train_on_frame.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "train_on_frame.ipynb",
      "provenance": [],
      "authorship_tag": "ABX9TyOsloUINZUH9qzBYEsv+cMf",
      "include_colab_link": true
    },
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/giacomoaccursi/Tirocinio/blob/main/train_on_frame.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "9Q1b2kSohKm-",
        "outputId": "25ebfbdc-5b1e-4606-c977-0b1532514bdb"
      },
      "source": [
        "!git clone https://github.com/giacomoaccursi/Tirocinio.git"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Cloning into 'Tirocinio'...\n",
            "remote: Enumerating objects: 16840, done.\u001b[K\n",
            "remote: Counting objects: 100% (16840/16840), done.\u001b[K\n",
            "remote: Compressing objects: 100% (16833/16833), done.\u001b[K\n",
            "remote: Total 16840 (delta 5), reused 16829 (delta 4), pack-reused 0\u001b[K\n",
            "Receiving objects: 100% (16840/16840), 7.03 MiB | 10.90 MiB/s, done.\n",
            "Resolving deltas: 100% (5/5), done.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qIkFKr-2Km0B"
      },
      "source": [
        "import json\n",
        "import numpy as np\n",
        "import pandas as pd"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "gFryVgFP4j1f"
      },
      "source": [
        "from os import listdir\n",
        "from os.path import isfile, join\n",
        "#lista tutti i file dato un percorso\n",
        "def list_all_files(input_path):\n",
        "  files = [join(input_path, f) for f in listdir(input_path) if isfile(join(input_path, f))]\n",
        "  return list(zip(files, listdir(input_path)))\n",
        "\n",
        "def list_all_in_dir(input_path):\n",
        "  files = [join(input_path, f) for f in listdir(input_path)]\n",
        "  return files"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "npZ6VUZHK39U"
      },
      "source": [
        "#estrae i dati dal file json\n",
        "def extract_json_detection(file):\n",
        "  with open(file, 'r') as f:\n",
        "    data = json.load(f)\n",
        "    data = data[\"people\"][0][\"pose_keypoints_2d\"]\n",
        "  return data"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BWMFpkbSA6KB"
      },
      "source": [
        "def coordinate_from_data(data): \n",
        "  del data[2:75:3] \n",
        "  new_data = [ [data[i], data[i+1]] for i in range(0, len(data), 2)]\n",
        "  return new_data"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "iVj0mut_K-6P"
      },
      "source": [
        "keypoints = np.arange(0, 25)\n",
        "matrix = []"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XIr6i3xBAfRh"
      },
      "source": [
        "outputs = sorted(list_all_in_dir(\"/content/Tirocinio/video_output\"))\n",
        "for output in outputs:\n",
        "  print(output)\n",
        "  files = sorted(list_all_files(output))\n",
        "  for file in files: \n",
        "    data = coordinate_from_data(extract_json_detection(file[0]))\n",
        "    matrix.append(data)\n",
        "\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YZe850JDiwSG"
      },
      "source": [
        "labels = []\n",
        "for output in outputs: \n",
        "  for file in sorted(list_all_files(output)):\n",
        "    labels.append((file[1].split(\"_\"))[2])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-yE5BaBWt5aj"
      },
      "source": [
        "matrix = np.array(matrix)\n",
        "labels = np.array(labels)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "LkuXvd5PkMEL"
      },
      "source": [
        "from tensorflow.keras.utils import to_categorical\n",
        "labels = to_categorical(labels)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "wa6jexMSllyJ"
      },
      "source": [
        "from sklearn.model_selection import train_test_split\n",
        "\n",
        "(X_train, X_test, y_train, y_test) = train_test_split(matrix, labels,\n",
        "\ttest_size=0.25, stratify=labels, random_state=42)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ykS7i8iupklB"
      },
      "source": [
        "from tensorflow.keras.models import Sequential\n",
        "from tensorflow.keras.layers import Dense, Flatten, Conv1D\n",
        "\n",
        "model = Sequential([\n",
        "                    Conv1D(16, 15, input_shape=(X_train.shape[1], X_train.shape[2])), \n",
        "                    Flatten(), \n",
        "                    Dense(128, activation = \"relu\"), \n",
        "                    Dense(128, activation = \"relu\"), \n",
        "                    Dense(128, activation = \"relu\"), \n",
        "                    Dense(y_train.shape[1], activation = \"softmax\")\n",
        "])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "eIsC2H7krS0U"
      },
      "source": [
        "model.compile(loss=\"binary_crossentropy\", optimizer=\"adam\", metrics=[\"accuracy\"])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "OwdhuzPhro9J",
