LFO_Classification.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "kernelspec": {
      "display_name": "Python 3",
      "language": "python",
      "name": "python3"
    },
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.7.1"
    },
    "colab": {
      "name": "LFO Classification.ipynb",
      "provenance": [],
      "collapsed_sections": [
        "PpT5hCwaGrhg"
      ]
    }
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "_q8gc1Q8Fnm9"
      },
      "source": [
        "try:\n",
        "  !git clone https://github.com/acen20/CNN-Tensorflow-Keras-for-sound-classification Data\n",
        "  import os\n",
        "  os.chdir('Data')\n",
        "except:\n",
        "  pass \n",
        "\n",
        "dataset = 'dataset.npy'"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ldfQc86dFmPz"
      },
      "source": [
        "import librosa, librosa.display\n",
        "import pandas as pd\n",
        "import os\n",
        "import matplotlib.pyplot as plt\n",
        "import numpy as np\n",
        "from sklearn.model_selection import train_test_split\n",
        "from tensorflow import keras\n",
        "from keras.utils.np_utils import to_categorical\n",
        "from tensorflow.python.keras.models import Sequential,Input,Model\n",
        "from keras.layers import Dense, Dropout, Flatten\n",
        "from keras.layers import Conv2D, MaxPooling2D\n",
        "from keras.layers import BatchNormalization\n",
        "from keras.layers.advanced_activations import LeakyReLU\n",
        "from sklearn.metrics import classification_report,ConfusionMatrixDisplay\n",
        "from sklearn.utils import shuffle"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "PpT5hCwaGrhg"
      },
      "source": [
        "# Feature Extractor\n",
        "Uncomment the following block to use your own dataset for training\n",
        "\n",
        "## Directory Heirarchy\n",
        "Relative to the location of this script, arrange the folders in the following format:\n",
        "\n",
        "<b>Drone Sounds:</b> audio/Drone/{drone sounds here}\n",
        "\n",
        "<b>Other Sounds:</b> audio/{Other}/{Other sounds here}"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "scrolled": false,
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "tBsKqBEQFmQE",
        "outputId": "3deb2c1a-3961-4e60-a7fb-4a5d70ee90a1"
      },
      "source": [
        "'''path = \"audio/\"\n",
        "min_samples = 22050\n",
        "\n",
        "waveforms = []\n",
        "ffts = []\n",
        "stfts = [] \n",
        "f_ffts = []\n",
        "log_specs = []\n",
        "MFCCs_list = []\n",
        "\n",
        "\n",
        "def visualize_signals():\n",
        "  sample_audio = 1\n",
        "  FIG_SIZE = (9, 5)\n",
        "\n",
        "  #Waveform\n",
        "  plt.figure(figsize = FIG_SIZE)\n",
        "  waveplot_ = waveforms[sample_audio]\n",
        "  librosa.display.waveplot(waveplot_[0], sr = waveplot_[1], alpha = 0.4)\n",
        "  plt.xlabel(\"Time (s)\")\n",
        "  plt.ylabel(\"Amplitude\")\n",
        "  plt.title(\"Waveform\")\n",
        "  plt.show()\n",
        "\n",
        "  #Fourier Transformation\n",
        "  fourier1 = f_ffts[sample_audio]\n",
        "  plt.figure(figsize=FIG_SIZE)\n",
        "  plt.plot(fourier1[0], fourier1[1], alpha=0.7, color = \"C5\")\n",
        "  plt.xlabel(\"Frequency\")\n",
        "  plt.ylabel(\"Magnitude\")\n",
        "  plt.title(\"FFT\")\n",
        "  plt.show()\n",
        "\n",
        "  #Power Spectrum\n",
        "  fourier2 = ffts[sample_audio]\n",
        "  plt.figure(figsize=FIG_SIZE)\n",
        "  plt.plot(fourier2[0], fourier2[1], alpha=0.7, color = \"C5\")\n",
        "  plt.xlabel(\"Frequency\")\n",
        "  plt.ylabel(\"Magnitude\")\n",
        "  plt.title(\"Power Spectrum\")\n",
        "  plt.show()\n",
        "\n",
        "  #Spectrogram\n",
        "  plt.figure(figsize = FIG_SIZE)\n",
        "  spectrogram_ = stfts[sample_audio]\n",
        "  librosa.display.specshow(spectrogram_[0], sr=spectrogram_[1], hop_length=spectrogram_[2])\n",
        "  plt.xlabel(\"Time\")\n",
        "  plt.ylabel(\"Frequency\")\n",
        "  plt.colorbar()\n",
        "  plt.title(\"Spectrogram\")\n",
        "\n",
        "  #Log Spectrogram\n",
        "  plt.figure(figsize = FIG_SIZE)\n",
        "  log_spectrogram_ = log_specs[sample_audio]\n",
        "  librosa.display.specshow(log_spectrogram_[0], sr=log_spectrogram_[1], hop_length=log_spectrogram_[2])\n",
        "  plt.xlabel(\"Time\")\n",
        "  plt.ylabel(\"Frequency\")\n",
        "  plt.colorbar()\n",
        "  plt.title(\"Log Spectrogram\")\n",
        "\n",
        "  #MFCCs\n",
        "  MFCCs_ = MFCCs_list[sample_audio]\n",
        "  plt.figure(figsize=FIG_SIZE)\n",
        "  librosa.display.specshow(MFCCs_[0], sr=MFCCs_[1], hop_length=MFCCs_[2])\n",
        "  plt.xlabel(\"Time\")\n",
        "  plt.ylabel(\"MFCC coefficients\")\n",
        "  plt.colorbar()\n",
        "  plt.title(\"MFCCs\")\n",
        "\n",
        "\n",
        "\n",
        "def extract_MFCCs(signal, sample_rate):\n",
        "    waveforms.append([signal, sample_rate]) \n",
        "    # FFT -> power spectrum\n",
        "    # perform Fourier transform\n",
        "    fft = np.fft.fft(signal)\n",
        "    magnitude = np.abs(fft)\n",
        "    frequency = np.linspace(0, sample_rate, len(magnitude))\n",
        "    f_ffts.append([frequency, magnitude])\n",
        "    \n",
        "    # calculate abs values on complex numbers to get magnitude\n",
        "    spectrum = np.abs(fft)\n",
        "    \n",
        "    # half frequency variable\n",
        "    f = np.linspace(0, sample_rate, len(spectrum))\n",
        "    left_spectrum = spectrum[:int(len(spectrum)/2)]\n",
        "    left_f = f[:int(len(spectrum)/2)]\n",
        "    \n",
        "    ffts.append([left_f, left_spectrum])\n",
        "    \n",
        "    num_mfcc=13 \n",
        "    n_fft=2048\n",
        "    hop_length=512\n",
        "    \n",
        "    #performing Short-time FT\n",
        "    stft = librosa.stft(signal, n_fft=n_fft, hop_length=hop_length)\n",
        "    spectrogram = np.abs(stft)\n",
        "    \n",
        "    stfts.append([spectrogram, sample_rate, hop_length])\n",
        "    \n",
        "    log_spectrogram = librosa.amplitude_to_db(spectrogram)\n",
        "    \n",
        "    log_specs.append([log_spectrogram, sample_rate, hop_length])\n",
        "    \n",
        "    MFCCs = librosa.feature.mfcc(signal, sample_rate, \n",
        "                                 n_mfcc=num_mfcc, n_fft=n_fft,\n",
        "                                 hop_length=hop_length)\n",
        "    MFCCs_list.append([MFCCs, sample_rate, hop_length])\n",
        "    return MFCCs\n",
        "\n",
        "\n",
        "\n",
        "def preprocess_dataset(dataset_path):\n",
        "    data = {\n",
        "        \"mapping\": [],\n",
        "        \"labels\": [],\n",
        "        \"MFCCs\": [],\n",
        "        \"files\": []\n",
        "    } \n",
        "    for i, (dirpath, dirnames, filenames) in enumerate(os.walk(dataset_path)):\n",
        "\n",
        "        if dirpath is not dataset_path:\n",
        "            label = dirpath.split(\"/\")[-1]\n",
        "            data[\"mapping\"].append(label)\n",
        "            print(\"\\nProcessing: '{}'\".format(label))\n",
        "\n",
        "            # process all audio files in sub-dir and store MFCCs\n",
        "            for f in filenames:\n",
        "                file_path = os.path.join(dirpath, f)\n",
        "                signal, sample_rate = librosa.load(file_path)\n",
        "                if len(signal) >= min_samples:\n",
        "\n",
        "                    signal = signal[:min_samples]\n",
        "\n",
        "                    MFCCs = extract_MFCCs(signal, sample_rate)\n",
        "                    if label == 'Drone':\n",
        "                        llabel = 1\n",
        "                    else:\n",
        "                        llabel = 0\n",
        "\n",
        "                    data[\"MFCCs\"].append(MFCCs.T.tolist())\n",
