test-encoder-output-with-cnn.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": [],
      "gpuType": "T4",
      "machine_shape": "hm"
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "code",
      "execution_count": 2,
      "metadata": {
        "id": "FT77L-YnoywR",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "43b8131f-e3fb-4bca-ca60-bc02051e1ab4"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Mounted at /gdrive\n",
            "(2500, 868)\n",
            "Min value after normalization: -11.391450116125661\n",
            "Max value after normalization: 12.467208378606498\n",
            "mu: -1.5717074339393922e-18\n",
            "sigma: 0.9999999999999999\n"
          ]
        }
      ],
      "source": [
        "import numpy as np\n",
        "import os\n",
        "from sklearn.model_selection import train_test_split\n",
        "from sklearn.utils.class_weight import compute_class_weight\n",
        "from keras.utils import to_categorical\n",
        "import matplotlib.pyplot as plt\n",
        "\n",
        "# Configuración Google Drive\n",
        "from google.colab import drive\n",
        "drive.mount('/gdrive')\n",
        "\n",
        "# Function to segment data into 5-second sections\n",
        "def segment_signal(data, sampling_rate=173.6, segment_length=5.0, overlap=1.4):\n",
        "    samples_per_segment = int(sampling_rate * segment_length)\n",
        "    overlap_samples = int(sampling_rate * overlap)\n",
        "\n",
        "    segments = []\n",
        "    for i in range(4):  # First four segments\n",
        "        start_idx = i * samples_per_segment\n",
        "        segments.append(data[start_idx:start_idx + samples_per_segment])\n",
        "\n",
        "    # Fifth segment with overlap\n",
        "    start_idx = 3 * samples_per_segment - overlap_samples\n",
        "    segments.append(data[start_idx:start_idx + samples_per_segment])\n",
        "\n",
        "    return segments\n",
        "\n",
        "# Function to read, label, and segment the data\n",
        "def read_label_and_segment(folder, label, sampling_rate=173.6):\n",
        "    base_path = '/gdrive/MyDrive/TEG-EEG/dataset_bonn/'\n",
        "    full_path = os.path.join(base_path, folder)\n",
        "    files = os.listdir(full_path)\n",
        "    segmented_data = []\n",
        "    labels = []\n",
        "    for file in files:\n",
        "        filepath = os.path.join(full_path, file)\n",
        "        with open(filepath, 'r') as f:\n",
        "            samples = np.loadtxt(f)\n",
        "            # Segment each file into 5-second sections\n",
        "            segments = segment_signal(samples, sampling_rate=sampling_rate)\n",
        "            segmented_data.extend(segments)\n",
        "            labels.extend([label] * len(segments))\n",
        "    return np.array(segmented_data), np.array(labels)\n",
        "\n",
        "# Read, label, and segment data\n",
        "data_ZO, labels_ZO = read_label_and_segment('Z', 0)\n",
        "data_O, labels_O = read_label_and_segment('O', 0)\n",
        "data_NF, labels_NF = read_label_and_segment('N', 1)\n",
        "data_F, labels_F = read_label_and_segment('F', 1)\n",
        "data_S, labels_S = read_label_and_segment('S', 2)\n",
        "\n",
        "# Combine the datasets\n",
        "all_data = np.concatenate([data_ZO, data_O, data_NF, data_F, data_S])\n",
        "all_labels = np.concatenate([labels_ZO, labels_O, labels_NF, labels_F, labels_S])\n",
        "\n",
        "print(all_data.shape)\n",
        "\n",
        "# Normalize the data\n",
        "mu, sigma = np.mean(all_data), np.std(all_data)\n",
        "all_data_normalized = (all_data - mu) / sigma\n",
        "\n",
        "# Find the minimum and maximum values in the dataset\n",
        "min_value = np.min(all_data)\n",
        "max_value = np.max(all_data)\n",
        "\n",
        "# Normalize the data between -1 and 1\n",
        "#all_data_normalized = 2 * (all_data - min_value) / (max_value - min_value) - 1\n",
        "\n",
        "# Check the new min and max values to ensure normalization\n",
        "print(f'Min value after normalization: {np.min(all_data_normalized)}')\n",
        "print(f'Max value after normalization: {np.max(all_data_normalized)}')\n",
        "\n",
        "print(f'mu: {np.mean(all_data_normalized)}')\n",
        "print(f'sigma: {np.std(all_data_normalized)}')\n",
        "\n",
        "# Calculate class weights\n",
