "Epoch 13/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.4907 - val_loss: 0.4163\n",
            "Epoch 14/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.4317 - val_loss: 0.3885\n",
            "Epoch 15/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.4624 - val_loss: 0.3644\n",
            "Epoch 16/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.4373 - val_loss: 0.3518\n",
            "Epoch 17/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.4510 - val_loss: 0.3256\n",
            "Epoch 18/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.3841 - val_loss: 0.3043\n",
            "Epoch 19/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.3761 - val_loss: 0.2997\n",
            "Epoch 20/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.3884 - val_loss: 0.2787\n",
            "Epoch 21/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.3966 - val_loss: 0.2705\n",
            "Epoch 22/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.3372 - val_loss: 0.2573\n",
            "Epoch 23/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.3459 - val_loss: 0.2478\n",
            "Epoch 24/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.3013 - val_loss: 0.2373\n",
            "Epoch 25/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.2995 - val_loss: 0.2233\n",
            "Epoch 26/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.3392 - val_loss: 0.2189\n",
            "Epoch 27/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.3355 - val_loss: 0.2106\n",
            "Epoch 28/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.2812 - val_loss: 0.2034\n",
            "Epoch 29/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.3095 - val_loss: 0.2088\n",
            "Epoch 30/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.2919 - val_loss: 0.1906\n",
            "Epoch 31/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.2277 - val_loss: 0.1798\n",
            "Epoch 32/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.2585 - val_loss: 0.1735\n",
            "Epoch 33/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.2489 - val_loss: 0.1679\n",
            "Epoch 34/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.3198 - val_loss: 0.1703\n",
            "Epoch 35/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.2378 - val_loss: 0.1646\n",
            "Epoch 36/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.2373 - val_loss: 0.1626\n",
            "Epoch 37/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.2568 - val_loss: 0.1653\n",
            "Epoch 38/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.2760 - val_loss: 0.1568\n",
            "Epoch 39/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.2130 - val_loss: 0.1576\n",
            "Epoch 40/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.2231 - val_loss: 0.1597\n",
            "Epoch 41/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.2471 - val_loss: 0.1523\n",
            "Epoch 42/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.2396 - val_loss: 0.1503\n",
            "Epoch 43/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.2042 - val_loss: 0.1447\n",
            "Epoch 44/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.2315 - val_loss: 0.1375\n",
            "Epoch 45/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.2484 - val_loss: 0.1447\n",
            "Epoch 46/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.1900 - val_loss: 0.1443\n",
            "Epoch 47/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1892 - val_loss: 0.1329\n",
            "Epoch 48/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.2204 - val_loss: 0.1378\n",
            "Epoch 49/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.2042 - val_loss: 0.1359\n",
            "Epoch 50/600\n",
            "14/14 [==============================] - 0s 16ms/step - loss: 0.2458 - val_loss: 0.1466\n",
            "Epoch 51/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.1527 - val_loss: 0.1317\n",
            "Epoch 52/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1839 - val_loss: 0.1274\n",
            "Epoch 53/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1610 - val_loss: 0.1212\n",
            "Epoch 54/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.1666 - val_loss: 0.1270\n",
            "Epoch 55/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1746 - val_loss: 0.1299\n",
            "Epoch 56/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1388 - val_loss: 0.1235\n",
            "Epoch 57/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1525 - val_loss: 0.1214\n",
            "Epoch 58/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1447 - val_loss: 0.1164\n",
            "Epoch 59/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.1386 - val_loss: 0.1141\n",
            "Epoch 60/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.1513 - val_loss: 0.1153\n",
            "Epoch 61/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1542 - val_loss: 0.1119\n",