        "outputId": "93a0a972-d2e8-4785-c1c8-426741b7fd8b"
      },
      "source": [
        "H = model.fit(X_train, y_train, epochs=100, batch_size=20,validation_data=(X_test, y_test) )"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/100\n",
            "642/642 [==============================] - 2s 3ms/step - loss: 3.8784 - accuracy: 0.6743 - val_loss: 0.3489 - val_accuracy: 0.8534\n",
            "Epoch 2/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.3752 - accuracy: 0.8607 - val_loss: 0.3069 - val_accuracy: 0.8698\n",
            "Epoch 3/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.2859 - accuracy: 0.8895 - val_loss: 0.2060 - val_accuracy: 0.9324\n",
            "Epoch 4/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.2681 - accuracy: 0.8984 - val_loss: 0.2397 - val_accuracy: 0.9095\n",
            "Epoch 5/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.2417 - accuracy: 0.9048 - val_loss: 0.1905 - val_accuracy: 0.9306\n",
            "Epoch 6/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.2205 - accuracy: 0.9152 - val_loss: 0.3151 - val_accuracy: 0.8796\n",
            "Epoch 7/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.2327 - accuracy: 0.9126 - val_loss: 0.1666 - val_accuracy: 0.9472\n",
            "Epoch 8/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.1840 - accuracy: 0.9306 - val_loss: 0.1811 - val_accuracy: 0.9294\n",
            "Epoch 9/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.1836 - accuracy: 0.9338 - val_loss: 0.1248 - val_accuracy: 0.9563\n",
            "Epoch 10/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.1542 - accuracy: 0.9432 - val_loss: 0.2364 - val_accuracy: 0.9114\n",
            "Epoch 11/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.1621 - accuracy: 0.9382 - val_loss: 0.1704 - val_accuracy: 0.9395\n",
            "Epoch 12/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.1706 - accuracy: 0.9353 - val_loss: 0.1495 - val_accuracy: 0.9465\n",
            "Epoch 13/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.1356 - accuracy: 0.9478 - val_loss: 0.1700 - val_accuracy: 0.9392\n",
            "Epoch 14/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.1409 - accuracy: 0.9469 - val_loss: 0.1194 - val_accuracy: 0.9647\n",
            "Epoch 15/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.1350 - accuracy: 0.9531 - val_loss: 0.1444 - val_accuracy: 0.9495\n",
            "Epoch 16/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.1266 - accuracy: 0.9533 - val_loss: 0.1484 - val_accuracy: 0.9516\n",
            "Epoch 17/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.1194 - accuracy: 0.9553 - val_loss: 0.1072 - val_accuracy: 0.9584\n",
            "Epoch 18/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.1189 - accuracy: 0.9580 - val_loss: 0.1516 - val_accuracy: 0.9444\n",
            "Epoch 19/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.1005 - accuracy: 0.9623 - val_loss: 0.0930 - val_accuracy: 0.9687\n",
            "Epoch 20/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0955 - accuracy: 0.9632 - val_loss: 0.0962 - val_accuracy: 0.9659\n",
            "Epoch 21/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.1150 - accuracy: 0.9552 - val_loss: 0.1094 - val_accuracy: 0.9591\n",
            "Epoch 22/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.1021 - accuracy: 0.9618 - val_loss: 0.0987 - val_accuracy: 0.9600\n",
            "Epoch 23/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0912 - accuracy: 0.9668 - val_loss: 0.0986 - val_accuracy: 0.9649\n",
            "Epoch 24/100\n",
            "642/642 [==============================] - 2s 3ms/step - loss: 0.0859 - accuracy: 0.9650 - val_loss: 0.1026 - val_accuracy: 0.9659\n",
            "Epoch 25/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0970 - accuracy: 0.9646 - val_loss: 0.1114 - val_accuracy: 0.9572\n",
            "Epoch 26/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0822 - accuracy: 0.9674 - val_loss: 0.0755 - val_accuracy: 0.9708\n",
            "Epoch 27/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0810 - accuracy: 0.9698 - val_loss: 0.0841 - val_accuracy: 0.9741\n",
            "Epoch 28/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0830 - accuracy: 0.9682 - val_loss: 0.1208 - val_accuracy: 0.9593\n",
            "Epoch 29/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0755 - accuracy: 0.9721 - val_loss: 0.1111 - val_accuracy: 0.9577\n",
            "Epoch 30/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0997 - accuracy: 0.9629 - val_loss: 0.1327 - val_accuracy: 0.9514\n",
            "Epoch 31/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0851 - accuracy: 0.9678 - val_loss: 0.0695 - val_accuracy: 0.9736\n",
            "Epoch 32/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0738 - accuracy: 0.9699 - val_loss: 0.0946 - val_accuracy: 0.9684\n",