        "                    data[\"labels\"].append(llabel)\n",
        "                    data[\"files\"].append(file_path)\n",
        "                    print(\"{}: {}\".format(file_path, i-1))\n",
        "                    \n",
        "    return data\n",
        "df = preprocess_dataset(path)\n",
        "mfcc = np.array(df[\"MFCCs\"])\n",
        "labels = np.array(df[\"labels\"])\n",
        "data = list(df.items())\n",
        "an_array = np.array(data)\n",
        "dataset = \"custom.npy\"\n",
        "visualize_signals()\n",
        "np.save(dataset, an_array)\n",
        "'''"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "string"
            },
            "text/plain": [
              "'path = \"audio/\"\\nmin_samples = 22050\\n\\nwaveforms = []\\nffts = []\\nstfts = [] \\nf_ffts = []\\nlog_specs = []\\nMFCCs_list = []\\n\\n\\ndef visualize_signals():\\n  sample_audio = 1\\n  FIG_SIZE = (9, 5)\\n\\n  #Waveform\\n  plt.figure(figsize = FIG_SIZE)\\n  waveplot_ = waveforms[sample_audio]\\n  librosa.display.waveplot(waveplot_[0], sr = waveplot_[1], alpha = 0.4)\\n  plt.xlabel(\"Time (s)\")\\n  plt.ylabel(\"Amplitude\")\\n  plt.title(\"Waveform\")\\n  plt.show()\\n\\n  #Fourier Transformation\\n  fourier1 = f_ffts[sample_audio]\\n  plt.figure(figsize=FIG_SIZE)\\n  plt.plot(fourier1[0], fourier1[1], alpha=0.7, color = \"C5\")\\n  plt.xlabel(\"Frequency\")\\n  plt.ylabel(\"Magnitude\")\\n  plt.title(\"FFT\")\\n  plt.show()\\n\\n  #Power Spectrum\\n  fourier2 = ffts[sample_audio]\\n  plt.figure(figsize=FIG_SIZE)\\n  plt.plot(fourier2[0], fourier2[1], alpha=0.7, color = \"C5\")\\n  plt.xlabel(\"Frequency\")\\n  plt.ylabel(\"Magnitude\")\\n  plt.title(\"Power Spectrum\")\\n  plt.show()\\n\\n  #Spectrogram\\n  plt.figure(figsize = FIG_SIZE)\\n  spectrogram_ = stfts[sample_audio]\\n  librosa.display.specshow(spectrogram_[0], sr=spectrogram_[1], hop_length=spectrogram_[2])\\n  plt.xlabel(\"Time\")\\n  plt.ylabel(\"Frequency\")\\n  plt.colorbar()\\n  plt.title(\"Spectrogram\")\\n\\n  #Log Spectrogram\\n  plt.figure(figsize = FIG_SIZE)\\n  log_spectrogram_ = log_specs[sample_audio]\\n  librosa.display.specshow(log_spectrogram_[0], sr=log_spectrogram_[1], hop_length=log_spectrogram_[2])\\n  plt.xlabel(\"Time\")\\n  plt.ylabel(\"Frequency\")\\n  plt.colorbar()\\n  plt.title(\"Log Spectrogram\")\\n\\n  #MFCCs\\n  MFCCs_ = MFCCs_list[sample_audio]\\n  plt.figure(figsize=FIG_SIZE)\\n  librosa.display.specshow(MFCCs_[0], sr=MFCCs_[1], hop_length=MFCCs_[2])\\n  plt.xlabel(\"Time\")\\n  plt.ylabel(\"MFCC coefficients\")\\n  plt.colorbar()\\n  plt.title(\"MFCCs\")\\n\\n\\n\\ndef extract_MFCCs(signal, sample_rate):\\n    waveforms.append([signal, sample_rate]) \\n    # FFT -> power spectrum\\n    # perform Fourier transform\\n    fft = np.fft.fft(signal)\\n    magnitude = np.abs(fft)\\n    frequency = np.linspace(0, sample_rate, len(magnitude))\\n    f_ffts.append([frequency, magnitude])\\n    \\n    # calculate abs values on complex numbers to get magnitude\\n    spectrum = np.abs(fft)\\n    \\n    # half frequency variable\\n    f = np.linspace(0, sample_rate, len(spectrum))\\n    left_spectrum = spectrum[:int(len(spectrum)/2)]\\n    left_f = f[:int(len(spectrum)/2)]\\n    \\n    ffts.append([left_f, left_spectrum])\\n    \\n    num_mfcc=13 \\n    n_fft=2048\\n    hop_length=512\\n    \\n    #performing Short-time FT\\n    stft = librosa.stft(signal, n_fft=n_fft, hop_length=hop_length)\\n    spectrogram = np.abs(stft)\\n    \\n    stfts.append([spectrogram, sample_rate, hop_length])\\n    \\n    log_spectrogram = librosa.amplitude_to_db(spectrogram)\\n    \\n    log_specs.append([log_spectrogram, sample_rate, hop_length])\\n    \\n    MFCCs = librosa.feature.mfcc(signal, sample_rate, \\n                                 n_mfcc=num_mfcc, n_fft=n_fft,\\n                                 hop_length=hop_length)\\n    MFCCs_list.append([MFCCs, sample_rate, hop_length])\\n    return MFCCs\\n\\n\\n\\ndef preprocess_dataset(dataset_path):\\n    data = {\\n        \"mapping\": [],\\n        \"labels\": [],\\n        \"MFCCs\": [],\\n        \"files\": []\\n    } \\n    for i, (dirpath, dirnames, filenames) in enumerate(os.walk(dataset_path)):\\n\\n        if dirpath is not dataset_path:\\n            label = dirpath.split(\"/\")[-1]\\n            data[\"mapping\"].append(label)\\n            print(\"\\nProcessing: \\'{}\\'\".format(label))\\n\\n            # process all audio files in sub-dir and store MFCCs\\n            for f in filenames:\\n                file_path = os.path.join(dirpath, f)\\n                signal, sample_rate = librosa.load(file_path)\\n                if len(signal) >= min_samples:\\n\\n                    signal = signal[:min_samples]\\n\\n                    MFCCs = extract_MFCCs(signal, sample_rate)\\n                    if label == \\'Drone\\':\\n                        llabel = 1\\n                    else:\\n                        llabel = 0\\n\\n                    data[\"MFCCs\"].append(MFCCs.T.tolist())\\n                    data[\"labels\"].append(llabel)\\n                    data[\"files\"].append(file_path)\\n                    print(\"{}: {}\".format(file_path, i-1))\\n                    \\n    return data\\ndf = preprocess_dataset(path)\\nmfcc = np.array(df[\"MFCCs\"])\\nlabels = np.array(df[\"labels\"])\\ndata = list(df.items())\\nan_array = np.array(data)\\ndataset = \"custom.npy\"\\nvisualize_signals()\\nnp.save(dataset, an_array)\\n'"
            ]
          },
          "metadata": {},
          "execution_count": 3
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "VEq31P8rG_Xd"
      },
      "source": [
        "# CNN Training\n",
        "From the top contended (SVM with RBF kernel, Naive Bayes and CNN) algorithms, CNN was chosen for deployment due to its highly discriminative performance"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "VDqtgGzOFmQR"
      },
      "source": [
        "dataset_ = np.load(dataset, allow_pickle = True)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "LXDKQsPHFmQS",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "b53c04cd-cb9b-4728-b69a-1f7f2769e221"
      },
      "source": [
        "labels = np.array(dataset_[1][1])\n",
        "mfcc = np.array(dataset_[2][1])\n",
        "labels, mfcc = shuffle(labels, mfcc, random_state = 42)\n",
        "breakpoint_ = 6 #To Manually split samples into traning and test sets. The integer specifies num of testing samples\n",
        "test_X = mfcc[:breakpoint_]\n",
        "mfcc = mfcc[breakpoint_:]\n",
        "test_y = labels[:breakpoint_]\n",
        "labels = labels[breakpoint_:]\n",
        "print(mfcc.shape)\n",
        "print(labels.shape)\n",
        "print(test_X.shape)\n",
        "print(test_y.shape)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "(172, 44, 13)\n",
            "(172,)\n",
            "(6, 44, 13)\n",
            "(6,)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zFQKtkzjFmQT",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "602322ba-2c77-4a35-8d49-5765de396aec"
      },
      "source": [
        "mfcc.shape"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(172, 44, 13)"
            ]
          },
          "metadata": {},
          "execution_count": 6
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "LEBSjC7ZFmQV",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "7e1e9a25-862f-49bb-a1d9-1b6ffb4d075b"
      },
      "source": [
        "unique, counts = np.unique(labels, return_counts=True)\n",
        "unique, counts"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(array([0, 1]), array([88, 84]))"
            ]
          },
          "metadata": {},
          "execution_count": 7
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "GZUWne-YFmQW",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "3c4224eb-ccb0-4c3e-cec8-771e4e92ac9b"
      },
      "source": [
        "classes = np.unique(labels)\n",
        "nClasses = len(classes)\n",
        "print('Total number of outputs : ', nClasses)\n",
        "print('Output classes : ', classes)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Total number of outputs :  2\n",
            "Output classes :  [0 1]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ePsfhGLhFmQY",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "8c964e20-16c7-4fdb-b84d-0b3f58a400d1"