            "Epoch 62/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1376 - val_loss: 0.1134\n",
            "Epoch 63/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1244 - val_loss: 0.1229\n",
            "Epoch 64/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1583 - val_loss: 0.1122\n",
            "Epoch 65/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1286 - val_loss: 0.1105\n",
            "Epoch 66/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1294 - val_loss: 0.1229\n",
            "Epoch 67/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1555 - val_loss: 0.1050\n",
            "Epoch 68/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1341 - val_loss: 0.1083\n",
            "Epoch 69/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.1334 - val_loss: 0.1054\n",
            "Epoch 70/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.1133 - val_loss: 0.1054\n",
            "Epoch 71/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.1372 - val_loss: 0.1168\n",
            "Epoch 72/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.1202 - val_loss: 0.1044\n",
            "Epoch 73/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1846 - val_loss: 0.1059\n",
            "Epoch 74/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0885 - val_loss: 0.1047\n",
            "Epoch 75/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.1595 - val_loss: 0.1049\n",
            "Epoch 76/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1087 - val_loss: 0.1152\n",
            "Epoch 77/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1077 - val_loss: 0.1090\n",
            "Epoch 78/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1286 - val_loss: 0.1007\n",
            "Epoch 79/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1078 - val_loss: 0.1022\n",
            "Epoch 80/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.1016 - val_loss: 0.1078\n",
            "Epoch 81/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1135 - val_loss: 0.1088\n",
            "Epoch 82/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1172 - val_loss: 0.1038\n",
            "Epoch 83/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1003 - val_loss: 0.1020\n",
            "Epoch 84/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0926 - val_loss: 0.1072\n",
            "Epoch 85/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0953 - val_loss: 0.1071\n",
            "Epoch 86/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.1110 - val_loss: 0.1024\n",
            "Epoch 87/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.1083 - val_loss: 0.1051\n",
            "Epoch 88/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0917 - val_loss: 0.1032\n",
            "Epoch 89/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.1302 - val_loss: 0.1070\n",
            "Epoch 90/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.1039 - val_loss: 0.0995\n",
            "Epoch 91/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1022 - val_loss: 0.1002\n",
            "Epoch 92/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1245 - val_loss: 0.1158\n",
            "Epoch 93/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0971 - val_loss: 0.0978\n",
            "Epoch 94/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1170 - val_loss: 0.1178\n",
            "Epoch 95/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0968 - val_loss: 0.0954\n",
            "Epoch 96/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1169 - val_loss: 0.0957\n",
            "Epoch 97/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0958 - val_loss: 0.1114\n",
            "Epoch 98/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1023 - val_loss: 0.0955\n",
            "Epoch 99/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0844 - val_loss: 0.1059\n",
            "Epoch 100/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1019 - val_loss: 0.1025\n",
            "Epoch 101/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0799 - val_loss: 0.1110\n",
            "Epoch 102/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0928 - val_loss: 0.0974\n",
            "Epoch 103/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0945 - val_loss: 0.0963\n",
            "Epoch 104/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1111 - val_loss: 0.1039\n",
            "Epoch 105/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0955 - val_loss: 0.1043\n",
            "Epoch 106/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1002 - val_loss: 0.1036\n",
            "Epoch 107/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0943 - val_loss: 0.1020\n",
            "Epoch 108/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0895 - val_loss: 0.1010\n",
            "Epoch 109/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0921 - val_loss: 0.0951\n",
            "Epoch 110/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0899 - val_loss: 0.1024\n",
            "Epoch 111/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0924 - val_loss: 0.0948\n",