            "Epoch 33/100\n",
            "642/642 [==============================] - 2s 3ms/step - loss: 0.0832 - accuracy: 0.9672 - val_loss: 0.0761 - val_accuracy: 0.9743\n",
            "Epoch 34/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.1021 - accuracy: 0.9643 - val_loss: 0.1083 - val_accuracy: 0.9654\n",
            "Epoch 35/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0678 - accuracy: 0.9767 - val_loss: 0.0747 - val_accuracy: 0.9724\n",
            "Epoch 36/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0741 - accuracy: 0.9725 - val_loss: 0.0857 - val_accuracy: 0.9701\n",
            "Epoch 37/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0549 - accuracy: 0.9795 - val_loss: 0.1013 - val_accuracy: 0.9663\n",
            "Epoch 38/100\n",
            "642/642 [==============================] - 2s 3ms/step - loss: 0.0719 - accuracy: 0.9738 - val_loss: 0.0955 - val_accuracy: 0.9701\n",
            "Epoch 39/100\n",
            "642/642 [==============================] - 2s 3ms/step - loss: 0.0676 - accuracy: 0.9748 - val_loss: 0.0824 - val_accuracy: 0.9719\n",
            "Epoch 40/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0617 - accuracy: 0.9760 - val_loss: 0.0577 - val_accuracy: 0.9799\n",
            "Epoch 41/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0531 - accuracy: 0.9796 - val_loss: 0.0800 - val_accuracy: 0.9769\n",
            "Epoch 42/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0591 - accuracy: 0.9756 - val_loss: 0.1225 - val_accuracy: 0.9684\n",
            "Epoch 43/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0714 - accuracy: 0.9766 - val_loss: 0.0711 - val_accuracy: 0.9769\n",
            "Epoch 44/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0537 - accuracy: 0.9790 - val_loss: 0.0930 - val_accuracy: 0.9708\n",
            "Epoch 45/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0651 - accuracy: 0.9736 - val_loss: 0.0812 - val_accuracy: 0.9743\n",
            "Epoch 46/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0546 - accuracy: 0.9794 - val_loss: 0.0606 - val_accuracy: 0.9820\n",
            "Epoch 47/100\n",
            "642/642 [==============================] - 2s 3ms/step - loss: 0.0564 - accuracy: 0.9781 - val_loss: 0.1080 - val_accuracy: 0.9642\n",
            "Epoch 48/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0605 - accuracy: 0.9779 - val_loss: 0.0784 - val_accuracy: 0.9738\n",
            "Epoch 49/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0608 - accuracy: 0.9777 - val_loss: 0.1052 - val_accuracy: 0.9731\n",
            "Epoch 50/100\n",
            "642/642 [==============================] - 2s 3ms/step - loss: 0.0644 - accuracy: 0.9766 - val_loss: 0.1593 - val_accuracy: 0.9497\n",
            "Epoch 51/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0477 - accuracy: 0.9836 - val_loss: 0.0805 - val_accuracy: 0.9752\n",
            "Epoch 52/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0482 - accuracy: 0.9803 - val_loss: 0.0897 - val_accuracy: 0.9771\n",
            "Epoch 53/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0581 - accuracy: 0.9800 - val_loss: 0.0734 - val_accuracy: 0.9778\n",
            "Epoch 54/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0625 - accuracy: 0.9759 - val_loss: 0.0716 - val_accuracy: 0.9783\n",
            "Epoch 55/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0547 - accuracy: 0.9802 - val_loss: 0.1010 - val_accuracy: 0.9708\n",
            "Epoch 56/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0565 - accuracy: 0.9790 - val_loss: 0.1032 - val_accuracy: 0.9701\n",
            "Epoch 57/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0607 - accuracy: 0.9811 - val_loss: 0.1082 - val_accuracy: 0.9642\n",
            "Epoch 58/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0497 - accuracy: 0.9810 - val_loss: 0.0661 - val_accuracy: 0.9839\n",
            "Epoch 59/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0550 - accuracy: 0.9791 - val_loss: 0.0584 - val_accuracy: 0.9822\n",
            "Epoch 60/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0475 - accuracy: 0.9828 - val_loss: 0.1077 - val_accuracy: 0.9673\n",
            "Epoch 61/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0654 - accuracy: 0.9757 - val_loss: 0.1138 - val_accuracy: 0.9659\n",
            "Epoch 62/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0530 - accuracy: 0.9814 - val_loss: 0.1151 - val_accuracy: 0.9696\n",
            "Epoch 63/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0496 - accuracy: 0.9830 - val_loss: 0.0825 - val_accuracy: 0.9691\n",
            "Epoch 64/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0423 - accuracy: 0.9840 - val_loss: 0.1029 - val_accuracy: 0.9645\n",
            "Epoch 65/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0746 - accuracy: 0.9705 - val_loss: 0.0635 - val_accuracy: 0.9808\n",