      },
      "source": [
        "train_X = mfcc.reshape(-1, 44, 13, 1)\n",
        "test_X = test_X.reshape(-1, 44, 13, 1)\n",
        "train_X.shape"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(172, 44, 13, 1)"
            ]
          },
          "metadata": {},
          "execution_count": 9
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "WYAlfqkMFmQZ"
      },
      "source": [
        "train_Y_one_hot = to_categorical(labels)\n",
        "test_y_one_hot = to_categorical(test_y)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "b0Qt9zzRFmQa",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "c97f6c4c-552c-4b39-a229-5638dffcced9"
      },
      "source": [
        "test_y_one_hot.shape"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(6, 2)"
            ]
          },
          "metadata": {},
          "execution_count": 11
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "De1qFHTFFmQb"
      },
      "source": [
        "train_X,valid_X,train_label,valid_label = train_test_split(train_X, train_Y_one_hot, test_size=0.20, random_state=42)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "oGI_H7JuFmQc",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "e9543562-45e6-445b-f857-bbe651198961"
      },
      "source": [
        "train_X.shape,valid_X.shape,train_label.shape,valid_label.shape"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "((137, 44, 13, 1), (35, 44, 13, 1), (137, 2), (35, 2))"
            ]
          },
          "metadata": {},
          "execution_count": 13
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ql6zE760FmQd"
      },
      "source": [
        "batch_size = 32\n",
        "epochs = 100\n",
        "num_classes = 2"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2jGehFYyFmQd"
      },
      "source": [
        "model_ = Sequential()\n",
        "model_.add(Conv2D(32, kernel_size=(3, 3),activation='linear',input_shape=(44,13,1),padding='same'))\n",
        "model_.add(LeakyReLU(alpha=0.1))\n",
        "model_.add(MaxPooling2D((2, 2),padding='same'))\n",
        "model_.add(Conv2D(64, (3, 3), activation='linear',padding='same'))\n",
        "model_.add(LeakyReLU(alpha=0.1))\n",
        "model_.add(MaxPooling2D(pool_size=(2, 2),padding='same'))\n",
        "model_.add(Conv2D(128, (3, 3), activation='linear',padding='same'))\n",
        "model_.add(LeakyReLU(alpha=0.1))                  \n",
        "model_.add(MaxPooling2D(pool_size=(2, 2),padding='same'))\n",
        "model_.add(Flatten())\n",
        "model_.add(Dense(128, activation='linear'))\n",
        "model_.add(LeakyReLU(alpha=0.1))                  \n",
        "model_.add(Dense(num_classes, activation='softmax'))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vREp9RV5FmQe"
      },
      "source": [
        "model_.compile(loss=keras.losses.categorical_crossentropy, optimizer='adam',metrics=['accuracy'])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "scrolled": false,
        "id": "GHcv23EsFmQg",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "20a30396-1c17-4f59-a313-662768e335b4"
      },
      "source": [
        "lfo_train = model_.fit(train_X, train_label, batch_size=batch_size,epochs=epochs,verbose=1,validation_data=(valid_X, valid_label))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/100\n",
            "5/5 [==============================] - 1s 115ms/step - loss: 23.2481 - accuracy: 0.5036 - val_loss: 8.4744 - val_accuracy: 0.4571\n",
            "Epoch 2/100\n",
            "5/5 [==============================] - 0s 61ms/step - loss: 7.9893 - accuracy: 0.5036 - val_loss: 0.4527 - val_accuracy: 0.7143\n",
            "Epoch 3/100\n",
            "5/5 [==============================] - 0s 55ms/step - loss: 1.0211 - accuracy: 0.6715 - val_loss: 0.5711 - val_accuracy: 0.6857\n",
            "Epoch 4/100\n",
            "5/5 [==============================] - 0s 56ms/step - loss: 0.5086 - accuracy: 0.7226 - val_loss: 0.5700 - val_accuracy: 0.6286\n",
            "Epoch 5/100\n",
            "5/5 [==============================] - 0s 60ms/step - loss: 0.4489 - accuracy: 0.7445 - val_loss: 0.3407 - val_accuracy: 0.8286\n",
            "Epoch 6/100\n",
            "5/5 [==============================] - 0s 56ms/step - loss: 0.2787 - accuracy: 0.8832 - val_loss: 0.4305 - val_accuracy: 0.8286\n",
            "Epoch 7/100\n",
            "5/5 [==============================] - 0s 55ms/step - loss: 0.2937 - accuracy: 0.8686 - val_loss: 0.2625 - val_accuracy: 0.8857\n",
            "Epoch 8/100\n",
            "5/5 [==============================] - 0s 59ms/step - loss: 0.2055 - accuracy: 0.9562 - val_loss: 0.2746 - val_accuracy: 0.9143\n",
            "Epoch 9/100\n",
            "5/5 [==============================] - 0s 55ms/step - loss: 0.1687 - accuracy: 0.9416 - val_loss: 0.2061 - val_accuracy: 0.9143\n",
            "Epoch 10/100\n",
            "5/5 [==============================] - 0s 55ms/step - loss: 0.1634 - accuracy: 0.9343 - val_loss: 0.2479 - val_accuracy: 0.8571\n",
            "Epoch 11/100\n",
            "5/5 [==============================] - 0s 55ms/step - loss: 0.1476 - accuracy: 0.9416 - val_loss: 0.3046 - val_accuracy: 0.9143\n",
            "Epoch 12/100\n",
            "5/5 [==============================] - 0s 55ms/step - loss: 0.1227 - accuracy: 0.9562 - val_loss: 0.1943 - val_accuracy: 0.9429\n",
            "Epoch 13/100\n",
            "5/5 [==============================] - 0s 55ms/step - loss: 0.0739 - accuracy: 0.9927 - val_loss: 0.1810 - val_accuracy: 0.9429\n",
            "Epoch 14/100\n",
            "5/5 [==============================] - 0s 55ms/step - loss: 0.0574 - accuracy: 1.0000 - val_loss: 0.1518 - val_accuracy: 0.9143\n",
            "Epoch 15/100\n",
            "5/5 [==============================] - 0s 58ms/step - loss: 0.0442 - accuracy: 1.0000 - val_loss: 0.1575 - val_accuracy: 0.9143\n",
            "Epoch 16/100\n",
            "5/5 [==============================] - 0s 54ms/step - loss: 0.0434 - accuracy: 1.0000 - val_loss: 0.1837 - val_accuracy: 0.9143\n",
            "Epoch 17/100\n",
            "5/5 [==============================] - 0s 55ms/step - loss: 0.0344 - accuracy: 1.0000 - val_loss: 0.1048 - val_accuracy: 0.9714\n",
            "Epoch 18/100\n",
            "5/5 [==============================] - 0s 55ms/step - loss: 0.0331 - accuracy: 0.9927 - val_loss: 0.1850 - val_accuracy: 0.9143\n",
            "Epoch 19/100\n",
            "5/5 [==============================] - 0s 55ms/step - loss: 0.0262 - accuracy: 1.0000 - val_loss: 0.1762 - val_accuracy: 0.8857\n",
            "Epoch 20/100\n",
            "5/5 [==============================] - 0s 56ms/step - loss: 0.0167 - accuracy: 1.0000 - val_loss: 0.1513 - val_accuracy: 0.9143\n",
            "Epoch 21/100\n",
            "5/5 [==============================] - 0s 58ms/step - loss: 0.0127 - accuracy: 1.0000 - val_loss: 0.2036 - val_accuracy: 0.9143\n",
            "Epoch 22/100\n",
            "5/5 [==============================] - 0s 55ms/step - loss: 0.0090 - accuracy: 1.0000 - val_loss: 0.1305 - val_accuracy: 0.9429\n",
            "Epoch 23/100\n",
            "5/5 [==============================] - 0s 56ms/step - loss: 0.0063 - accuracy: 1.0000 - val_loss: 0.1231 - val_accuracy: 0.9429\n",
            "Epoch 24/100\n",
            "5/5 [==============================] - 0s 54ms/step - loss: 0.0050 - accuracy: 1.0000 - val_loss: 0.1571 - val_accuracy: 0.9143\n",
            "Epoch 25/100\n",
            "5/5 [==============================] - 0s 56ms/step - loss: 0.0038 - accuracy: 1.0000 - val_loss: 0.1522 - val_accuracy: 0.9143\n",
            "Epoch 26/100\n",
            "5/5 [==============================] - 0s 59ms/step - loss: 0.0033 - accuracy: 1.0000 - val_loss: 0.1630 - val_accuracy: 0.9143\n",
            "Epoch 27/100\n",
            "5/5 [==============================] - 0s 55ms/step - loss: 0.0030 - accuracy: 1.0000 - val_loss: 0.1704 - val_accuracy: 0.9143\n",
            "Epoch 28/100\n",
            "5/5 [==============================] - 0s 54ms/step - loss: 0.0025 - accuracy: 1.0000 - val_loss: 0.1330 - val_accuracy: 0.9429\n",
            "Epoch 29/100\n",