            "Epoch 112/600\n",
            "14/14 [==============================] - 0s 16ms/step - loss: 0.1138 - val_loss: 0.0919\n",
            "Epoch 113/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1217 - val_loss: 0.1052\n",
            "Epoch 114/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0804 - val_loss: 0.1110\n",
            "Epoch 115/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1117 - val_loss: 0.0999\n",
            "Epoch 116/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0921 - val_loss: 0.0976\n",
            "Epoch 117/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0889 - val_loss: 0.1037\n",
            "Epoch 118/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0934 - val_loss: 0.0965\n",
            "Epoch 119/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0923 - val_loss: 0.0957\n",
            "Epoch 120/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0978 - val_loss: 0.0955\n",
            "Epoch 121/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.0799 - val_loss: 0.0998\n",
            "Epoch 122/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0762 - val_loss: 0.1056\n",
            "Epoch 123/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0701 - val_loss: 0.1085\n",
            "Epoch 124/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0704 - val_loss: 0.0915\n",
            "Epoch 125/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0875 - val_loss: 0.1032\n",
            "Epoch 126/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0797 - val_loss: 0.1077\n",
            "Epoch 127/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1190 - val_loss: 0.0977\n",
            "Epoch 128/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0691 - val_loss: 0.1040\n",
            "Epoch 129/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0708 - val_loss: 0.1009\n",
            "Epoch 130/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.0910 - val_loss: 0.1147\n",
            "Epoch 131/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1032 - val_loss: 0.0948\n",
            "Epoch 132/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.1083 - val_loss: 0.0980\n",
            "Epoch 133/600\n",
            "14/14 [==============================] - 0s 4ms/step - loss: 0.1384 - val_loss: 0.0979\n",
            "Epoch 134/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1047 - val_loss: 0.1078\n",
            "Epoch 135/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0607 - val_loss: 0.0967\n",
            "Epoch 136/600\n",
            "14/14 [==============================] - 0s 6ms/step - loss: 0.1055 - val_loss: 0.0925\n",
            "Epoch 137/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1297 - val_loss: 0.0969\n",
            "Epoch 138/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0843 - val_loss: 0.1168\n",
            "Epoch 139/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1087 - val_loss: 0.0980\n",
            "Epoch 140/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0781 - val_loss: 0.0943\n",
            "Epoch 141/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1033 - val_loss: 0.1009\n",
            "Epoch 142/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1179 - val_loss: 0.0964\n",
            "Epoch 143/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0866 - val_loss: 0.0963\n",
            "Epoch 144/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0608 - val_loss: 0.0980\n",
            "Epoch 145/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.1387 - val_loss: 0.0956\n",
            "Epoch 146/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0833 - val_loss: 0.1007\n",
            "Epoch 147/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0709 - val_loss: 0.1034\n",
            "Epoch 148/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0930 - val_loss: 0.0955\n",
            "Epoch 149/600\n",
            "14/14 [==============================] - 0s 5ms/step - loss: 0.0606 - val_loss: 0.1102\n",
            "Epoch 00149: early stopping\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tensorflow.python.keras.callbacks.History at 0x7f5896c30748>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 57
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "I3xs_ghNSiEj"
      },
      "source": [
        "## Check history of losses"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "n1gg9sB1Swq9"
      },
      "source": [
        "model_loss = pd.DataFrame(model.history.history)"
      ],
      "execution_count": 58,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 265
        },
        "id": "KPY5imRFS8HL",
        "outputId": "ccc544c7-7261-4042-dc71-4fa6e8c0b00e"
      },
      "source": [
        "model_loss.plot();"
      ],