            "Epoch 66/100\n",
            "642/642 [==============================] - 2s 3ms/step - loss: 0.0476 - accuracy: 0.9820 - val_loss: 0.0672 - val_accuracy: 0.9808\n",
            "Epoch 67/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0519 - accuracy: 0.9825 - val_loss: 0.0891 - val_accuracy: 0.9670\n",
            "Epoch 68/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0404 - accuracy: 0.9828 - val_loss: 0.0930 - val_accuracy: 0.9675\n",
            "Epoch 69/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0542 - accuracy: 0.9787 - val_loss: 0.0726 - val_accuracy: 0.9773\n",
            "Epoch 70/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0394 - accuracy: 0.9838 - val_loss: 0.0855 - val_accuracy: 0.9792\n",
            "Epoch 71/100\n",
            "642/642 [==============================] - 2s 3ms/step - loss: 0.0534 - accuracy: 0.9806 - val_loss: 0.0615 - val_accuracy: 0.9797\n",
            "Epoch 72/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0431 - accuracy: 0.9835 - val_loss: 0.0705 - val_accuracy: 0.9797\n",
            "Epoch 73/100\n",
            "642/642 [==============================] - 2s 3ms/step - loss: 0.0521 - accuracy: 0.9829 - val_loss: 0.0741 - val_accuracy: 0.9787\n",
            "Epoch 74/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0385 - accuracy: 0.9854 - val_loss: 0.0846 - val_accuracy: 0.9778\n",
            "Epoch 75/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0430 - accuracy: 0.9827 - val_loss: 0.1427 - val_accuracy: 0.9750\n",
            "Epoch 76/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0484 - accuracy: 0.9812 - val_loss: 0.0834 - val_accuracy: 0.9719\n",
            "Epoch 77/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0481 - accuracy: 0.9827 - val_loss: 0.0967 - val_accuracy: 0.9764\n",
            "Epoch 78/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0488 - accuracy: 0.9814 - val_loss: 0.1267 - val_accuracy: 0.9638\n",
            "Epoch 79/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0499 - accuracy: 0.9810 - val_loss: 0.0808 - val_accuracy: 0.9808\n",
            "Epoch 80/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0431 - accuracy: 0.9834 - val_loss: 0.0838 - val_accuracy: 0.9773\n",
            "Epoch 81/100\n",
            "642/642 [==============================] - 2s 3ms/step - loss: 0.0416 - accuracy: 0.9860 - val_loss: 0.0801 - val_accuracy: 0.9832\n",
            "Epoch 82/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0491 - accuracy: 0.9855 - val_loss: 0.1186 - val_accuracy: 0.9708\n",
            "Epoch 83/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0435 - accuracy: 0.9834 - val_loss: 0.0839 - val_accuracy: 0.9839\n",
            "Epoch 84/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0381 - accuracy: 0.9838 - val_loss: 0.0890 - val_accuracy: 0.9736\n",
            "Epoch 85/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0402 - accuracy: 0.9849 - val_loss: 0.1660 - val_accuracy: 0.9523\n",
            "Epoch 86/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0400 - accuracy: 0.9853 - val_loss: 0.0860 - val_accuracy: 0.9790\n",
            "Epoch 87/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0425 - accuracy: 0.9812 - val_loss: 0.1482 - val_accuracy: 0.9649\n",
            "Epoch 88/100\n",
            "642/642 [==============================] - 2s 3ms/step - loss: 0.0452 - accuracy: 0.9810 - val_loss: 0.1461 - val_accuracy: 0.9712\n",
            "Epoch 89/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0630 - accuracy: 0.9767 - val_loss: 0.0729 - val_accuracy: 0.9799\n",
            "Epoch 90/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0382 - accuracy: 0.9841 - val_loss: 0.1230 - val_accuracy: 0.9684\n",
            "Epoch 91/100\n",
            "642/642 [==============================] - 2s 3ms/step - loss: 0.0329 - accuracy: 0.9854 - val_loss: 0.0775 - val_accuracy: 0.9836\n",
            "Epoch 92/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0297 - accuracy: 0.9868 - val_loss: 0.0998 - val_accuracy: 0.9790\n",
            "Epoch 93/100\n",
            "642/642 [==============================] - 2s 3ms/step - loss: 0.0365 - accuracy: 0.9850 - val_loss: 0.0840 - val_accuracy: 0.9846\n",
            "Epoch 94/100\n",
            "642/642 [==============================] - 2s 3ms/step - loss: 0.0421 - accuracy: 0.9828 - val_loss: 0.1021 - val_accuracy: 0.9780\n",
            "Epoch 95/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0352 - accuracy: 0.9873 - val_loss: 0.1821 - val_accuracy: 0.9717\n",
            "Epoch 96/100\n",
            "642/642 [==============================] - 2s 3ms/step - loss: 0.1748 - accuracy: 0.9582 - val_loss: 0.0899 - val_accuracy: 0.9750\n",
            "Epoch 97/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0442 - accuracy: 0.9833 - val_loss: 0.0779 - val_accuracy: 0.9862\n",