            "5/5 [==============================] - 0s 53ms/step - loss: 0.0022 - accuracy: 1.0000 - val_loss: 0.1141 - val_accuracy: 0.9429\n",
            "Epoch 30/100\n",
            "5/5 [==============================] - 0s 54ms/step - loss: 0.0021 - accuracy: 1.0000 - val_loss: 0.1108 - val_accuracy: 0.9429\n",
            "Epoch 31/100\n",
            "5/5 [==============================] - 0s 56ms/step - loss: 0.0019 - accuracy: 1.0000 - val_loss: 0.1154 - val_accuracy: 0.9429\n",
            "Epoch 32/100\n",
            "5/5 [==============================] - 0s 56ms/step - loss: 0.0017 - accuracy: 1.0000 - val_loss: 0.1267 - val_accuracy: 0.9429\n",
            "Epoch 33/100\n",
            "5/5 [==============================] - 0s 56ms/step - loss: 0.0015 - accuracy: 1.0000 - val_loss: 0.1156 - val_accuracy: 0.9429\n",
            "Epoch 34/100\n",
            "5/5 [==============================] - 0s 55ms/step - loss: 0.0015 - accuracy: 1.0000 - val_loss: 0.1171 - val_accuracy: 0.9429\n",
            "Epoch 35/100\n",
            "5/5 [==============================] - 0s 54ms/step - loss: 0.0014 - accuracy: 1.0000 - val_loss: 0.1323 - val_accuracy: 0.9429\n",
            "Epoch 36/100\n",
            "5/5 [==============================] - 0s 56ms/step - loss: 0.0012 - accuracy: 1.0000 - val_loss: 0.1238 - val_accuracy: 0.9429\n",
            "Epoch 37/100\n",
            "5/5 [==============================] - 0s 55ms/step - loss: 0.0012 - accuracy: 1.0000 - val_loss: 0.1159 - val_accuracy: 0.9429\n",
            "Epoch 38/100\n",
            "5/5 [==============================] - 0s 57ms/step - loss: 0.0011 - accuracy: 1.0000 - val_loss: 0.1191 - val_accuracy: 0.9429\n",
            "Epoch 39/100\n",
            "5/5 [==============================] - 0s 58ms/step - loss: 0.0010 - accuracy: 1.0000 - val_loss: 0.1217 - val_accuracy: 0.9429\n",
            "Epoch 40/100\n",
            "5/5 [==============================] - 0s 57ms/step - loss: 9.7301e-04 - accuracy: 1.0000 - val_loss: 0.1132 - val_accuracy: 0.9429\n",
            "Epoch 41/100\n",
            "5/5 [==============================] - 0s 53ms/step - loss: 9.1975e-04 - accuracy: 1.0000 - val_loss: 0.1070 - val_accuracy: 0.9429\n",
            "Epoch 42/100\n",
            "5/5 [==============================] - 0s 56ms/step - loss: 9.0661e-04 - accuracy: 1.0000 - val_loss: 0.1124 - val_accuracy: 0.9429\n",
            "Epoch 43/100\n",
            "5/5 [==============================] - 0s 53ms/step - loss: 8.2984e-04 - accuracy: 1.0000 - val_loss: 0.1223 - val_accuracy: 0.9429\n",
            "Epoch 44/100\n",
            "5/5 [==============================] - 0s 56ms/step - loss: 8.1465e-04 - accuracy: 1.0000 - val_loss: 0.1273 - val_accuracy: 0.9429\n",
            "Epoch 45/100\n",
            "5/5 [==============================] - 0s 56ms/step - loss: 7.6158e-04 - accuracy: 1.0000 - val_loss: 0.1107 - val_accuracy: 0.9429\n",
            "Epoch 46/100\n",
            "5/5 [==============================] - 0s 56ms/step - loss: 7.3683e-04 - accuracy: 1.0000 - val_loss: 0.1050 - val_accuracy: 0.9429\n",
            "Epoch 47/100\n",
            "5/5 [==============================] - 0s 54ms/step - loss: 7.0156e-04 - accuracy: 1.0000 - val_loss: 0.1162 - val_accuracy: 0.9429\n",
            "Epoch 48/100\n",
            "5/5 [==============================] - 0s 56ms/step - loss: 6.7082e-04 - accuracy: 1.0000 - val_loss: 0.1102 - val_accuracy: 0.9429\n",
            "Epoch 49/100\n",
            "5/5 [==============================] - 0s 54ms/step - loss: 6.4026e-04 - accuracy: 1.0000 - val_loss: 0.1067 - val_accuracy: 0.9429\n",
            "Epoch 50/100\n",
            "5/5 [==============================] - 0s 54ms/step - loss: 6.1867e-04 - accuracy: 1.0000 - val_loss: 0.1061 - val_accuracy: 0.9429\n",
            "Epoch 51/100\n",
            "5/5 [==============================] - 0s 59ms/step - loss: 5.8692e-04 - accuracy: 1.0000 - val_loss: 0.1132 - val_accuracy: 0.9429\n",
            "Epoch 52/100\n",
            "5/5 [==============================] - 0s 55ms/step - loss: 5.6952e-04 - accuracy: 1.0000 - val_loss: 0.1145 - val_accuracy: 0.9429\n",
            "Epoch 53/100\n",
            "5/5 [==============================] - 0s 56ms/step - loss: 5.5192e-04 - accuracy: 1.0000 - val_loss: 0.1108 - val_accuracy: 0.9429\n",
            "Epoch 54/100\n",
            "5/5 [==============================] - 0s 55ms/step - loss: 5.2882e-04 - accuracy: 1.0000 - val_loss: 0.1161 - val_accuracy: 0.9429\n",
            "Epoch 55/100\n",
            "5/5 [==============================] - 0s 56ms/step - loss: 5.1166e-04 - accuracy: 1.0000 - val_loss: 0.1108 - val_accuracy: 0.9429\n",
            "Epoch 56/100\n",
            "5/5 [==============================] - 0s 54ms/step - loss: 4.8885e-04 - accuracy: 1.0000 - val_loss: 0.1051 - val_accuracy: 0.9429\n",
            "Epoch 57/100\n",
            "5/5 [==============================] - 0s 54ms/step - loss: 4.7323e-04 - accuracy: 1.0000 - val_loss: 0.1046 - val_accuracy: 0.9429\n",
            "Epoch 58/100\n",
            "5/5 [==============================] - 0s 54ms/step - loss: 4.6117e-04 - accuracy: 1.0000 - val_loss: 0.1005 - val_accuracy: 0.9429\n",
            "Epoch 59/100\n",
            "5/5 [==============================] - 0s 55ms/step - loss: 4.4321e-04 - accuracy: 1.0000 - val_loss: 0.1040 - val_accuracy: 0.9429\n",
            "Epoch 60/100\n",
            "5/5 [==============================] - 0s 54ms/step - loss: 4.3631e-04 - accuracy: 1.0000 - val_loss: 0.1113 - val_accuracy: 0.9429\n",
            "Epoch 61/100\n",
            "5/5 [==============================] - 0s 56ms/step - loss: 4.1607e-04 - accuracy: 1.0000 - val_loss: 0.1020 - val_accuracy: 0.9429\n",
            "Epoch 62/100\n",
            "5/5 [==============================] - 0s 56ms/step - loss: 4.0134e-04 - accuracy: 1.0000 - val_loss: 0.0941 - val_accuracy: 0.9714\n",
            "Epoch 63/100\n",
            "5/5 [==============================] - 0s 56ms/step - loss: 3.9843e-04 - accuracy: 1.0000 - val_loss: 0.0957 - val_accuracy: 0.9714\n",
            "Epoch 64/100\n",
            "5/5 [==============================] - 0s 56ms/step - loss: 3.8603e-04 - accuracy: 1.0000 - val_loss: 0.1070 - val_accuracy: 0.9429\n",
            "Epoch 65/100\n",
            "5/5 [==============================] - 0s 56ms/step - loss: 3.6960e-04 - accuracy: 1.0000 - val_loss: 0.1157 - val_accuracy: 0.9429\n",
            "Epoch 66/100\n",
            "5/5 [==============================] - 0s 59ms/step - loss: 3.6035e-04 - accuracy: 1.0000 - val_loss: 0.1101 - val_accuracy: 0.9429\n",
            "Epoch 67/100\n",
            "5/5 [==============================] - 0s 55ms/step - loss: 3.4304e-04 - accuracy: 1.0000 - val_loss: 0.1018 - val_accuracy: 0.9429\n",
            "Epoch 68/100\n",
            "5/5 [==============================] - 0s 59ms/step - loss: 3.3031e-04 - accuracy: 1.0000 - val_loss: 0.0971 - val_accuracy: 0.9714\n",
            "Epoch 69/100\n",
            "5/5 [==============================] - 0s 57ms/step - loss: 3.2518e-04 - accuracy: 1.0000 - val_loss: 0.0950 - val_accuracy: 0.9714\n",
            "Epoch 70/100\n",
            "5/5 [==============================] - 0s 56ms/step - loss: 3.1355e-04 - accuracy: 1.0000 - val_loss: 0.1013 - val_accuracy: 0.9429\n",
            "Epoch 71/100\n",
            "5/5 [==============================] - 0s 54ms/step - loss: 3.0291e-04 - accuracy: 1.0000 - val_loss: 0.1067 - val_accuracy: 0.9429\n",
            "Epoch 72/100\n",
            "5/5 [==============================] - 0s 56ms/step - loss: 2.9671e-04 - accuracy: 1.0000 - val_loss: 0.1050 - val_accuracy: 0.9429\n",
            "Epoch 73/100\n",
            "5/5 [==============================] - 0s 58ms/step - loss: 2.8602e-04 - accuracy: 1.0000 - val_loss: 0.1021 - val_accuracy: 0.9429\n",
            "Epoch 74/100\n",
            "5/5 [==============================] - 0s 54ms/step - loss: 2.7821e-04 - accuracy: 1.0000 - val_loss: 0.0973 - val_accuracy: 0.9429\n",
            "Epoch 75/100\n",
            "5/5 [==============================] - 0s 55ms/step - loss: 2.6434e-04 - accuracy: 1.0000 - val_loss: 0.1007 - val_accuracy: 0.9429\n",