      "execution_count": 59,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXQAAAD4CAYAAAD8Zh1EAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nOzdd3hUZfbA8e87k95DeiEFEkpIqKF3FERBEFwEbKCLrg3ruurPsq5ldcW2rhVd10oTGyACKr0TILTQQiAhJEAaJCSkv78/bggJBEhgkiHJ+TxPnmTuvXPvmUHPvHPuW5TWGiGEEI2fydoBCCGEsAxJ6EII0URIQhdCiCZCEroQQjQRktCFEKKJsLHWhb29vXVYWJi1Li+EEI3S5s2bM7XWPjXts1pCDwsLIy4uzlqXF0KIRkkplXyhfVJyEUKIJkISuhBCNBGS0IUQoomwWg1dCNE8lZSUkJqaSmFhobVDuao5ODgQHByMra1trZ8jCV0I0aBSU1NxdXUlLCwMpZS1w7kqaa3JysoiNTWV8PDwWj9PSi5CiAZVWFiIl5eXJPOLUErh5eVV528xktCFEA1OkvmlXc571OgSetyhbF7/dQ8y7a8QQlTX6BL6jiMn+XjFATJOFVk7FCFEI+Xi4mLtEOpFo0voEb7GP0Ti8VNWjkQIIa4ujTahH5CELoS4QlprnnzySaKjo4mJiWH27NkApKenM2DAADp37kx0dDSrVq2irKyMyZMnVx77zjvvWDn689Wq26JSajjwb8AMfKa1fv2c/e8AgyseOgG+WmsPSwZ6hr+bAy72NtJCF6IJ+Mf8XSSk5Vr0nFGBbvz9xg61OvaHH34gPj6ebdu2kZmZSffu3RkwYAAzZszguuuu49lnn6WsrIyCggLi4+M5cuQIO3fuBODEiRMWjdsSLpnQlVJm4ANgKJAKbFJKzdNaJ5w5Rmv9WJXjpwJd6iHWM+entY8ziRmS0IUQV2b16tVMnDgRs9mMn58fAwcOZNOmTXTv3p27776bkpISbrrpJjp37kyrVq1ISkpi6tSpjBgxgmHDhlk7/PPUpoXeA0jUWicBKKVmAaOBhAscPxH4u2XCq1lrXxfWJGbW5yWEEA2gti3phjZgwABWrlzJL7/8wuTJk3n88ce588472bZtG4sXL+bjjz9mzpw5fP7559YOtZra1NCDgMNVHqdWbDuPUioUCAeWXmD/vUqpOKVUXEZGRl1jrRTh68Kx3CJyC0su+xxCCNG/f39mz55NWVkZGRkZrFy5kh49epCcnIyfnx/33HMPU6ZMYcuWLWRmZlJeXs7NN9/MK6+8wpYtW6wd/nksPfR/AjBXa11W006t9XRgOkBsbOxldySP8Dl7Y7RLiOflnkYI0cyNGTOGdevW0alTJ5RSvPHGG/j7+/Pll18ybdo0bG1tcXFx4auvvuLIkSPcddddlJeXA/Daa69ZOfrz1SahHwFaVnkcXLGtJhOAB680qEup2nVREroQoq5OnTLuwSmlmDZtGtOmTau2f9KkSUyaNOm8512NrfKqalNy2QREKqXClVJ2GEl73rkHKaXaAZ7AOsuGeA6tCTFlYmc2yY1RIYSo4pIJXWtdCjwELAZ2A3O01ruUUi8ppUZVOXQCMEvX95j8lW9i81Evor3KpS+6EEJUUasautZ6IbDwnG0vnPP4RcuFdRFtroNlrzDRbg0fHB/aIJcUQojGoNGNFCWgIwR3Z0j+L6Rk55NfVGrtiIQQ4qrQ+BI6QOzdeJ0+RA+1m2d+2CEzLwohBI01oXcYAw4evBy0gXnb0vhoxQFrRySEEFbXOBO6rSN0uZ2IrGXc2sGBaYv3kpyVb+2ohBDCqhpnQgfoeieqvJQn/LehNazaL1MBCCEs72Jzpx86dIjo6OgGjObiGm9C92kLQbG02P8d/q72rEvKsnZEQghhVZYe+t+wutyGWvAYt4RlMiNJo7WWtQqFaEx+fRqO7rDsOf1j4PrXL7j76aefpmXLljz4oDGo/cUXX8TGxoZly5aRk5NDSUkJr7zyCqNHj67TZQsLC7n//vuJi4vDxsaGt99+m8GDB7Nr1y7uuusuiouLKS8v5/vvvycwMJBbbrmF1NRUysrKeP755xk/fvwVvWxozC10gA5jwcaBG/VyMk8Vs18GGgkhLmH8+PHMmTOn8vGcOXOYNGkSP/74I1u2bGHZsmU88cQTde4998EHH6CUYseOHcycOZNJkyZRWFjIxx9/zCOPPEJ8fDxxcXEEBwezaNEiAgMD2bZtGzt37mT48OEWeW2Nu4Xu6AHtRtJq/6/Ycx3rDmTRxs/V2lEJIWrrIi3p+tKlSxeOHz9OWloaGRkZeHp64u/vz2OPPcbKlSsxmUwcOXKEY8eO4e/vX+vzrl69mqlTpwLQrl07QkND2bdvH7179+bVV18lNTWVsWPHEhkZSUxMDE888QRPPfUUI0eOpH///hZ5bY27hQ7Q9Q7MRSe402Uz6w5IHV0IcWnjxo1j7ty5zJ49m/Hjx/Ptt9+SkZHB5s2biY+Px8/Pj8LCQotc69Zbb2XevHk4Ojpyww03sHTpUtq0acOWLVuIiYnhueee46WXXrLItRp/Qg8fCL5R3GVeyLoDmTwxZxsxf1/MerlJKoS4gPHjxzNr1izmzp3LuHHjOHnyJL6+vtja2rJs2TKSk5PrfM7+/fvz7bffArBv3z5SUlJo27YtSUlJtGrViocffpjRo0ezfft20tLScHJy4vbbb+fJJ5+02CyOjT+hKwW9HiCw6ABRxdv4ZUcatjYmnv9pJyVl5daOTghxFerQoQN5eXkEBQUREBDAbbfdRlxcHDExMXz11Ve0a9euzud84IEHKC8vJyYmhvHjx/PFF19gb2/PnDlziI6OpnPnzuzcuZM777yTHTt20KNHDzp37sw//vEPnnvuOYu8LmWtYfOxsbE6Li7OMicrKUS/G80xlw7Y3jGHLSknuOerOJ4b0Z4p/VtZ5hpCCIvYvXs37du3t3YYjUJN75VSarPWOram4xt/Cx3A1gEV+2f8jy3Hq/Aw17b3ZVBbH979fT/H8yxTBxNCiKtd00joAN3/DGY72PARSimeG9GeU0WlzN+Wbu3IhBCN3I4dO+jcuXO1n549e1o7rPM07m6LVbn4QswtED8DBj9LhG8Lwr2dWb0/gz/3C7d2dEKIKhrbIMCYmBji4+Mb9JqXUw5vOi10gN4PQEkBbP4CgP6R3qxPyqaotMY1q4UQVuDg4EBWVpZMe30RWmuysrJwcHCo0/OaTgsdwK+D0Y1x46fQZyr9I334al0yW5JP0Lu1l7WjE0IAwcHBpKamkpGRYe1QrmoODg4EBwfX6TlNK6ED9H4QZtwCu+fTK+JGzCbF6sQMSehCXCVsbW0JD5cyaH1oWiUXgIih4B4CW77E1cGWriEeMrWuEKJZaHoJ3WSCrndA0nLIPki/CB92HDlJTn6xtSMTQoh6VauErpQarpTaq5RKVEo9fYFjblFKJSildimlZlg2zDrqfBsoE2z5iv5tvNEaVuyTep0Qomm7ZEJXSpmBD4DrgShgolIq6pxjIoFngL5a6w7Ao/UQa+25B0HkMIj/lk4BzoR6OfHVukNWDUkIIepbbVroPYBErXWS1roYmAWcO/P7PcAHWuscAK31ccuGeRm6TYZTxzAnLuGuPmFsSTnB5uQca0clhBD1pjYJPQg4XOVxasW2qtoAbZRSa5RS65VSNc7WrpS6VykVp5SKq/cuSxFDwTUAtnzJuNiWuDrY8Pnqg/V7TSGEsCJL3RS1ASKBQcBE4FOllMe5B2mtp2utY7XWsT4+Pha69AWYbaDL7ZD4O86n07m1Rwi/7kzncHZB/V5XCCGspDYJ/QjQssrj4IptVaUC87TWJVrrg8A+jARvXV3uAK1h6zdM6hMGwHebU60bkxBC1JPaJPRNQKRSKlwpZQdMAOadc8xPGK1zlFLeGCWYJAvGeXk8Q6H1YNj6DYFudsSGtmDJrqPWjkoIIerFJRO61roUeAhYDOwG5mitdymlXlJKjao4bDGQpZRKAJYBT2qtr44lg7pOgtxUOLCMYR382HM0j5QsKbsIIZqeWtXQtdYLtdZttNattdavVmx7QWs9r+JvrbV+XGsdpbWO0VrPqs+g66Tt9WDrBPuXMDTKD4AlCdJKF0I0PU1vpOi5bOwhtA8kLSfUy5l2/q4sSThm7aiEEMLimn5CB2g1CDL3Qm4awzr4E3com6xTRdaOSgghLKp5JPTwgcbvgysZFuVHuYY/dlt/7JMQQlhS80joftHg5AVJy+kQ6EawpyMLd8rSdEKIpqV5JHSTyWilJy1HASM6BrB6fyYnCmQGRiFE09E8EjoYdfS8dMjcx8iYQErLNYulT7oQoglpRgm9oo6etILoIDdCWjixYLuUXYQQTUfzSegeocZkXambUEoxsmMAaw9kkS0LXwghmojmk9CVguBYSN0EGHX0Mim7CCGakOaT0AGCYiHnIORnERXgRri3Mwu2p1k7KiGEsIjmldCDuxu/j8ShlGJETADrDmSRKYOMhBBNQPNK6IGdQZkhNQ6AkZ0CKNewaKeUXYQQjV/zSuh2zuAXVVlHb+vnSmsfKbsIIZqG5pXQwaijH9kM5eVG2aVjIBsOZnM8r9DakQkhxBVpfgk9uDsU5ULWfgBGdgxAS9lFCNEENMOEHmv8rii7tPFzpY2fiwwyEkI0es0voXtFgmMLOLS6ctOImEA2HcrmWK6UXYQQjVfzS+gmkzGvy4GlxgLSGIOMtIaFO6SVLoRovJpfQgeIuAZOHYPjCcZDXxfa+bvyi5RdhBCNWPNM6K0GG78PLK3cNLJjAHHJOaSdOG2loIQQ4so0z4TuHgQ+7aol9BtiAgApuwghGq9aJXSl1HCl1F6lVKJS6uka9k9WSmUopeIrfqZYPlQLaz0EktdCidEib+XjQnSQGz9uPWLlwIQQ4vJcMqErpczAB8D1QBQwUSkVVcOhs7XWnSt+PrNwnJbXegiUFkLKuspNt8S2ZFdaLjtST1oxMCGEuDy1aaH3ABK11kla62JgFjC6fsNqAKF9wGxXrewyunMQDrYmZm5KsWJgQghxeWqT0IOAw1Uep1ZsO9fNSqntSqm5SqmWNZ1IKXWvUipOKRWXkZFxGeFakJ2zMWq0Sn90d0dbbogJYF58GgXFpVYMTggh6s5SN0XnA2Fa647Ab8CXNR2ktZ6utY7VWsf6+PhY6NJXIKwfpG+DwrMllok9QjhVVCojR4UQjU5tEvoRoGqLO7hiWyWtdZbW+syk4p8B3SwTXj0L6we6HFLWV26KDfWklbez9EkXQjQ6tUnom4BIpVS4UsoOmADMq3qAUiqgysNRwG7LhViPgrsbdfRDqyo3KaWIDnInKfOUFQMTQoi6u2RC11qXAg8BizES9Ryt9S6l1EtKqVEVhz2slNqllNoGPAxMrq+ALcrW8bw6OkCYlxNHck5TXFpupcCEEKLubGpzkNZ6IbDwnG0vVPn7GeAZy4bWQML6wcppRh3dwR2AEC9nyjUcOXGacG9nKwcohBC10zxHilZVQx09zMsJgENZ+daKSggh6kwSeg119FAvo1WeklVgraiEEKLOJKHbOkJQN0g+O2LU28UOJzuztNCFEI2KJHSAkF6QHg/FRotcKUWolzPJ0kIXQjQiktABQvpAeamxeHSFMC8nkqWFLoRoRCShA7TsDqhqE3WFeDlxOPs0ZeWa08VlnDxdYr34hBCiFiShAzh6gm9UtYQe5uVMcVk56SdPM3XmViZMX3+REwghhPVJQj8jtDcc3ghlxqRcoRVdF5fuOc7vu4+xOz1XVjMSQlzVJKGfEdIbik/BsZ3A2a6L7/y2D6WMQ9YkZlorOiGEuCRJ6GeE9DJ+VwwwCnBzwM7GRE5BCeO6BePtYicJXQhxVZOEfoZ7MLiHVNbRTSZFSAuj7HJ3v3D6RnizOjELrbU1oxRCiAuShF5VSC8joVck7X4R3oyICaCdvxt9I7zJPFXE3mN5Vg5SCCFqVqvJuZqNkF6wYw7kHIQWrXhxVIfKXX0jvAFYvT+Tdv5u1opQCCEuSFroVYX2MX6nnN9FMcjDkVbezlJHF0JctSShV+XdFhw8IHltjbv7RXqzPimbwpKyBg5MCCEuTRJ6VSZTRR295kFEg9r6cLqkjE2Hshs4MCGEuDRJ6OcK6Q1Z+yH//NJK71be2NmYWLYnwwqBCSHExUlCP1dIb+N3lWkAznC0M9O7lRfL9x5v4KCEEOLSJKGfK7AzmO0vWHYZ3NaHpMx8DmXKTIxCiKuLJPRz2dhDcCwkr6lx96C2vgDSShdCXHUkodcktA+kb4PC3PN2hXk708rbmeX7pI4uhLi61CqhK6WGK6X2KqUSlVJPX+S4m5VSWikVa7kQrSC0r7Fw9OENNe6+pr0vaxIzOShlFyHEVeSSCV0pZQY+AK4HooCJSqmoGo5zBR4Bas6CjUnLHmCygUOra9x9z4BW2NuYeWVBQgMHJoQQF1abFnoPIFFrnaS1LgZmAaNrOO5l4F9AoQXjsw47ZwjsesE6uq+rA1OHRPDHnuNSSxdCXDVqk9CDgMNVHqdWbKuklOoKtNRa/3KxEyml7lVKxSml4jIyrvIadFhfSNsKxTWXVe7qG064tzMvL0iQGRiFEFeFK74pqpQyAW8DT1zqWK31dK11rNY61sfH50ovXb9C+xkLR1+gjm5nY2JK/3AOZORzKKuggYMTQojz1SahHwFaVnkcXLHtDFcgGliulDoE9ALmNfoboyE9QZnhUM1lF4DY0BYAbEnOaaiohBDigmqT0DcBkUqpcKWUHTABmHdmp9b6pNbaW2sdprUOA9YDo7TWcfUScUOxd4WATnBo1QUPifR1wdXehi0pktCFENZ3yYSutS4FHgIWA7uBOVrrXUqpl5RSo+o7QKtqPQRS4+D0iRp3m0yKziEebEmpeb8QQjSkWtXQtdYLtdZttNattdavVmx7QWs9r4ZjBzX61vkZEdeCLoOk5Rc8pEuIJ3uP5nKqqLTh4hJCiBrISNGLCe4O9u6Q+PsFD+ka4kG5hu2HpZUuhLAuSegXY7aB1oMg8Y/KdUbP1aWlJ4DU0YUQVicJ/VIiroW8NDhe86hQdydbInxd2JJygqLSMhbvOsrk/22k68u/kZwlUwMIIRqOJPRLibjW+H2JssvqxEy6vfw7f/l6M7vTc8k9XcKXa5MbKEghhJCEfmlugeDbAfYuuuAhIzoGEuThyI2dAvh8cixrnhrC9TEBfLf5MAXFcrNUCNEwJKHXRsdbIGXtBSfrGtjGh2V/HcRrYzsypJ0fNmYTk3qHkldYyk9b0xo4WCFEcyUJvTZ6/gXcgmDJ8xe8OXqubqGeRAW48dW6Q5SXy1wvQoj6Jwm9NmwdYfD/QdoW2PVjrZ6ilGJSn1D2HM0j+sXF/OmjtaSdOF3PgQohmjNJ6LXVaSL4RsHy12v9lHHdWvLu+M6M6RJEXHIOK2SVIyFEPZKEXlsmM3S9EzL3womU2j3FpLipSxAvj47G1d6GhLTzl7QTQghLkYReF+EDjd8HV9bpaSaTon2AG7vTz0/oR08WUlpWbonohBDNnCT0uvBtD84+kLSizk9tH+DK7vTcajdI84tKGfzmcr7dULsWvxBCXIwk9LpQCsIHwMEVte7tckZUoBv5xWWkZJ9dDCMlu4DTJWVslvnUhRAWIAm9rsIHwqljkLmvTk+LCnAHIKFK2eVwRXJPqKEUI4QQdSUJva7CBxi/61h2ifRzwWxS1W6MpuYY3RiTMk5xurjMYiEKIZonSeh11SIcPEKMsksdONiaifBxqd5CzzFa6OUa9h7Ls2iYQojmRxL65QgfCAdXQVlJnZ525sboGYezT+PmYAMgXRqFEFdMEvrlaHs9FJ284NwuFxIV6Eb6yUKy84sBSM0poHtYC1ztbWrs0iiEEHUhCf1ytB4Ctk6we36dnlZ5YzQtF601qTmnadnCifYBbnJjVAhxxSShXw5bR4gcCnsWQHntBwXFBBkJfWtKDicKSjhVVEqwpyNRgW7n9VEXQoi6koR+udqPMrovpm6q9VPcnWxp6+fKxkPZlTdEW7ZwIirAjYLiMpKr9FEXQoi6qlVCV0oNV0rtVUolKqWermH/fUqpHUqpeKXUaqVUlOVDvcpEDgOzHeyeV6en9QhvwZbkHA5lGcn7TAsd5MaoEOLKXDKhK6XMwAfA9UAUMLGGhD1Dax2jte4MvAG8bfFIrzYObtBqEOz6CXLTa/207uEtyC8u47eEY4DRQo/wdcHOxsRP8UfQWlNYUsZfvo7j63WH6iNyIUQTVZsWeg8gUWudpLUuBmYBo6seoLWu2rR0BppHMbjX/VCQBR/2hB1za/WUHmEtAFiy6yjujra4OdjiYGvmiaFt+C3hGN9sSOHJudtZvOsYn60+iL7EFAOLdqbzyoKaF7AWQjQvtUnoQcDhKo9TK7ZVo5R6UCl1AKOF/nBNJ1JK3auUilNKxWVkNIG5wVsPgftWg3cb+H4K5By65FP83R1o2cKRotJyWrZwrNx+T/9WDGjjwws/72T+tjQ6BruTnFVA4vFTFz3f3M1H+GpdstxQFUJY7qao1voDrXVr4CnguQscM11rHau1jvXx8bHUpa3LOwLGfgpoSKhdPb17RSs92MOpcpvJpHj7lk4EeTjyp27BfHJHNwB+3338oudKyjxFcVk5mflFlxe/EKLJqE1CPwK0rPI4uGLbhcwCbrqSoBqdFuHg37HWN0h7hhsJvWoLHcDbxZ4VTw7mzXGdCHB3pEOgG3/sPnbB85SUlZNScXM17UThZQYvhGgqapPQNwGRSqlwpZQdMAGolrmUUpFVHo4A9lsuxEYiarTRhfHkxT7rDD3DvQAI93Y5b5/ZpCr/vqa9H1tScsg6VXPr+3B2AaUVpRZZr1QIccmErrUuBR4CFgO7gTla611KqZeUUqMqDntIKbVLKRUPPA5MqreIr1ZRFV9KajF6NMzbmR8e6MPYrufdiqjm2va+lGtYtrfm+w1JGfmVf0tCF0LY1OYgrfVCYOE5216o8vcjFo6r8fGOMBaRTvgZet13ycO7hnhe8pjoQHd8Xe1Ztuc4f+oWfN7+AxnGDVM7s4kjktCFaPZkpKglRY2GlHWQttUipzOZFH0jvNlwMKvG7otJGfl4u9gR6uUkLXQhhCR0i+o+BdyDYeatkHfUIqfsGd6CzFPFHKhSXjkjKfMUrbxdCPRwlJuiQghJ6Bbl7A0TZkDhCZh1G5QWX/Epe1T0iNl4MPu8fUkZ+bTyca5I6NJCF6K5k4RuaQEdYfQHcCQOtnx5xacL93bGx9WeDQezqm0/UVBMVn4xrX1cCPJwICu/mMISWcZOiOZMEnp96DAGQvvCymlQfGUzKCql6BHegg1J2dXq6GdKMGda6ADpJ6XsIkRzJgm9PigFQ543ptfdOP2KT9crvAVHcws5nH22rJJU0cOllY9LZUKXsosQzZsk9PoS2hsihsKad+H0iSs6Vc9WxkCkM2WXkrJyth4+ga1Z0dLTkaCKhC5dF4Vo3mrVD11cpmtegOmDYOFf4ebPLvs0ET4ueDrZ8u8/9vNdXCp7juaSW1hK55Ye2JhN+Lk5oJTRQi8v12TlF+Pjam+51yGEaBSkhV6fAjrCoGdgx3ewbfZln8ZkUvy5XzgeTrYoBcOj/fn49m7MuKcnAHY2Jnxd7Uk7cZqnf9hOn9f/YE1iJgCFJWXsPZpnkZcjhLi6qUvNt11fYmNjdVxcnFWu3aDKy+CLEXB0J9y/BjxD6+UyYz5cw6HMfHIKSnCyM2NWilfGRPPBskT2HTvFokf7087frV6uLYRoOEqpzVrr2Jr2SQu9vpnMMLbixuj8h6GePkADPRzJKSihQ6Abix4ZgLO9DY/MiienoASlYPHOC8/aKIRoGiShNwSPEBj6IiQth63f1MslWns7Y2dj4p3xnQnxcuKbKT14aHAESx4dQJeWHvy2u/rI1bJyzexNKZwulr7rQjQVktAbSre7jb7pi5+12LQAVd0/KIKlTwykjZ8rABG+rvz1urZ4OtsxNMqfnUdyq3VrXLbnOE99v4NvNyRbPBYhhHVIQm8oJhOM+g+U5MOqtyx+ekc7M8GeTjXuGxrlB1BtsYxle42VkOZuTrV4LEII65CE3pC8WkOnibD5S8hruJp2ax9nwr2dWZJgXFNrzfK9GTjYmthzNI9daScbLBYhRP2RhN7Q+j0G5SWw9r0Gu6RSiqFRfqxPyuLk6RL2Hz/FkROnefiaSOzMJr7fXPMqS8dzCykqlRq7EI2FJPSG5tUaYsZB3OdwquaViOrDqE6BlJRp/v37fpbtMcotN3UO4pr2vvwcf4SSsvJqxxeWlHHt2yv4zx+JDRajEOLKSEK3hv5/hbJi+HQw7P+9QS4ZHeTO7b1C+GLtQb7ZkEw7f1cCPRy5uWswWfnFrDhnmbu4QznkFpZW1tqFEFc/SejW4NMGJi8EWyf49mZY/U6