            "Epoch 98/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0361 - accuracy: 0.9864 - val_loss: 0.0983 - val_accuracy: 0.9752\n",
            "Epoch 99/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0401 - accuracy: 0.9841 - val_loss: 0.1019 - val_accuracy: 0.9832\n",
            "Epoch 100/100\n",
            "642/642 [==============================] - 2s 3ms/step - loss: 0.0351 - accuracy: 0.9887 - val_loss: 0.1319 - val_accuracy: 0.9748\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "W6YhKEWer0rY",
        "outputId": "802019a4-b3c3-4605-a963-f82cf3efc341"
      },
      "source": [
        "model.evaluate(X_test, y_test)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "134/134 [==============================] - 0s 1ms/step - loss: 0.1319 - accuracy: 0.9748\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.13193227350711823, 0.9747545719146729]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 45
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "QYtDX1Y4jNeS"
      },
      "source": [
        "from sklearn.preprocessing import StandardScaler\n",
        "scaler = StandardScaler()\n",
        "X_train_scaled = scaler.fit_transform(X_train.reshape(-1, X_train.shape[-1])).reshape(X_train.shape)\n",
        "X_test_scaled = scaler.transform(X_test.reshape(-1, X_test.shape[-1])).reshape(X_test.shape)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zFvqp6AKjWdJ"
      },
      "source": [
        "model_scaled = Sequential([\n",
        "                    Conv1D(16, 15, input_shape=(X_train.shape[1], X_train.shape[2])), \n",
        "                    Flatten(), \n",
        "                    Dense(128, activation = \"relu\"), \n",
        "                    Dense(128, activation = \"relu\"), \n",
        "                    Dense(128, activation = \"relu\"), \n",
        "                    Dense(y_train.shape[1], activation = \"softmax\")\n",
        "])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DO6zzLUejYDC"
      },
      "source": [
        "model_scaled.compile(loss=\"binary_crossentropy\", optimizer=\"adam\", metrics=[\"accuracy\"])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "X4tDBZqCjZqx",
        "outputId": "f59f6899-c518-4913-8571-b4209f6474b4"
      },
      "source": [
        "H_scaled = model_scaled.fit(X_train_scaled, y_train, epochs=100, batch_size=20,validation_data=(X_test_scaled, y_test) )"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/100\n",
            "642/642 [==============================] - 2s 3ms/step - loss: 0.6076 - accuracy: 0.6408 - val_loss: 0.2258 - val_accuracy: 0.9107\n",
            "Epoch 2/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.2406 - accuracy: 0.9034 - val_loss: 0.1785 - val_accuracy: 0.9306\n",
            "Epoch 3/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.1850 - accuracy: 0.9255 - val_loss: 0.1560 - val_accuracy: 0.9371\n",
            "Epoch 4/100\n",
            "642/642 [==============================] - 1s 2ms/step - loss: 0.1567 - accuracy: 0.9357 - val_loss: 0.1494 - val_accuracy: 0.9371\n",
            "Epoch 5/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.1341 - accuracy: 0.9469 - val_loss: 0.0955 - val_accuracy: 0.9687\n",
            "Epoch 6/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.1254 - accuracy: 0.9506 - val_loss: 0.3312 - val_accuracy: 0.8946\n",
            "Epoch 7/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.1251 - accuracy: 0.9539 - val_loss: 0.1014 - val_accuracy: 0.9596\n",
            "Epoch 8/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.1036 - accuracy: 0.9598 - val_loss: 0.0884 - val_accuracy: 0.9698\n",
            "Epoch 9/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0959 - accuracy: 0.9640 - val_loss: 0.0876 - val_accuracy: 0.9694\n",
            "Epoch 10/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0915 - accuracy: 0.9646 - val_loss: 0.0769 - val_accuracy: 0.9729\n",
            "Epoch 11/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0910 - accuracy: 0.9662 - val_loss: 0.1041 - val_accuracy: 0.9647\n",
            "Epoch 12/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0760 - accuracy: 0.9713 - val_loss: 0.1239 - val_accuracy: 0.9556\n",
            "Epoch 13/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0749 - accuracy: 0.9708 - val_loss: 0.0974 - val_accuracy: 0.9649\n",
            "Epoch 14/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0690 - accuracy: 0.9733 - val_loss: 0.0600 - val_accuracy: 0.9811\n",
            "Epoch 15/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0656 - accuracy: 0.9721 - val_loss: 0.0666 - val_accuracy: 0.9773\n",
            "Epoch 16/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0754 - accuracy: 0.9718 - val_loss: 0.1094 - val_accuracy: 0.9668\n",
            "Epoch 17/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0731 - accuracy: 0.9735 - val_loss: 0.0660 - val_accuracy: 0.9778\n",