            "Epoch 76/100\n",
            "5/5 [==============================] - 0s 56ms/step - loss: 2.6102e-04 - accuracy: 1.0000 - val_loss: 0.0986 - val_accuracy: 0.9429\n",
            "Epoch 77/100\n",
            "5/5 [==============================] - 0s 57ms/step - loss: 2.5344e-04 - accuracy: 1.0000 - val_loss: 0.0937 - val_accuracy: 0.9714\n",
            "Epoch 78/100\n",
            "5/5 [==============================] - 0s 55ms/step - loss: 2.4282e-04 - accuracy: 1.0000 - val_loss: 0.0900 - val_accuracy: 0.9714\n",
            "Epoch 79/100\n",
            "5/5 [==============================] - 0s 55ms/step - loss: 2.3927e-04 - accuracy: 1.0000 - val_loss: 0.0874 - val_accuracy: 0.9429\n",
            "Epoch 80/100\n",
            "5/5 [==============================] - 0s 61ms/step - loss: 2.3455e-04 - accuracy: 1.0000 - val_loss: 0.0888 - val_accuracy: 0.9429\n",
            "Epoch 81/100\n",
            "5/5 [==============================] - 0s 56ms/step - loss: 2.2667e-04 - accuracy: 1.0000 - val_loss: 0.0913 - val_accuracy: 0.9429\n",
            "Epoch 82/100\n",
            "5/5 [==============================] - 0s 55ms/step - loss: 2.1895e-04 - accuracy: 1.0000 - val_loss: 0.0951 - val_accuracy: 0.9429\n",
            "Epoch 83/100\n",
            "5/5 [==============================] - 0s 55ms/step - loss: 2.1455e-04 - accuracy: 1.0000 - val_loss: 0.0997 - val_accuracy: 0.9429\n",
            "Epoch 84/100\n",
            "5/5 [==============================] - 0s 56ms/step - loss: 2.0883e-04 - accuracy: 1.0000 - val_loss: 0.0979 - val_accuracy: 0.9429\n",
            "Epoch 85/100\n",
            "5/5 [==============================] - 0s 56ms/step - loss: 2.0317e-04 - accuracy: 1.0000 - val_loss: 0.0937 - val_accuracy: 0.9429\n",
            "Epoch 86/100\n",
            "5/5 [==============================] - 0s 55ms/step - loss: 1.9972e-04 - accuracy: 1.0000 - val_loss: 0.0909 - val_accuracy: 0.9429\n",
            "Epoch 87/100\n",
            "5/5 [==============================] - 0s 57ms/step - loss: 1.9447e-04 - accuracy: 1.0000 - val_loss: 0.0924 - val_accuracy: 0.9429\n",
            "Epoch 88/100\n",
            "5/5 [==============================] - 0s 57ms/step - loss: 1.9018e-04 - accuracy: 1.0000 - val_loss: 0.0942 - val_accuracy: 0.9429\n",
            "Epoch 89/100\n",
            "5/5 [==============================] - 0s 58ms/step - loss: 1.8467e-04 - accuracy: 1.0000 - val_loss: 0.0934 - val_accuracy: 0.9429\n",
            "Epoch 90/100\n",
            "5/5 [==============================] - 0s 55ms/step - loss: 1.8454e-04 - accuracy: 1.0000 - val_loss: 0.0982 - val_accuracy: 0.9429\n",
            "Epoch 91/100\n",
            "5/5 [==============================] - 0s 57ms/step - loss: 1.7441e-04 - accuracy: 1.0000 - val_loss: 0.0934 - val_accuracy: 0.9429\n",
            "Epoch 92/100\n",
            "5/5 [==============================] - 0s 58ms/step - loss: 1.7025e-04 - accuracy: 1.0000 - val_loss: 0.0896 - val_accuracy: 0.9429\n",
            "Epoch 93/100\n",
            "5/5 [==============================] - 0s 56ms/step - loss: 1.6761e-04 - accuracy: 1.0000 - val_loss: 0.0859 - val_accuracy: 0.9429\n",
            "Epoch 94/100\n",
            "5/5 [==============================] - 0s 58ms/step - loss: 1.6449e-04 - accuracy: 1.0000 - val_loss: 0.0851 - val_accuracy: 0.9429\n",
            "Epoch 95/100\n",
            "5/5 [==============================] - 0s 55ms/step - loss: 1.5994e-04 - accuracy: 1.0000 - val_loss: 0.0852 - val_accuracy: 0.9429\n",
            "Epoch 96/100\n",
            "5/5 [==============================] - 0s 55ms/step - loss: 1.5476e-04 - accuracy: 1.0000 - val_loss: 0.0848 - val_accuracy: 0.9429\n",
            "Epoch 97/100\n",
            "5/5 [==============================] - 0s 56ms/step - loss: 1.5084e-04 - accuracy: 1.0000 - val_loss: 0.0844 - val_accuracy: 0.9429\n",
            "Epoch 98/100\n",
            "5/5 [==============================] - 0s 61ms/step - loss: 1.4619e-04 - accuracy: 1.0000 - val_loss: 0.0821 - val_accuracy: 0.9429\n",
            "Epoch 99/100\n",
            "5/5 [==============================] - 0s 56ms/step - loss: 1.4311e-04 - accuracy: 1.0000 - val_loss: 0.0800 - val_accuracy: 0.9429\n",
            "Epoch 100/100\n",
            "5/5 [==============================] - 0s 56ms/step - loss: 1.4010e-04 - accuracy: 1.0000 - val_loss: 0.0825 - val_accuracy: 0.9429\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "BFpACxWXZxrr",
        "outputId": "3f871cab-2467-47b8-b9d1-f7d517480c8b"
      },
      "source": [
        "model_.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "module_wrapper (ModuleWrappe (None, 44, 13, 32)        320       \n",
            "_________________________________________________________________\n",
            "module_wrapper_1 (ModuleWrap (None, 44, 13, 32)        0         \n",
            "_________________________________________________________________\n",
            "module_wrapper_2 (ModuleWrap (None, 22, 7, 32)         0         \n",
            "_________________________________________________________________\n",
            "module_wrapper_3 (ModuleWrap (None, 22, 7, 64)         18496     \n",
            "_________________________________________________________________\n",
            "module_wrapper_4 (ModuleWrap (None, 22, 7, 64)         0         \n",
            "_________________________________________________________________\n",
            "module_wrapper_5 (ModuleWrap (None, 11, 4, 64)         0         \n",
            "_________________________________________________________________\n",
            "module_wrapper_6 (ModuleWrap (None, 11, 4, 128)        73856     \n",
            "_________________________________________________________________\n",
            "module_wrapper_7 (ModuleWrap (None, 11, 4, 128)        0         \n",
            "_________________________________________________________________\n",
            "module_wrapper_8 (ModuleWrap (None, 6, 2, 128)         0         \n",
            "_________________________________________________________________\n",
            "module_wrapper_9 (ModuleWrap (None, 1536)              0         \n",
            "_________________________________________________________________\n",
            "module_wrapper_10 (ModuleWra (None, 128)               196736    \n",
            "_________________________________________________________________\n",
            "module_wrapper_11 (ModuleWra (None, 128)               0         \n",
            "_________________________________________________________________\n",
            "module_wrapper_12 (ModuleWra (None, 2)                 258       \n",
            "=================================================================\n",
            "Total params: 289,666\n",
            "Trainable params: 289,666\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jz3VkXGvFmQg",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "b82c85d7-b3d2-46a5-859d-38391e584ce9"
      },
      "source": [
        "test_X.shape\n",
        "test_y_one_hot.shape"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(6, 2)"
            ]
          },
          "metadata": {},
          "execution_count": 19
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xoNsoTkPFmQh"
      },
      "source": [
        "test_eval = model_.evaluate(test_X, test_y_one_hot, verbose=0)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "mXI_oDQSFmQi",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "581f5e53-f797-4a1f-9a40-9d1afa0dda5a"
      },
      "source": [
        "test_eval"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[6.901703454786912e-05, 1.0]"
            ]
          },
          "metadata": {},
          "execution_count": 21
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "scrolled": false,
        "id": "CmEVhiMWFmQi",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 545
        },
        "outputId": "07abe691-289b-47c7-97e3-61f97145b5f2"
      },
      "source": [
        "accuracy = lfo_train.history['accuracy']\n",
        "val_accuracy = lfo_train.history['val_accuracy']\n",