DXPZvw9vh42rP4ezTDGrrC8DAtj54ONkyf3tatWNXJRoJPiE9l5z8K5/XXQhR/yShW0tIT7hvFUTfDL+/CHt+qfdLujnY8vLoaMwmxfXR/gDYmk1cHx3AbwnHqvVJX7UvE08nW7SG9UlZFzqlEOIqIgndmmzsjcUwArvC9/fA0R31fslhHfzZ8eIwOrX0qNx2Y6cACorLWFpRW888VURCei6T+oThbGdmzYHM886jtWbP0VwOZ1/ZfO9CCMupVUJXSg1XSu1VSiUqpZ6uYf/jSqkEpdR2pdQfSqn6mbCkKbJ1NJatc3CHr8dC1gHI2AfTBxsLZNQDJ7vqk2z2DPfCx9We+duMssuZib0Gt/WlZysv1h4420IvLCnjs1VJXPv2Coa/u4oxH67leK4srCHE1eCSCV0pZQY+AK4HooCJSqmocw7bCsRqrTsCc4E3LB1ok+YWAHf+BLoMvhgJnw6BtC2w7gMoLar3y5tNihExASzde5y8whJW7c/Ew8mW6CB3+rT2Iikjn7QTp/k5/gjXvLWCV37ZTQtnO56+vh35RaU8NHMrpef0khFCNLzatNB7AIla6yStdTEwCxhd9QCt9TKt9Znv3uuBYMuG2SNZfdAAACAASURBVAz4tIU7fjRGknpHwI3vwekc2LeoQS5/Y6dAikvLueG9Vfy6I52+rb0xmxS9WxuLa4z+YA2PzIrH3dGWGVN68t19fbhvYGteGxvDxoPZvP3bvgaJUwhxYbVJ6EHA4SqPUyu2XcifgV9r2qGUulcpFaeUisvIaLg+2I1GQCd4dCdM+QO63A6uARA/o0Eu3TXEg5dviiY60J1QL2du6d4SgPb+bgR5OGJWijfHdWLB1H70ifCufN5NXYIY0TGAr9cnSytdCCuz6IpFSqnbgVhgYE37tdbTgelgzIduyWs3GQ5V5izvNAHWvGdME+DqV6+XVUpxR69Q7uhV/faHyaRY+Eh/7G1MONiaa3zuDdEB/LI9nW2pJ+gW2qJe4xRCXFhtWuhHgJZVHgdXbKtGKXUt8CwwSmtd/4Xf5qDTrUZdffvlr3ZkCe6OthdM5gD9IrwxKVix7/zeMEKIhlObhL4JiFRKhSul7IAJwLyqByilugCfYCRzGVpoKT5tjCl311bU069S7k62dGrpwYp9UkYTwpoumdC11qXAQ8BiYDcwR2u9Syn1klJqVMVh0wAX4DulVLxSat4FTifqavjrUJANv71g7UguakCkD9tTT8ioUiGsqFb90LXWC7XWbbTWrbXWr1Zse0FrPa/i72u11n5a684VP6MufkZRawEdofeDsOUrOLTG2tFc0IA2PmgNqxOl7CKEtchI0cZg0DPgEQrfTYb07daOpkadgt1xc7BhpZRdhLAaSeiNgZ0T3PYdmG3hixFwaLW1IzqPjdnEwLa+/LIjnW2HT1g7HCGaJUnojYVPW7h7Mbj6w1c3GYtNaw1HNhs/V4Fnb2iPl4sdk/63kb1H887bv+lQNu8v3X/e3OtCCMtQWlunO3hsbKyOi4uzyrUbtYJsmHsXJC0HtyDIrehBOugZGPgUKGXV8FKyChj3yVoUit8eH4Crgy1Zp4p4cX5C5VwxTwxtw9RrIq0apxCNlVJqs9Y6tqZ90kJvbJxawG3fQ7/HwTfKWHi600RY/hrMuhV2z4eUDTDvYfh8eIOuXQoQ4uXEx7d342huIf9ZmkhpWTn3f7uFxbuO8vCQCEZ0DOC9pftJSMtt0LiEaA4sOlJUNBCzDVz797OPu9xhlGRWvQ17FxrbbJ1Al8OcO2HSfLCxM7aXl0NeGrjX33Q7XUI8GR/bks9XHyTzVBEbD2bz1rhO3NwtmBMFxWw8mM3jc+L56cG+Fx2wJISoGym5NCVlJXB4A+SmQZvrIPEPozwTfbMxT0zGXti/BPIzYOjL0Pfhegsl61QRg99cTm5hKTd3DeatWzpV7lu65xh3fxHH8A7+fHBbV8ym6mWiFfsyeP6nnTjammkX4MrfhrcjyMOxVtdduCOd0nLNqE6BFn09QlwtLlZykRZ6U2K2hbB+Zx9HjzUWzVj9Nuz8HhxbQOvBUJQHvz0Pjp7Q9Y56CcXLxZ5/jo3hp61HeGl0h2r7hrTz44WRUby0IIG/z9vJy6OjUUqhtWb6yiT+tWgPrX1cCPZ0ZNHOo5iU4p3xnS95zYLiUp7+fjuOdmZu7BiAsvL9BCEamiT0pu7av0PsXcYCGg7uxrbSYpg5AeY/DLt+hMhhxuyO9i4WvfTIjoGM7FhzS/nufuEczyvi4xUHOHm6lH+M6sCL83Yxb1saI2ICmDauI052Njz/005mxx3mhZFReDobZaPycs03G5KJCXKnS4hn5Tl/2ppGbmEpuYWlHMzMp5WPZV+PEFc7uSnaHHiEnE3mYNTTx38NvR+CEymw6Cn4/DrISb6885cUwum69z1/anhbnhrejgXb0+j1zz+Yvz2Nvw1vy/u3dqlcVWlijxCKS8v5YavRm6ewpIyHZ23lhZ93MXXmVopKjXVQtdZ8ufYQfm72AKxPyr681yJEIyYJvbmyc4ZhL8PUOKPXzMnDMH0QpFbc1ygrgd/+bqx1uvBvcHx3zecpKzU+DD67xvi7DpRS3D+oNV/c1YMOQW58Pqk7DwyKqFYqiQp0o3NLD2ZsSOZgZj63fbaBBdvTGdsliNSc03y9zvgQWp+Uzd5jeTwxtC2+rvaysLVoliShC4i8Fu5ZZrTivxkL6dvgh3thzbtweD1s+RJmTjRa4uda9z6kx0NWIuy+vDnZBrbx4ccH+jK4nW+N+2/tGcKBjHyGvr2CfcfyeP/WLrw9vjP9I715f1kiaxIz+efC3Xg62TKqcyC9WnmxPimLmm74FxSXsj317LeJVfszGP/JOk4Xl11W7EJcTSShC4NXa5g0D+xcjDVNd/0AQ1+CR3fAxFmQc9CYxre83Biluvod2LfY6P/edgR4RcCafxujVy1sZMcAWvk4M6yDH388PrCyLv/U8HacKCjhts82cDingBdHdcDB1kyvVl4czyviYGY+QLXE/vqvexj1/hoWbE8j61QRj82OZ8PBbBLST1o8biEamtwUFWd5hMAdP8HM8dBxAvR9xNjeejB0GAOr3oKDK+HQqrPPsXeDEW8Z3SHnP2zsb1XjglWXzcnOhqVPDDpve3SQOy/eGEVhaTm39wrFxd74z7lXK2PVpAXb09l0KJsTBSX89GBftNb8sj0dpeCJOduICXInu2K6333HTslqS6LRk4QuqvNpAw9vPX/7df+EfUuMeWNGvgttr4ekFeAZBm4B0HE8LH0F/vgH+Mww5pw5V2kRnEw1vg1YyOS+4edtC/d2xtfVnrd/24dSxpeGZXuO42BrJiu/mFfHRPPR8gPEJefwzPXtePf3/ew7dv7cM0I0NpLQRe24BcKU342bqZ4V6452Gn92v62DkfR/fhDe7wHD/2l0hQRj/pl1H8DmL6AgE254E3rcU2+hKqUY0TGAFfsyeHd8Z/7y9Wb+t/YgLT2dcLYzc3PXYPq09mbZnuNM7hPGgu3p7D92qtbnP1lQwp6juRw5cZr+kT74uNrX22sRoi4koYva84u6+P6O4yCwC8x/xEjsWYnQ+TaYcQtkHzRa9WUlsPCvRjfHPlOND4K6KCkEG/vzJyErKwGUMS0C8MLIqMreMnf0DuWNRXtxtD3BdR38cLA1E+7tTHg/o3Uf6efC6v3GwhwnCop5c8le2vq5MqCND6FeztUus/9YHmM+XMupIqNHz/XR/nx0e7e6vYbavtSycrannqRriIcMkhK1IjdFhWV5Rxg3V7vdZdw4/bA3FJ6EuxfBxJnGT8w4WPYKvBEOs2+HrAO1O/fJVHg3Bn79W/Xt5eXwv+th7uTKTVUT4MTuITjYmjhdUsaIGgY6tfFz5XheEScLSvhx6xG+WZ/C8z/vYtCby1my62jlcYUlZUyduRV7GxP/m9ydyX3CWLTrKPvPKddsOpTNwh3p5BaW1O511aC0rJxHZm3l5o/W8uHyWr4/otmThC4sz2SGke/ANS9AcHeY8geE9DL2mW1hzHSj73vn24ybqJ8MhO1zLt5DprwMfvgL5B+HjdOrL/KxYw6kboI9v8DJI+c91dPZjnHdWuLtYseACA/jg6GKNn7GiNJ9x/NYuuc4rXycWf7XQbT1c+XFebsoKDZa4/9atIc9R/OYNq4jg9v58vA1kTjYmKsl3ILiUu7+YhMPfLuFri/9xovzdlX2svlp6xH+tWhP5fnOlX7yNNNXHmDhjnSe+n4HC3ccpa2fK28u2ctvCceMexCr34FCmamy3uVnNfhMpZYgJRdRP5SC/k8YP+cymYy+75HXGj1pvp8CP9wDy16FtjdAQZZRookcBj3/AmY7WPUmJK826u9r/2OUde5bA2j442Vo0RqyD8C2mTDgr+dd8vmRUTw+tA32Cx4ypjt4YH3lzdk2fq4AbDt8gg1J2dzZO5Qwmyxmuv2HPx/rw9tLAigpK+fLdclM6h3KkHZ+ALRwtuP2XiH8d/VBHr02klAvZ+bFp5FXWMrLozsQf/gkX6w9RKSfC6EtnHniu22UlWsW7zrKexO6EB10dvRuTn4xt326gaSKrpYAr3c/zS2H/48pfn/n4Zlbedw/nnsyXqMIO+z7PVjj2344uwBfN3vsbZrxLJYlp2HFv6DXA+BS89iGS5pzh3Hv54F1Vl9joC6khS6sy6MlTP4FbvrISMobPjFa3+WlRlnm7fbwWjCsnGbMGtl9itH6z0qEL26AuXdDbirc+C6E9Yf4b2ts6dvZmPDM2AjbZ0NZsdEbp0KQhyPOdma+WpdMcVk5Q9r6wM8P4ZmyhFkOr3Fg7Q98uS6ZKf3CeW5k9fsI9/Rvha3ZxEvzEygv13y1Lpl2/q7c3iuUaX/qyMA2PvxjXgIPfLuZCB8X/jspltPFZdzx3w0UlhiDmQpLyrj36zhST5zmmz/3ZP5D/fjuvt6ML5qD6cQh3mu7g2uj/IjJ+R2AY3E/VX9xWnPkx+d45qNZ9H9jGS/OSzj/fT6wDNJq6L1kaaXFsOkzyk6mc/NHa5mxIaX+r3muTf81vsls+uzynn8iBZLXQMbu+nnPdn5vTJBXD2qV0JVSw5VSe5VSiUqpp2vYP0