            "Epoch 18/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0569 - accuracy: 0.9790 - val_loss: 0.1385 - val_accuracy: 0.9600\n",
            "Epoch 19/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0643 - accuracy: 0.9779 - val_loss: 0.0902 - val_accuracy: 0.9731\n",
            "Epoch 20/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0598 - accuracy: 0.9781 - val_loss: 0.0768 - val_accuracy: 0.9766\n",
            "Epoch 21/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0711 - accuracy: 0.9718 - val_loss: 0.0727 - val_accuracy: 0.9778\n",
            "Epoch 22/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0531 - accuracy: 0.9793 - val_loss: 0.0846 - val_accuracy: 0.9708\n",
            "Epoch 23/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0543 - accuracy: 0.9776 - val_loss: 0.0784 - val_accuracy: 0.9752\n",
            "Epoch 24/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0484 - accuracy: 0.9808 - val_loss: 0.0861 - val_accuracy: 0.9741\n",
            "Epoch 25/100\n",
            "642/642 [==============================] - 1s 2ms/step - loss: 0.0532 - accuracy: 0.9810 - val_loss: 0.0656 - val_accuracy: 0.9820\n",
            "Epoch 26/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0463 - accuracy: 0.9826 - val_loss: 0.0764 - val_accuracy: 0.9771\n",
            "Epoch 27/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0368 - accuracy: 0.9861 - val_loss: 0.0793 - val_accuracy: 0.9792\n",
            "Epoch 28/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0443 - accuracy: 0.9836 - val_loss: 0.0860 - val_accuracy: 0.9729\n",
            "Epoch 29/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0477 - accuracy: 0.9804 - val_loss: 0.1043 - val_accuracy: 0.9663\n",
            "Epoch 30/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0410 - accuracy: 0.9876 - val_loss: 0.0736 - val_accuracy: 0.9813\n",
            "Epoch 31/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0469 - accuracy: 0.9839 - val_loss: 0.0943 - val_accuracy: 0.9736\n",
            "Epoch 32/100\n",
            "642/642 [==============================] - 2s 3ms/step - loss: 0.0466 - accuracy: 0.9838 - val_loss: 0.0674 - val_accuracy: 0.9860\n",
            "Epoch 33/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0400 - accuracy: 0.9846 - val_loss: 0.0578 - val_accuracy: 0.9832\n",
            "Epoch 34/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0285 - accuracy: 0.9898 - val_loss: 0.1166 - val_accuracy: 0.9577\n",
            "Epoch 35/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0490 - accuracy: 0.9810 - val_loss: 0.0619 - val_accuracy: 0.9853\n",
            "Epoch 36/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0400 - accuracy: 0.9869 - val_loss: 0.0582 - val_accuracy: 0.9862\n",
            "Epoch 37/100\n",
            "642/642 [==============================] - 1s 2ms/step - loss: 0.0593 - accuracy: 0.9801 - val_loss: 0.0578 - val_accuracy: 0.9869\n",
            "Epoch 38/100\n",
            "642/642 [==============================] - 1s 2ms/step - loss: 0.0382 - accuracy: 0.9873 - val_loss: 0.1046 - val_accuracy: 0.9691\n",
            "Epoch 39/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0358 - accuracy: 0.9884 - val_loss: 0.1209 - val_accuracy: 0.9663\n",
            "Epoch 40/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0445 - accuracy: 0.9838 - val_loss: 0.0614 - val_accuracy: 0.9860\n",
            "Epoch 41/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0323 - accuracy: 0.9890 - val_loss: 0.0715 - val_accuracy: 0.9801\n",
            "Epoch 42/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0322 - accuracy: 0.9891 - val_loss: 0.0687 - val_accuracy: 0.9799\n",
            "Epoch 43/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0395 - accuracy: 0.9856 - val_loss: 0.0871 - val_accuracy: 0.9727\n",
            "Epoch 44/100\n",
            "642/642 [==============================] - 1s 2ms/step - loss: 0.0310 - accuracy: 0.9883 - val_loss: 0.0634 - val_accuracy: 0.9815\n",
            "Epoch 45/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0325 - accuracy: 0.9880 - val_loss: 0.0695 - val_accuracy: 0.9839\n",
            "Epoch 46/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0305 - accuracy: 0.9900 - val_loss: 0.0609 - val_accuracy: 0.9878\n",
            "Epoch 47/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0285 - accuracy: 0.9895 - val_loss: 0.0598 - val_accuracy: 0.9832\n",
            "Epoch 48/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0264 - accuracy: 0.9915 - val_loss: 0.0631 - val_accuracy: 0.9855\n",
            "Epoch 49/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0246 - accuracy: 0.9908 - val_loss: 0.0505 - val_accuracy: 0.9881\n",
            "Epoch 50/100\n",
            "642/642 [==============================] - 1s 2ms/step - loss: 0.0259 - accuracy: 0.9914 - val_loss: 0.0619 - val_accuracy: 0.9862\n",