        "loss = lfo_train.history['loss']\n",
        "val_loss = lfo_train.history['val_loss']\n",
        "epochss = range(len(accuracy))\n",
        "plt.plot(epochss, accuracy, 'bo', label='Training accuracy')\n",
        "plt.plot(epochss, val_accuracy, 'b', label='Validation accuracy')\n",
        "plt.title('Training and validation accuracy')\n",
        "plt.legend()\n",
        "plt.figure()\n",
        "plt.plot(epochss, loss, 'bo', label='Training loss')\n",
        "plt.plot(epochss, val_loss, 'b', label='Validation loss')\n",
        "plt.title('Training and validation loss')\n",
        "plt.legend()\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 296
        },
        "id": "VhDiyFdSbcmJ",
        "outputId": "acd17b0f-2cfa-4ede-b735-64f6ca96d1d7"
      },
      "source": [
        "predicted_classes = model_.predict(test_X)\n",
        "predicted_classes = np.argmax(np.round(predicted_classes),axis=1)\n",
        "ConfusionMatrixDisplay.from_predictions(test_y, predicted_classes, labels = [1,0], cmap=\"cubehelix\", colorbar = 0)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<sklearn.metrics._plot.confusion_matrix.ConfusionMatrixDisplay at 0x7f8ffa5d7f90>"
            ]
          },
          "metadata": {},
          "execution_count": 23
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "r1K-ow22mHKn",
        "outputId": "c5b607a9-7db7-4865-d6a8-f3be5985565c"
      },
      "source": [
        "test_acc = len(np.where(predicted_classes == test_y)[0]) / len(test_y)\n",
        "print(f'CM Acc: {test_acc}')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "CM Acc: 1.0\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "6IgRXDYTJMI0"
      },
      "source": [
        "# Adding the dropout layer to evaluate generalization"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "h0kjwuDUFmQj"
      },
      "source": [
        "#adding dropout layer\n",
        "model_ = Sequential()\n",
        "model_.add(Conv2D(32, kernel_size=(3, 3),activation='linear',padding='same',input_shape=(44,13,1)))\n",
        "model_.add(LeakyReLU(alpha=0.1))\n",
        "model_.add(MaxPooling2D((2, 2),padding='same'))\n",
        "model_.add(Dropout(0.25))\n",
        "model_.add(Conv2D(64, (3, 3), activation='linear',padding='same'))\n",
        "model_.add(LeakyReLU(alpha=0.1))\n",
        "model_.add(MaxPooling2D(pool_size=(2, 2),padding='same'))\n",
        "model_.add(Dropout(0.25))\n",
        "model_.add(Conv2D(128, (3, 3), activation='linear',padding='same'))\n",
        "model_.add(LeakyReLU(alpha=0.1))                  \n",
        "model_.add(MaxPooling2D(pool_size=(2, 2),padding='same'))\n",
        "model_.add(Dropout(0.4))\n",
        "model_.add(Flatten())\n",
        "model_.add(Dense(128, activation='linear'))\n",
        "model_.add(LeakyReLU(alpha=0.1))           \n",
        "model_.add(Dropout(0.3))\n",
        "model_.add(Dense(num_classes, activation='softmax'))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "k3z83aadFmQk"
      },
      "source": [
        "model_.compile(loss=keras.losses.categorical_crossentropy, optimizer=\"adam\",metrics=['accuracy'])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jHvNFGa1FmQl",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "2ca524cd-0fec-4510-992f-16c436b7a136"
      },
      "source": [
        "lfo_train_dropout = model_.fit(train_X, train_label, batch_size=batch_size,epochs=epochs,verbose=1,validation_data=(valid_X, valid_label))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/100\n",
            "5/5 [==============================] - 1s 104ms/step - loss: 21.3382 - accuracy: 0.5474 - val_loss: 9.1844 - val_accuracy: 0.5429\n",
            "Epoch 2/100\n",
            "5/5 [==============================] - 0s 62ms/step - loss: 10.9645 - accuracy: 0.4599 - val_loss: 2.0882 - val_accuracy: 0.4857\n",
            "Epoch 3/100\n",
            "5/5 [==============================] - 0s 58ms/step - loss: 3.8279 - accuracy: 0.5839 - val_loss: 1.6194 - val_accuracy: 0.5429\n",
            "Epoch 4/100\n",
            "5/5 [==============================] - 0s 58ms/step - loss: 2.0410 - accuracy: 0.6058 - val_loss: 0.5017 - val_accuracy: 0.7429\n",
            "Epoch 5/100\n",
            "5/5 [==============================] - 0s 60ms/step - loss: 1.0280 - accuracy: 0.6642 - val_loss: 0.4534 - val_accuracy: 0.7143\n",
            "Epoch 6/100\n",
            "5/5 [==============================] - 0s 58ms/step - loss: 1.2182 - accuracy: 0.6058 - val_loss: 0.6422 - val_accuracy: 0.5714\n",
            "Epoch 7/100\n",
            "5/5 [==============================] - 0s 60ms/step - loss: 0.8217 - accuracy: 0.6496 - val_loss: 0.5019 - val_accuracy: 0.8000\n",
            "Epoch 8/100\n",
            "5/5 [==============================] - 0s 57ms/step - loss: 1.0613 - accuracy: 0.5912 - val_loss: 0.4260 - val_accuracy: 0.8000\n",
            "Epoch 9/100\n",
            "5/5 [==============================] - 0s 60ms/step - loss: 0.9150 - accuracy: 0.7080 - val_loss: 0.3975 - val_accuracy: 0.7429\n",
            "Epoch 10/100\n",
            "5/5 [==============================] - 0s 59ms/step - loss: 0.8667 - accuracy: 0.6350 - val_loss: 0.4207 - val_accuracy: 0.8000\n",
            "Epoch 11/100\n",
            "5/5 [==============================] - 0s 56ms/step - loss: 0.7279 - accuracy: 0.6934 - val_loss: 0.4055 - val_accuracy: 0.8000\n",
            "Epoch 12/100\n",
            "5/5 [==============================] - 0s 59ms/step - loss: 0.6167 - accuracy: 0.7226 - val_loss: 0.4019 - val_accuracy: 0.8000\n",
            "Epoch 13/100\n",
            "5/5 [==============================] - 0s 56ms/step - loss: 0.5353 - accuracy: 0.7737 - val_loss: 0.4002 - val_accuracy: 0.8000\n",
            "Epoch 14/100\n",
            "5/5 [==============================] - 0s 59ms/step - loss: 0.5475 - accuracy: 0.7372 - val_loss: 0.4081 - val_accuracy: 0.7714\n",
            "Epoch 15/100\n",
            "5/5 [==============================] - 0s 57ms/step - loss: 0.5749 - accuracy: 0.7372 - val_loss: 0.3803 - val_accuracy: 0.8000\n",
            "Epoch 16/100\n",
            "5/5 [==============================] - 0s 60ms/step - loss: 0.5641 - accuracy: 0.7445 - val_loss: 0.3850 - val_accuracy: 0.8857\n",
            "Epoch 17/100\n",
            "5/5 [==============================] - 0s 58ms/step - loss: 0.4610 - accuracy: 0.7664 - val_loss: 0.4197 - val_accuracy: 0.8000\n",
            "Epoch 18/100\n",
            "5/5 [==============================] - 0s 58ms/step - loss: 0.4440 - accuracy: 0.8102 - val_loss: 0.3719 - val_accuracy: 0.8286\n",
            "Epoch 19/100\n",
            "5/5 [==============================] - 0s 59ms/step - loss: 0.5251 - accuracy: 0.7299 - val_loss: 0.3691 - val_accuracy: 0.8000\n",
            "Epoch 20/100\n",
            "5/5 [==============================] - 0s 57ms/step - loss: 0.4778 - accuracy: 0.7591 - val_loss: 0.4524 - val_accuracy: 0.7714\n",
            "Epoch 21/100\n",
            "5/5 [==============================] - 0s 57ms/step - loss: 0.4672 - accuracy: 0.7372 - val_loss: 0.3795 - val_accuracy: 0.8000\n",
            "Epoch 22/100\n",
            "5/5 [==============================] - 0s 58ms/step - loss: 0.4368 - accuracy: 0.8175 - val_loss: 0.3710 - val_accuracy: 0.8000\n",
            "Epoch 23/100\n",
            "5/5 [==============================] - 0s 57ms/step - loss: 0.3867 - accuracy: 0.8394 - val_loss: 0.4403 - val_accuracy: 0.7143\n",
            "Epoch 24/100\n",
            "5/5 [==============================] - 0s 59ms/step - loss: 0.4430 - accuracy: 0.7956 - val_loss: 0.3511 - val_accuracy: 0.8857\n",
            "Epoch 25/100\n",
            "5/5 [==============================] - 0s 56ms/step - loss: 0.4586 - accuracy: 0.8248 - val_loss: 0.3411 - val_accuracy: 0.8857\n",
            "Epoch 26/100\n",
            "5/5 [==============================] - 0s 61ms/step - loss: 0.4128 - accuracy: 0.7956 - val_loss: 0.3178 - val_accuracy: 0.8286\n",
            "Epoch 27/100\n",
            "5/5 [==============================] - 0s 56ms/step - loss: 0.3716 - accuracy: 0.8321 - val_loss: 0.2980 - val_accuracy: 0.8857\n",
            "Epoch 28/100\n",