AptUUpVaqU+pPlwxRNmtkGOt8Kd/wAz2fA4wlw7zK4d7nR/733gzBhBoz5xGgtRVwDw/9l3Ajd+6sxsCl8gNGrJjsJ9iyA5a/Dr0/DsV3GNcpK4Je/Gn3t+z4KCT/D4Y2AUW+P8HMlJbsAF3sbeuTMh4MrYMhzmHzb8pndW/x4zUmeGxmFrdl09nyr38F3wWTWuz9L98R3eWzmRhLSc7m9VyhKKUwZCXzYYTe3O66hrUrh0ztjuaa9H2/8qSM5BSVGGQWYvmQz+clbeWtcJ/pFehMT7E531xzUvsVgssFlz3f8Z3QovcrjKVIOBJ7YTEl+TuXbl7Z5AUHb/sMNx6cTG+rJnLjDHKpo6e89mkfG9iXwzc0w89aaR/vWwamiUr5Yc7ByDp0zTheX8f7CTRR9eRP88gRF393D5uRsXvh5J5uTcy5wtovIToJF/2f8m537AZ11AOL+Z5ThzlVcYAxwAyNxVjz3REFx9ZHD8TPh+J7znj57Uwo7l3xhPDDZGg0AS0rbajRC4v5n2fNWuGTJRSllBj4AhgKpwCal1DytddVmQAowGTj/u64QdWGqUioI7AKj36/5uF73GT+FuWBbMVd6+1FGAphd0V3SbA8bPjL6ypeXGfPVTJhpJP9tM41Sj70r5CTzocmfbbaeBLjYYfN7vHFM/79i0+Mv8OWNdNn0JHTsbPT0Kcg2Fg45tAp82uPhE8h9+QvYvncX2fZ3cVPHIbD2ffj97ziXl/ICoJUJFX8UOoyh74GZ/NdpC4eXx3I6x4NJG9/iYft8UL7AOCP29R8ZpaahLxuTpy18EspLSO7yN9psfZW45d8TO2IKANlL3yMQ6K83EzXCg2s+PY7psyHkuLfgwyPdedH8OafsPHDJS4MtX0HPeyvfxrl/rMW0fzFjotxQaHDxM7qlBnUzuqieY/qKA7y3NJHScs2U/q0AKCvXPPntah49eB825gzoMAanXT9yo6kTK+0H8tCMLSyY2g+vU/uN8QnO3hf+9y8rgUVPG61sKhJwp4kQ3M34N1z/oTHeobTQaOX2fdgosez5BVoPMRJw/nFjYNz2WXB0OxsLW3L7ZxsY2zWI18bGoHbPg5/ugxat4P51lT2tyso1r/26h+/4yXj97sGwYy4Me8W493M5ysshaz94tzEaI2v+bQzG6zbp8s53CbWpofcAErXWSQBKqVnAaKAyoWutD1Xsk9V/RcNycDv7t52TMV3w0e3Q60Gjfhr/LRzZYvzP1ONeo+ukUkaf+cXPGlMWhPSmLCmBNgUH8TY5Q3CssbSfUsb5J840Ji6bOR5C+kDSMjidAzd9DJ0nooDCHT8T/sMDfK1fgLf/aSScdiPhWqO0o1a/Y9wHWPUmJpMtPWw8uSZnLSyH9eWd6BFkh/NP9xkzYboFQfwMiP4TdJsMy/8JO+eCZxgRIx4nZ+v7FOyYDyOmsHfnFqILNrLN/2Y6HZ+H166v+CToBCFH95B12o1/26zltMmFG079H5+4f0mbVW8Zc+DbOlJw7AADV03EhxNwtPrbqk025Hu0xSUoyvhg7XEv+aWKQ2vn8pvdt7y69H7Gd78PF3sbXpq/i1aJXxFhm8Y7fm/w2M1TSD+0hxf01xy97V5u/iKBX79+g9uPv218Q7prkTEYrUJ5uaZca2yKTxoflAdXQs/7IPbPMH0gbP7cSOjL/mm8h22uB10GS1+Glj3g9xchZR3YOoPJxii9DX8Nds4lL24298cPxsasmLXpMJ08ipgY9yi4hxjfAla/A4OfAWBX2kk8T6cQaX+A0qhXsfFqZXyTO7DUWDAGIHO/MbguZlz1xkcVuYUluJlKjA+fLV/CiRTK+z7GnLLBjE/4GdVnavXZTy2oNgk9CDhc5XEq0LNeohHiSp07YKnP1JqPi/mT8VOh+Hgez/ywg+l3xELFvOuV3AKNks+XNxr/cwd3h/6PG4m/gkPMaBwiBhrJ/uBK8IuG2LvP3lC76QOIHmPc7I26iZxCJ258aybOFOHbpjuDJ7aFL0YYCQ2MxNTrfqP1GDPOqAd3GIvJxobjAYPolPYbz36/lT773yIcG1qPexmWm2HLl/QoOc1sPZRv3P/CrF7JOLfsxLCdLvx99Qlm2r5szKLZ/kZKvnsAW13CJNt/kWIO49dH+mHKz+D3VatI3fYH7TMS6VG8Fvsd38G+xaxzHc00/Tb2plLeKX+dmQsj2VsawO9b9rDB+Ve2O/ZjRlZrHlUmPnR5kBfzH8bnu34s8Y4i7Nh6cv164Jqzm6wPhzMzYhpD+/Zg77F8/vXrHsY6b+eJkumogkyjtNZpgvE+RN9slE66T4G171EePQ7TzZ9CfiZlH/TE/Pl1aJMtavjrxqRye36Bwc+CUwtKwgZxeusc/Mva8HXXPew8kILnigTKbPMx370IVrxhLP4SMw68I1i1P5ORpnWUa8VBv6FEhoUbi8Js+BhC+3I6fQ98MwbH0lzSf3+fsuteIzg4DIpyjQ+UnGR2OvXgb78e4Uff/2Kfs8/4cPHviGnNO3Qun0u52Yy51wNX8B/4xV1yCbqKmvhwrfWUisd3AD211g/VcOwXwAKt9dwLnOte4F6AkJCQbsnJlzn/thDWUFpklEEs1Oth/Cfr2HAwm7n39SY2rIVRytm70PhK7htl9OkHo9Y7+zaYOBu8Izi9/Wccf7iTdO2FNyc4FHg9kX/5FlI3w2dDwC2Y9NuW4e7ZonJe+X3H8hj2zkpWtpxOSMZyAAqx4x+e/2T0jWOZMH09/SO92Z2eR+apIq5t78u21JOEtHBibq8kmP8oqryEFJtQQu76gtzPx5Jbaub5kru4P/w43Y98xbze3/HIsmLWPD2EG/+zmikhx3jAfR1632IWFHbkE7epDHM7xL3JT+KgSijStpzAGa1s8CeTUx5tcRn3kVHuOOPIZmOyOAcPSkqKGVb2Dp/cP4IIHxdee+/f3JH9Pmn9XqPXsHHV3tvE46eY/dkbPFtcUU832VLu2ILsghLe1RO4+8FnaeWQD+93N74VDX2ZOX+s4aa8mWwqa0vG2O+4qUuQMX/RHy9R7OBNUWEBJ7UTPzuO5dbTM/BU54wuVmbjmwNQYNsCp/GfGfd7ykrZ9db1dCjYyEbPkfR45Nsr+u/mSpegOwK0rPI4uGJbnWmtpwPTwVhT9HLOIYTV2Fh2iP/T17djXVKWkcwBnFqcnS6hKt92MHVz5UPH9tdB10n4lxRQaHIiYtDjxo7gbjD8dQjpTYBf9e56bfxciQpw4xH9BD9OfZstW9bz5NJ8nhgykl6tvBjRMYCFO9IZ0taXu/qG0y/Sm5kbU3jmhx18FzuQ9MA3CEv+Du/RbxASFEPu2Bk4fn87/1PTjGzQYSxhUd1h2RoW7TxKdn4xzpH9oM/tKEBtT2PnjK3sPOqLb/+Z3OiexMHEBFx1HsGuZv6z25FVtmOZHdiVah+XgV3BvyMc3c4C779wMNWV+77ZzK09Qvj0aCT/Vf/mppwgelV5yvG8QsZ+uAZ3UyxTA/ri1roH9Lwfk6sfp7MLWPjBGtZ+FcePD/TF/a5fYP6j8NN93ALsbDGUx4/dzE1Hc7mJIOj/BNm+PTg48294qVzyxs3iwQ4xHD78EP/89BO6BDpza59I41ubix/T//shRUd2sNFzLF9HXANAXonm9tz7mUwwmxnDV1f4383F1CahbwIilVLhGP90E4Bb6zEmIZqFLiGe1ZbQqzVbBxj1Hgo4b+nsXvdf8GljugTx6sLd/JrmxD82+WPjoRgWZfSpf/uWTrx4Y4dq89KM6xbMf1cf5G/fb8ds8uexa99jVGfjW0Nwhz7Qdg/smQ97FsI1z9Pe1Q07GxMzNhjfvDsEnr2/MSImgLg+OUT4ujChlzEXUHSV5W891iez8aedrE7MpH+kz9kdSsHgZ9Hx3/BW4hDa+DmSePwUr/yym57hLQhwd2DZ3uOUlevKxcYX7TxKbmEpMx8eiFvgqGrvQcsWTnx0W1du+2wDLy9I4M1xneDPv5Gw+FP+tSqLe2+4F88FCew9anQr1Frz+Fp71pW+wC9T+xLjZ7ymli1D8e17J/+37AAxI/sR08KoiX+T143UsvaUpxsDxQLcjXVxc0rtWR12D9tST1JSVn62t5SFXfKsWutS4CFgMbAbmKO13qWUekkpNQpAKdVdKZWKcYv+E6XUrnqJVghx2UZ1DkQpuP/bLQBMvyMWm4rEYm9jPm+SMRuziVdviqZvhBff39+Hh4ZEVp9TxsbOqHH/6b/gGYadjYnoQDcOZOSjFLQLOJvQlVK8OKoDt1ck83PdEhtMgLsDD8/cyoPfbmH+trSzO9sOJ+26T0nNK+PWHiH83w3t8XCy5dUx0Qxp70dOQQnxh88uWrJ411Fa+zjTIbDmG489W3kxuU8YP2xJNaZtMJmYU9KPDeYudAv1pH2AW2VCn7ExheV7M3jm+nZE+LlVO89fBramhbMdr/1qrOZ1qqiUlOwCRnUyppf4vaJb6s/xaYR6OXFrT2M5xaSMfOpLrUaKaq0XAgvP2fZClb83YZRihBBXKT83B26ICSAlq4Dpd3YjwP289v15erby4ttWXrW+RueWnmxJOUGYl3Pl/PS1YW9j5oPbuvK/NYfYfCibX3akcyy3sLJr5Jm+7LFhLYgOcmdynzBszCZ8XBwwmxRL9xyjW6gnJwtKWJ+Uzb0DWl30eg8MjmDWpsO8uWQvDw2O5Kf4I/Rq5YWDrZm2/q78uPUIx/MK+ffv++kR1oI7e4eddw43B1vuH9iaVxfuZt+xPPIq5u4Z0TGQbaknWZJwjNY+Lqw9kMlDgyMqP2AS0k/S1t+11u9NXcjQfyGakf9M6ILJVH9D2TuHeMAaYy3Yuuoa4knXEE9KKiYme+WX3ZhNirv6hrP5UDaOtmbaVSTCM98s3J1s6RbqyR+7j/Pkde1YuvcYZeW6spR0IS2c7binfyve+X0fq/Zn4ulkx99v7ABQeY1pi/ZyPK+IaeM6XfA9G9U5kFcX7ua3hGO4Oxp91dsHuDI0yo9PVyWxan8mIS2cmNgzBB8Xe+xsTCSk5TKmS53fnlqRof9CNCP1mcwBuoZ4ABATdPn9rG3NJv49oQvDovx4eUECO4+cZHNKDp1belQm8qquaefLnqN5rE3MZPHOY/i52dMp2OOS1/lz/3B8Xe0J9nTk+/v7EO5tDKRq5298GH23OZV2/q4MiLzwQCg/Nwc6tfRgScIxdqfn4upgQ5CHIzd1DsLd0Zb7B7Vm8aMDCHB3xMZsop2/Kwnp9Te5mrTQhRAWE+zpxDd/7mm01K+ArdnEtD914pq3l/P0D9vZnZ7HA4NqXulqTJcgvt2Qwm3/3YBZKSb0aFmrDy4XexuWPDYARztztcnM/Nzs8XCy5URBCfcOaHXJueiHRfkxbfFeThQU097fDaUUUYFuxL8w7LxjowLcWLzrKFrrepnjXlroQgiL6hfpXaf6+YW4O9nyfze0Z+eRXMrKNV1Da+4R5OvmwK+P9OeOXqFo4KbOQbW+hoeT3XkzUyqliA50x9/NoXJB8osZWlHeSc4qoH3AxWvjUYFu5BSUcDT3yubUuRBpoQshrlpjugQxe9NhNh7KpmvLC3fxdLa34aXR0Tw7or1Fpg5+/eYYSss0djaXbvNG+roQ5uXEoayCaj17ahJVsT8hLbdWN6XrShK6EOKqpZTiP7d2ISEtF3enS0+QZal54IM9nWp9rFKq4iboQdpfIqG3C3DD3dGWnILLX83qYiShCyGuar6uDvi2rePasw1sct9wTEoRfYnePS72NsS/MLTe1oiVhC6EEFcoyMORZ25oX6tj63PBb7kpKoQQTYQkdCGEaCIkoQshRBMhCV0IIZoISehCCNFESEIXQogmQhK6EEI0EZLQhRCiibjkItH1dmGlMoDLXSXaG8i0YDj1oTHECI0jTonRMiRGy7B2jKFaa5+adlgtoV8JpVTchVa9vlo0hhihccQpMVqGxGgZV3OMUnIRQogmQhK6EEI0EY01oU+3dgC10BhihMYRp8RoGRKjZVy1MTbKGroQQojzNdYWuhBCiHNIQhdCiCai0SV0pdRwpdRepVSiUuppa8cDoJRqqZRappRKUErtUko9UrG9hVLqN6XU/orfF14UseFiNSultiqlFlQ8DldKbah4P2cr9f/tnV9ollUcxz9fNBcaNE2y5YRZjWJJpXThqIvon05ECbqYCBkJ3QT9QYjWIOgyisrALOgfxDBoWY1BSa2urYyco7kylNzQ5kUZ1I3St4tzhk8v78u62jnvy/nAgeec3/PCl++z3+953t95XqYlifW1SxqWdFzSpKTe3HyU9FS8zhOSDki6PAcfJb0jaVbSRGWtrncKvBb1jkvakFDji/F6j0v6WFJ7JTYQNU5J2pRKYyW2R5IlrYzzJD42oqkKuqRFwD6gD+gBdkjqSasKgIvAHts9wEbgsajrGWDMdjcwNgGLkQAAA01JREFUFuepeQKYrMxfAF6xfQPwO7A7iapL7AU+t30TcCtBazY+SloNPA7cbnsdsAjoJw8f3wM216w18q4P6I7jUWB/Qo1fAOts3wL8BAwAxBzqB26On3k91oAUGpG0Brgf+LWynMrH+thumgH0Aocq8wFgILWuOjo/Be4DpoCOuNYBTCXW1UlI6ruBUUCEX7wtrudvAn1XAieJm/WV9Wx8BFYDp4EVhH/hOApsysVHoAuYmM874E1gR73zFlpjTewBYCge/ye/gUNAbyqNwDDhIeMUsDK1j/VGUz2hcymZ5piOa9kgqQtYDxwGVtk+E0NngVWJZM3xKvA08E+cXwX8YftinKf2cy1wDng3toXekrSMjHy0PQO8RHhKOwOcB46Ql49VGnmXay49AnwWj7PRKGk7MGP7aE0oG43QZC2X3JF0BfAR8KTtP6sxh9t3sndEJW0FZm0fSaXhf7AY2ADst70e+Iua9koGPi4HthNuPtcCy6jz9TxHUns3H5IGCe3LodRaqkhaCjwLPJday3w0W0GfAdZU5p1xLTmSLiMU8yHbB+Pyb5I6YrwDmE2lD7gD2CbpFPABoe2yF2iXtDiek9rPaWDa9uE4HyYU+Jx8vBc4afuc7QvAQYK3OflYpZF3WeWSpIeBrcDOeOOBfDReT7iBH4350wl8L+ka8tEINF9B/xbojm8ULCFsmIwk1oQkAW8Dk7ZfroRGgF3xeBeht54E2wO2O213EXz7yvZO4GvgwXhaao1ngdOSboxL9wA/kpGPhFbLRklL43Wf05iNjzU08m4EeCi+pbEROF9pzSwokjYTWoHbbP9dCY0A/ZLaJK0lbDx+s9D6bB+zfbXtrpg/08CG+PeajY9zYptqAFsIO+G/AIOp9URNdxK+yo4DP8SxhdCjHgN+Br4EVqTWGvXeBYzG4+sISXIC+BBoS6ztNuC76OUnwPLcfASeB44DE8D7QFsOPgIHCH39C4Sis7uRd4QN8X0xj44R3tpJpfEEoQ89lztvVM4fjBqngL5UGmvip7i0KZrEx0aj/PS/UCgUWoRma7kUCoVCoQGloBcKhUKLUAp6oVAotAiloBcKhUKLUAp6oVAotAiloBcKhUKLUAp6oVAotAj/AjuiJpdHJRG9AAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "MWh_RrB-S9jx"
      },
      "source": [
        "We can see that both `loss` and `val_loss` are decreasing almost together and flattening out. This is a great improvment from our original model.\r\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "EGVnRhqMTT6Y"
      },
      "source": [
        "# Model Evaluation"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "c2cwiwTsTXCL"
      },
      "source": [
        "# for keras, the way it use BINARY CLASSFICATION prediction is as follow\r\n",
        "\r\n",
        "predictions = (model.predict(X_test) > 0.2).astype(\"int32\")"
      ],
      "execution_count": 123,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "LLfXeVVbTvSI",
        "outputId": "138b5b69-9f32-4d6e-8df0-146e425d8422"
      },
      "source": [
        "predictions[:5]"
      ],
      "execution_count": 124,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([[1],\n",
              "       [1],\n",
              "       [1],\n",
              "       [0],\n",
              "       [1]], dtype=int32)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 124
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "IUtlbPVGUIjB"
      },
      "source": [
        "from sklearn.metrics import classification_report, confusion_matrix, plot_confusion_matrix"
      ],
      "execution_count": 125,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ku5dd3dqUROb",
        "outputId": "a336682b-987c-4351-a581-7bd7839284db"
      },
      "source": [
        "print(classification_report(y_test, predictions))"
      ],
      "execution_count": 126,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.98      0.98      0.98        55\n",
            "           1       0.99      0.99      0.99        88\n",
            "\n",
            "    accuracy                           0.99       143\n",
            "   macro avg       0.99      0.99      0.99       143\n",
            "weighted avg       0.99      0.99      0.99       143\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "SU6hg5UcUVaS",
        "outputId": "a584e9d1-0093-44bd-fca0-a36659ddea3c"
      },
      "source": [
        "print(confusion_matrix(y_test, predictions))"
      ],
      "execution_count": 127,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[[54  1]\n",
            " [ 1 87]]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "FeAAurfxVZ6k"
      },
      "source": [
        "There are about 2 points which are misclassified."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-2GBpkcyVeSZ"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}