            "Epoch 51/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0341 - accuracy: 0.9873 - val_loss: 0.0556 - val_accuracy: 0.9834\n",
            "Epoch 52/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0245 - accuracy: 0.9899 - val_loss: 0.0584 - val_accuracy: 0.9867\n",
            "Epoch 53/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0247 - accuracy: 0.9925 - val_loss: 0.0591 - val_accuracy: 0.9871\n",
            "Epoch 54/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0232 - accuracy: 0.9924 - val_loss: 0.0526 - val_accuracy: 0.9890\n",
            "Epoch 55/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0277 - accuracy: 0.9899 - val_loss: 0.0699 - val_accuracy: 0.9881\n",
            "Epoch 56/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0220 - accuracy: 0.9930 - val_loss: 0.0861 - val_accuracy: 0.9811\n",
            "Epoch 57/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0234 - accuracy: 0.9912 - val_loss: 0.0816 - val_accuracy: 0.9853\n",
            "Epoch 58/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0257 - accuracy: 0.9906 - val_loss: 0.0774 - val_accuracy: 0.9794\n",
            "Epoch 59/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0319 - accuracy: 0.9889 - val_loss: 0.0597 - val_accuracy: 0.9883\n",
            "Epoch 60/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0188 - accuracy: 0.9938 - val_loss: 0.0691 - val_accuracy: 0.9883\n",
            "Epoch 61/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0209 - accuracy: 0.9926 - val_loss: 0.1128 - val_accuracy: 0.9696\n",
            "Epoch 62/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0225 - accuracy: 0.9903 - val_loss: 0.0765 - val_accuracy: 0.9848\n",
            "Epoch 63/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0289 - accuracy: 0.9889 - val_loss: 0.0623 - val_accuracy: 0.9857\n",
            "Epoch 64/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0187 - accuracy: 0.9935 - val_loss: 0.0952 - val_accuracy: 0.9727\n",
            "Epoch 65/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0242 - accuracy: 0.9901 - val_loss: 0.0745 - val_accuracy: 0.9860\n",
            "Epoch 66/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0205 - accuracy: 0.9934 - val_loss: 0.0637 - val_accuracy: 0.9867\n",
            "Epoch 67/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0198 - accuracy: 0.9924 - val_loss: 0.0857 - val_accuracy: 0.9822\n",
            "Epoch 68/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0261 - accuracy: 0.9904 - val_loss: 0.0642 - val_accuracy: 0.9902\n",
            "Epoch 69/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0244 - accuracy: 0.9923 - val_loss: 0.0711 - val_accuracy: 0.9850\n",
            "Epoch 70/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0165 - accuracy: 0.9935 - val_loss: 0.0720 - val_accuracy: 0.9906\n",
            "Epoch 71/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0204 - accuracy: 0.9928 - val_loss: 0.0702 - val_accuracy: 0.9871\n",
            "Epoch 72/100\n",
            "642/642 [==============================] - 1s 2ms/step - loss: 0.0192 - accuracy: 0.9938 - val_loss: 0.0540 - val_accuracy: 0.9918\n",
            "Epoch 73/100\n",
            "642/642 [==============================] - 2s 3ms/step - loss: 0.0203 - accuracy: 0.9939 - val_loss: 0.0725 - val_accuracy: 0.9885\n",
            "Epoch 74/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0236 - accuracy: 0.9914 - val_loss: 0.1442 - val_accuracy: 0.9677\n",
            "Epoch 75/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0299 - accuracy: 0.9889 - val_loss: 0.0681 - val_accuracy: 0.9867\n",
            "Epoch 76/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0259 - accuracy: 0.9912 - val_loss: 0.0732 - val_accuracy: 0.9848\n",
            "Epoch 77/100\n",
            "642/642 [==============================] - 2s 3ms/step - loss: 0.0193 - accuracy: 0.9926 - val_loss: 0.0651 - val_accuracy: 0.9871\n",
            "Epoch 78/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0211 - accuracy: 0.9924 - val_loss: 0.0663 - val_accuracy: 0.9904\n",
            "Epoch 79/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0180 - accuracy: 0.9949 - val_loss: 0.0697 - val_accuracy: 0.9832\n",
            "Epoch 80/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0207 - accuracy: 0.9924 - val_loss: 0.0933 - val_accuracy: 0.9853\n",
            "Epoch 81/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0169 - accuracy: 0.9951 - val_loss: 0.0836 - val_accuracy: 0.9806\n",
            "Epoch 82/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0235 - accuracy: 0.9912 - val_loss: 0.0919 - val_accuracy: 0.9846\n",
            "Epoch 83/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0219 - accuracy: 0.9918 - val_loss: 0.0879 - val_accuracy: 0.9883\n",
            "Epoch 84/100\n",