            "5/5 [==============================] - 0s 61ms/step - loss: 0.3908 - accuracy: 0.8321 - val_loss: 0.3511 - val_accuracy: 0.8571\n",
            "Epoch 29/100\n",
            "5/5 [==============================] - 0s 59ms/step - loss: 0.4265 - accuracy: 0.8248 - val_loss: 0.3672 - val_accuracy: 0.8857\n",
            "Epoch 30/100\n",
            "5/5 [==============================] - 0s 56ms/step - loss: 0.3868 - accuracy: 0.8175 - val_loss: 0.2980 - val_accuracy: 0.8857\n",
            "Epoch 31/100\n",
            "5/5 [==============================] - 0s 58ms/step - loss: 0.3747 - accuracy: 0.8321 - val_loss: 0.2814 - val_accuracy: 0.8571\n",
            "Epoch 32/100\n",
            "5/5 [==============================] - 0s 59ms/step - loss: 0.3706 - accuracy: 0.8175 - val_loss: 0.3207 - val_accuracy: 0.8571\n",
            "Epoch 33/100\n",
            "5/5 [==============================] - 0s 59ms/step - loss: 0.3137 - accuracy: 0.8686 - val_loss: 0.2327 - val_accuracy: 0.8857\n",
            "Epoch 34/100\n",
            "5/5 [==============================] - 0s 62ms/step - loss: 0.2948 - accuracy: 0.8467 - val_loss: 0.2253 - val_accuracy: 0.8571\n",
            "Epoch 35/100\n",
            "5/5 [==============================] - 0s 59ms/step - loss: 0.2646 - accuracy: 0.8686 - val_loss: 0.2550 - val_accuracy: 0.8857\n",
            "Epoch 36/100\n",
            "5/5 [==============================] - 0s 61ms/step - loss: 0.2494 - accuracy: 0.8832 - val_loss: 0.2792 - val_accuracy: 0.8857\n",
            "Epoch 37/100\n",
            "5/5 [==============================] - 0s 60ms/step - loss: 0.2056 - accuracy: 0.9051 - val_loss: 0.2098 - val_accuracy: 0.8857\n",
            "Epoch 38/100\n",
            "5/5 [==============================] - 0s 58ms/step - loss: 0.2610 - accuracy: 0.8686 - val_loss: 0.2005 - val_accuracy: 0.8857\n",
            "Epoch 39/100\n",
            "5/5 [==============================] - 0s 58ms/step - loss: 0.1854 - accuracy: 0.9343 - val_loss: 0.2267 - val_accuracy: 0.8571\n",
            "Epoch 40/100\n",
            "5/5 [==============================] - 0s 57ms/step - loss: 0.2177 - accuracy: 0.8832 - val_loss: 0.2009 - val_accuracy: 0.9143\n",
            "Epoch 41/100\n",
            "5/5 [==============================] - 0s 58ms/step - loss: 0.2284 - accuracy: 0.9270 - val_loss: 0.1980 - val_accuracy: 0.9143\n",
            "Epoch 42/100\n",
            "5/5 [==============================] - 0s 57ms/step - loss: 0.2498 - accuracy: 0.9051 - val_loss: 0.1875 - val_accuracy: 0.9143\n",
            "Epoch 43/100\n",
            "5/5 [==============================] - 0s 60ms/step - loss: 0.1869 - accuracy: 0.9416 - val_loss: 0.1795 - val_accuracy: 0.9143\n",
            "Epoch 44/100\n",
            "5/5 [==============================] - 0s 57ms/step - loss: 0.1877 - accuracy: 0.9051 - val_loss: 0.1734 - val_accuracy: 0.9143\n",
            "Epoch 45/100\n",
            "5/5 [==============================] - 0s 58ms/step - loss: 0.1918 - accuracy: 0.9051 - val_loss: 0.1587 - val_accuracy: 0.9143\n",
            "Epoch 46/100\n",
            "5/5 [==============================] - 0s 57ms/step - loss: 0.1802 - accuracy: 0.9051 - val_loss: 0.1781 - val_accuracy: 0.9429\n",
            "Epoch 47/100\n",
            "5/5 [==============================] - 0s 59ms/step - loss: 0.2152 - accuracy: 0.9270 - val_loss: 0.1588 - val_accuracy: 0.9143\n",
            "Epoch 48/100\n",
            "5/5 [==============================] - 0s 56ms/step - loss: 0.2621 - accuracy: 0.8759 - val_loss: 0.1391 - val_accuracy: 0.9143\n",
            "Epoch 49/100\n",
            "5/5 [==============================] - 0s 60ms/step - loss: 0.1911 - accuracy: 0.9343 - val_loss: 0.1609 - val_accuracy: 0.9143\n",
            "Epoch 50/100\n",
            "5/5 [==============================] - 0s 59ms/step - loss: 0.1383 - accuracy: 0.9416 - val_loss: 0.1379 - val_accuracy: 0.9143\n",
            "Epoch 51/100\n",
            "5/5 [==============================] - 0s 57ms/step - loss: 0.1500 - accuracy: 0.9270 - val_loss: 0.1755 - val_accuracy: 0.8857\n",
            "Epoch 52/100\n",
            "5/5 [==============================] - 0s 57ms/step - loss: 0.1283 - accuracy: 0.9562 - val_loss: 0.1607 - val_accuracy: 0.9143\n",
            "Epoch 53/100\n",
            "5/5 [==============================] - 0s 59ms/step - loss: 0.1047 - accuracy: 0.9562 - val_loss: 0.1555 - val_accuracy: 0.9429\n",
            "Epoch 54/100\n",
            "5/5 [==============================] - 0s 61ms/step - loss: 0.1440 - accuracy: 0.9343 - val_loss: 0.1614 - val_accuracy: 0.8857\n",
            "Epoch 55/100\n",
            "5/5 [==============================] - 0s 58ms/step - loss: 0.1289 - accuracy: 0.9562 - val_loss: 0.1472 - val_accuracy: 0.9143\n",
            "Epoch 56/100\n",
            "5/5 [==============================] - 0s 58ms/step - loss: 0.0855 - accuracy: 0.9562 - val_loss: 0.1396 - val_accuracy: 0.9429\n",
            "Epoch 57/100\n",
            "5/5 [==============================] - 0s 60ms/step - loss: 0.1121 - accuracy: 0.9489 - val_loss: 0.1105 - val_accuracy: 0.9429\n",
            "Epoch 58/100\n",
            "5/5 [==============================] - 0s 61ms/step - loss: 0.1020 - accuracy: 0.9489 - val_loss: 0.0898 - val_accuracy: 0.9714\n",
            "Epoch 59/100\n",
            "5/5 [==============================] - 0s 57ms/step - loss: 0.1056 - accuracy: 0.9781 - val_loss: 0.1653 - val_accuracy: 0.9143\n",
            "Epoch 60/100\n",
            "5/5 [==============================] - 0s 61ms/step - loss: 0.1329 - accuracy: 0.9416 - val_loss: 0.1110 - val_accuracy: 0.9143\n",
            "Epoch 61/100\n",
            "5/5 [==============================] - 0s 57ms/step - loss: 0.1114 - accuracy: 0.9635 - val_loss: 0.1388 - val_accuracy: 0.9429\n",
            "Epoch 62/100\n",
            "5/5 [==============================] - 0s 58ms/step - loss: 0.0627 - accuracy: 0.9781 - val_loss: 0.1194 - val_accuracy: 0.9429\n",
            "Epoch 63/100\n",
            "5/5 [==============================] - 0s 57ms/step - loss: 0.0959 - accuracy: 0.9635 - val_loss: 0.1594 - val_accuracy: 0.9429\n",
            "Epoch 64/100\n",
            "5/5 [==============================] - 0s 56ms/step - loss: 0.0893 - accuracy: 0.9489 - val_loss: 0.1152 - val_accuracy: 0.9429\n",
            "Epoch 65/100\n",
            "5/5 [==============================] - 0s 59ms/step - loss: 0.0442 - accuracy: 0.9854 - val_loss: 0.1032 - val_accuracy: 0.9429\n",
            "Epoch 66/100\n",
            "5/5 [==============================] - 0s 57ms/step - loss: 0.0548 - accuracy: 0.9854 - val_loss: 0.0867 - val_accuracy: 0.9143\n",
            "Epoch 67/100\n",
            "5/5 [==============================] - 0s 59ms/step - loss: 0.1017 - accuracy: 0.9635 - val_loss: 0.0859 - val_accuracy: 0.9714\n",
            "Epoch 68/100\n",
            "5/5 [==============================] - 0s 58ms/step - loss: 0.0364 - accuracy: 1.0000 - val_loss: 0.0804 - val_accuracy: 0.9429\n",
            "Epoch 69/100\n",
            "5/5 [==============================] - 0s 63ms/step - loss: 0.0282 - accuracy: 1.0000 - val_loss: 0.0958 - val_accuracy: 0.9429\n",
            "Epoch 70/100\n",
            "5/5 [==============================] - 0s 57ms/step - loss: 0.0675 - accuracy: 0.9781 - val_loss: 0.0864 - val_accuracy: 0.9714\n",
            "Epoch 71/100\n",
            "5/5 [==============================] - 0s 57ms/step - loss: 0.0500 - accuracy: 0.9854 - val_loss: 0.0810 - val_accuracy: 0.9714\n",
            "Epoch 72/100\n",
            "5/5 [==============================] - 0s 58ms/step - loss: 0.0549 - accuracy: 0.9854 - val_loss: 0.1122 - val_accuracy: 0.9143\n",
            "Epoch 73/100\n",
            "5/5 [==============================] - 0s 60ms/step - loss: 0.0552 - accuracy: 0.9708 - val_loss: 0.0674 - val_accuracy: 0.9714\n",
            "Epoch 74/100\n",
            "5/5 [==============================] - 0s 58ms/step - loss: 0.1087 - accuracy: 0.9635 - val_loss: 0.0721 - val_accuracy: 0.9714\n",
            "Epoch 75/100\n",
            "5/5 [==============================] - 0s 58ms/step - loss: 0.1720 - accuracy: 0.9416 - val_loss: 0.1208 - val_accuracy: 0.9714\n",
            "Epoch 76/100\n",
            "5/5 [==============================] - 0s 57ms/step - loss: 0.0409 - accuracy: 0.9927 - val_loss: 0.2069 - val_accuracy: 0.9143\n",
            "Epoch 77/100\n",
            "5/5 [==============================] - 0s 60ms/step - loss: 0.0741 - accuracy: 0.9708 - val_loss: 0.1572 - val_accuracy: 0.9143\n",
            "Epoch 78/100\n",
            "5/5 [==============================] - 0s 59ms/step - loss: 0.0565 - accuracy: 0.9708 - val_loss: 0.1623 - val_accuracy: 0.9429\n",
            "Epoch 79/100\n",
            "5/5 [==============================] - 0s 57ms/step - loss: 0.0595 - accuracy: 0.9635 - val_loss: 0.1702 - val_accuracy: 0.9143\n",
            "Epoch 80/100\n",
            "5/5 [==============================] - 0s 60ms/step - loss: 0.0353 - accuracy: 0.9781 - val_loss: 0.1735 - val_accuracy: 0.9143\n",
            "Epoch 81/100\n",
            "5/5 [==============================] - 0s 59ms/step - loss: 0.0643 - accuracy: 0.9708 - val_loss: 0.1745 - val_accuracy: 0.9429\n",
            "Epoch 82/100\n",
            "5/5 [==============================] - 0s 59ms/step - loss: 0.0411 - accuracy: 0.9927 - val_loss: 0.1302 - val_accuracy: 0.9429\n",
            "Epoch 83/100\n",
            "5/5 [==============================] - 0s 59ms/step - loss: 0.0615 - accuracy: 0.9708 - val_loss: 0.1122 - val_accuracy: 0.9429\n",
            "Epoch 84/100\n",
            "5/5 [==============================] - 0s 59ms/step - loss: 0.0533 - accuracy: 0.9854 - val_loss: 0.1328 - val_accuracy: 0.9429\n",
            "Epoch 85/100\n",
            "5/5 [==============================] - 0s 61ms/step - loss: 0.0253 - accuracy: 0.9927 - val_loss: 0.1185 - val_accuracy: 0.9429\n",
            "Epoch 86/100\n",
            "5/5 [==============================] - 0s 60ms/step - loss: 0.0309 - accuracy: 0.9854 - val_loss: 0.1055 - val_accuracy: 0.9143\n",
            "Epoch 87/100\n",
            "5/5 [==============================] - 0s 58ms/step - loss: 0.0042 - accuracy: 1.0000 - val_loss: 0.1323 - val_accuracy: 0.9429\n",
            "Epoch 88/100\n",
            "5/5 [==============================] - 0s 63ms/step - loss: 0.0605 - accuracy: 0.9781 - val_loss: 0.1193 - val_accuracy: 0.9429\n",
            "Epoch 89/100\n",
            "5/5 [==============================] - 0s 59ms/step - loss: 0.0922 - accuracy: 0.9635 - val_loss: 0.3152 - val_accuracy: 0.9143\n",
            "Epoch 90/100\n",
            "5/5 [==============================] - 0s 59ms/step - loss: 0.0949 - accuracy: 0.9781 - val_loss: 0.1063 - val_accuracy: 0.9714\n",
            "Epoch 91/100\n",
            "5/5 [==============================] - 0s 60ms/step - loss: 0.0595 - accuracy: 0.9854 - val_loss: 0.0932 - val_accuracy: 0.9429\n",
            "Epoch 92/100\n",
            "5/5 [==============================] - 0s 59ms/step - loss: 0.0581 - accuracy: 0.9854 - val_loss: 0.1438 - val_accuracy: 0.9714\n",
            "Epoch 93/100\n",
            "5/5 [==============================] - 0s 59ms/step - loss: 0.1657 - accuracy: 0.9489 - val_loss: 0.0474 - val_accuracy: 0.9714\n",
            "Epoch 94/100\n",
            "5/5 [==============================] - 0s 59ms/step - loss: 0.1019 - accuracy: 0.9562 - val_loss: 0.1292 - val_accuracy: 0.9143\n",
            "Epoch 95/100\n",
            "5/5 [==============================] - 0s 60ms/step - loss: 0.1097 - accuracy: 0.9635 - val_loss: 0.0833 - val_accuracy: 0.9429\n",
            "Epoch 96/100\n",
            "5/5 [==============================] - 0s 59ms/step - loss: 0.0570 - accuracy: 0.9854 - val_loss: 0.0725 - val_accuracy: 0.9429\n",
            "Epoch 97/100\n",
            "5/5 [==============================] - 0s 58ms/step - loss: 0.0159 - accuracy: 1.0000 - val_loss: 0.0566 - val_accuracy: 0.9714\n",
            "Epoch 98/100\n",
            "5/5 [==============================] - 0s 62ms/step - loss: 0.0683 - accuracy: 0.9781 - val_loss: 0.0585 - val_accuracy: 0.9714\n",
            "Epoch 99/100\n",
            "5/5 [==============================] - 0s 58ms/step - loss: 0.0738 - accuracy: 0.9781 - val_loss: 0.0598 - val_accuracy: 0.9429\n",
            "Epoch 100/100\n",
            "5/5 [==============================] - 0s 58ms/step - loss: 0.0445 - accuracy: 0.9854 - val_loss: 0.0621 - val_accuracy: 0.9714\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "IRDlmyq3FmQm",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "717adb84-d218-4dc6-9550-fb6f292c4c72"
      },
      "source": [
        "test_eval = model_.evaluate(test_X, test_y_one_hot, verbose=1)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "1/1 [==============================] - 0s 21ms/step - loss: 2.9913e-04 - accuracy: 1.0000\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "scrolled": true,
        "id": "Y1LIPBfNFmQn",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "ecd43fac-f614-4516-d5e1-425ade813597"
      },
      "source": [
        "test_eval"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.00029913129401393235, 1.0]"
            ]
          },
          "metadata": {},
          "execution_count": 29
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "scrolled": false,
        "id": "qlM59W_HFmQn",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 545
        },
        "outputId": "9fc62ad6-181e-4ba0-ec09-646a40346641"
      },
      "source": [
        "accuracy = lfo_train_dropout.history['accuracy']\n",
        "val_accuracy = lfo_train_dropout.history['val_accuracy']\n",
        "loss = lfo_train_dropout.history['loss']\n",
        "val_loss = lfo_train_dropout.history['val_loss']\n",
        "epochss = range(len(accuracy))\n",
        "plt.plot(epochss, accuracy, 'bo', label='Training accuracy')\n",
        "plt.plot(epochss, val_accuracy, 'b', label='Validation accuracy')\n",
        "plt.title('Training and validation accuracy')\n",
        "plt.legend()\n",
        "plt.figure()\n",
        "plt.plot(epochss, loss, 'bo', label='Training loss')\n",
        "plt.plot(epochss, val_loss, 'b', label='Validation loss')\n",
        "plt.title('Training and validation loss')\n",
        "plt.legend()\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lXEAaUMjFmQo",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "1cbd1d49-705d-4e9c-a2eb-e07a38c6da3b"
      },
      "source": [
        "predicted_classes = model_.predict(test_X)\n",
        "predicted_classes = np.argmax(np.round(predicted_classes),axis=1)\n",
        "correct = np.where(predicted_classes==test_y)[0]\n",
        "print(len(correct)/len(predicted_classes))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "1.0\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 296
        },
        "id": "uxls3wa0hGc0",
        "outputId": "5e2d6bd9-6c02-4661-d104-991b78b6880d"
      },
      "source": [
        "predicted_classes = model_.predict(test_X)\n",
        "predicted_classes = np.argmax(np.round(predicted_classes),axis=1)\n",
        "ConfusionMatrixDisplay.from_predictions(test_y, predicted_classes, labels = [1,0], cmap=\"cubehelix\", colorbar = 0)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<sklearn.metrics._plot.confusion_matrix.ConfusionMatrixDisplay at 0x7f90009eded0>"
            ]
          },
          "metadata": {},
          "execution_count": 32
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "KNIyFDg3FmQp",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "f292d7b8-4359-492b-93b1-1b6341350286"
      },
      "source": [
        "target_names = [\"Class {}\".format(i) for i in range(num_classes)]\n",
        "print(classification_report(test_y, predicted_classes, target_names=target_names))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "     Class 0       1.00      1.00      1.00         1\n",
            "     Class 1       1.00      1.00      1.00         5\n",
            "\n",
            "    accuracy                           1.00         6\n",
            "   macro avg       1.00      1.00      1.00         6\n",
            "weighted avg       1.00      1.00      1.00         6\n",
            "\n"
          ]
        }
      ]
    }
  ]
}