            "642/642 [==============================] - 2s 3ms/step - loss: 0.0182 - accuracy: 0.9939 - val_loss: 0.0706 - val_accuracy: 0.9904\n",
            "Epoch 85/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0146 - accuracy: 0.9954 - val_loss: 0.0671 - val_accuracy: 0.9909\n",
            "Epoch 86/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0258 - accuracy: 0.9910 - val_loss: 0.0710 - val_accuracy: 0.9916\n",
            "Epoch 87/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0211 - accuracy: 0.9930 - val_loss: 0.0977 - val_accuracy: 0.9815\n",
            "Epoch 88/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0246 - accuracy: 0.9933 - val_loss: 0.0608 - val_accuracy: 0.9883\n",
            "Epoch 89/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0129 - accuracy: 0.9957 - val_loss: 0.0656 - val_accuracy: 0.9909\n",
            "Epoch 90/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0193 - accuracy: 0.9944 - val_loss: 0.0635 - val_accuracy: 0.9885\n",
            "Epoch 91/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0145 - accuracy: 0.9955 - val_loss: 0.0667 - val_accuracy: 0.9911\n",
            "Epoch 92/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0136 - accuracy: 0.9953 - val_loss: 0.0731 - val_accuracy: 0.9897\n",
            "Epoch 93/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0239 - accuracy: 0.9916 - val_loss: 0.0592 - val_accuracy: 0.9892\n",
            "Epoch 94/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0090 - accuracy: 0.9963 - val_loss: 0.0644 - val_accuracy: 0.9902\n",
            "Epoch 95/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0204 - accuracy: 0.9936 - val_loss: 0.0708 - val_accuracy: 0.9892\n",
            "Epoch 96/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0166 - accuracy: 0.9947 - val_loss: 0.0820 - val_accuracy: 0.9892\n",
            "Epoch 97/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0183 - accuracy: 0.9934 - val_loss: 0.0772 - val_accuracy: 0.9867\n",
            "Epoch 98/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0201 - accuracy: 0.9937 - val_loss: 0.0817 - val_accuracy: 0.9841\n",
            "Epoch 99/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0243 - accuracy: 0.9920 - val_loss: 0.0769 - val_accuracy: 0.9860\n",
            "Epoch 100/100\n",
            "642/642 [==============================] - 2s 2ms/step - loss: 0.0289 - accuracy: 0.9895 - val_loss: 0.0764 - val_accuracy: 0.9906\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "tzsXI-8g79YW",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 598
        },
        "outputId": "d9c1e82b-9665-4f57-89f4-efa378af15af"
      },
      "source": [
        "import matplotlib.pyplot as plt\n",
        "\n",
        "N = 100\n",
        "plt.style.use(\"ggplot\")\n",
        "plt.figure()\n",
        "plt.plot(np.arange(0, N), H.history[\"accuracy\"], label=\"train_acc\")\n",
        "plt.plot(np.arange(0, N), H.history[\"val_accuracy\"], label=\"val_acc\")\n",
        "plt.title(\"Accuracy on Dataset\")\n",
        "plt.xlabel(\"Epoch #\")\n",
        "plt.ylabel(\"Accuracy\")\n",
        "plt.legend(loc=\"lower left\")\n",
        "\n",
        "plt.style.use(\"ggplot\")\n",
        "plt.figure()\n",
        "plt.plot(np.arange(0, N), H.history[\"loss\"], label=\"loss\")\n",
        "plt.plot(np.arange(0, N), H.history[\"val_loss\"], label=\"val_loss\")\n",
        "plt.title(\"Loss on Dataset\")\n",
        "plt.xlabel(\"Epoch #\")\n",
        "plt.ylabel(\"Accuracy\")\n",
        "plt.legend(loc=\"lower left\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.legend.Legend at 0x7f3d2c568dd0>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 53
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 598
        },
        "id": "z9y6Q_0HXeYi",
        "outputId": "86503304-0c78-4471-b7d4-65df28c5e58b"
      },
      "source": [
        "plt.style.use(\"ggplot\")\n",
        "plt.figure()\n",
        "plt.plot(np.arange(0, N), H_scaled.history[\"accuracy\"], label=\"train_acc\")\n",
        "plt.plot(np.arange(0, N), H_scaled.history[\"val_accuracy\"], label=\"val_acc\")\n",
        "plt.title(\"Accuracy on Dataset\")\n",
        "plt.xlabel(\"Epoch #\")\n",
        "plt.ylabel(\"Accuracy\")\n",
        "plt.legend(loc=\"lower left\")\n",
        "\n",
        "plt.style.use(\"ggplot\")\n",
        "plt.figure()\n",
        "plt.plot(np.arange(0, N), H_scaled.history[\"loss\"], label=\"loss\")\n",
        "plt.plot(np.arange(0, N), H_scaled.history[\"val_loss\"], label=\"val_loss\")\n",
        "plt.title(\"Loss on Dataset\")\n",
        "plt.xlabel(\"Epoch #\")\n",
        "plt.ylabel(\"Accuracy\")\n",
        "plt.legend(loc=\"lower left\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.legend.Legend at 0x7f3d37ba8790>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 54
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    }
  ]
}