        "class_weights = compute_class_weight(class_weight='balanced', classes=np.unique(all_labels), y=all_labels)\n",
        "class_weights_dict = dict(enumerate(class_weights))\n"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "\n",
        "import tensorflow as tf\n",
        "\n",
        "from keras.utils import to_categorical\n",
        "\n",
        "def window_data(X, Y, wsize, overlap):\n",
        "    X_v = []\n",
        "    Y_v = []\n",
        "\n",
        "    # Calculate stride based on overlap\n",
        "    stride = wsize - overlap\n",
        "\n",
        "    for x, y in zip(X, Y):\n",
        "        # Window the signal \"x\" and create corresponding labels\n",
        "        for start in range(0, len(x) - wsize + 1, stride):\n",
        "            x_v = x[start:start + wsize]\n",
        "            X_v.append(x_v)\n",
        "            Y_v.append(y)  # Repeat the label for each window\n",
        "\n",
        "    # Convert the lists X_v, Y_v to NumPy arrays\n",
        "    X_v = np.array(X_v)\n",
        "    Y_v = np.array(Y_v)\n",
        "\n",
        "    # Shuffling the data\n",
        "    #p = np.random.permutation(len(X_v))\n",
        "    #X_v = X_v[p]\n",
        "    #Y_v = Y_v[p]\n",
        "\n",
        "    return X_v, Y_v.flatten()\n",
        "\n",
        "\n",
        "def split_data(data, labels, num_classes=3):\n",
        "    # Convert labels to one-hot encoding\n",
        "    one_hot_labels = to_categorical(labels, num_classes=num_classes)\n",
        "\n",
        "    # Split the dataset into training, validation, and test sets\n",
        "    train_data, test_data, train_labels, test_labels = train_test_split(data, one_hot_labels, test_size=0.2, random_state=42, shuffle=True)\n",
        "    test_data, val_data, test_labels, val_labels = train_test_split(test_data, test_labels, test_size=0.5, random_state=42, shuffle=True)\n",
        "\n",
        "\n",
        "    # Split the dataset into training, validation, and test sets\n",
        "    return train_data, val_data, test_data, train_labels, val_labels, test_labels\n",
        "\n",
        "# Example usage\n",
        "window_size = 128  # Example value\n",
        "overlap = 0        # Overlap can be set to zero\n",
        "\n",
        "windowed_data, windowed_labels = window_data(all_data_normalized, all_labels, window_size, overlap)\n",
        "\n",
        "print(windowed_data.shape)\n",
        "print(windowed_labels.shape)\n",
        "\n",
        "X_train, X_val, X_test, y_train, y_val, y_test = split_data(windowed_data, windowed_labels, num_classes=3)  # Assuming 3 classes\n",
        "\n",
        "print(X_train.shape)\n",
        "print(X_val.shape)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "uQ6bMz5MpMGf",
        "outputId": "36d81eaa-317a-419b-fa94-ff2ffa6c4133"
      },
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "(15000, 128)\n",
            "(15000,)\n",
            "(12000, 128)\n",
            "(1500, 128)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# test new autoencoder\n",
        "\n",
        "import tensorflow as tf\n",
        "\n",
        "func1 = 'relu'\n",
        "func2 = 'linear'\n",
        "\n",
        "tf.random.set_seed(157) # Reproducibilidad del entrenamiento\n",
        "# Definimos la entrada\n",
        "entrada = tf.keras.layers.Input(shape=(window_size,))\n",
        "\n",
        "def encoder_function(entrada):\n",
        "\n",
        "  # Agregamos una capa de convolución con 32 filtros, tamaño de kernel de 3, y activación relu\n",
        "  entrada_convolucional = tf.keras.layers.Reshape((window_size, 1))(entrada)\n",
        "  conv1 = tf.keras.layers.Conv1D(16, 3, padding='same', activation=func1)(entrada_convolucional)\n",
        "\n",
        "  # Agregamos una capa de pooling con tamaño de pool de 2\n",
        "  maxpool1 = tf.keras.layers.MaxPooling1D(2)(conv1)\n",
        "  maxpool1 = tf.keras.layers.Dropout(0.05)(maxpool1)\n",
        "\n",
        "  encoder = tf.keras.layers.Dense(window_size, activation = func1,\n",
        "                                  name='bottleneck')(maxpool1)\n",
        "  encoder = tf.keras.layers.ActivityRegularization(l1=0.004)(encoder)\n",
        "\n",
        "  # Aplicamos dropout para intentar reducir el overfitting\n",
        "  encoder = tf.keras.layers.Dropout(0.05)(encoder)\n",
        "\n",
        "\n",
        "  return encoder\n",
        "\n",
        "def decorder_function(encoder_input):\n",
        "  # Agregamos una capa de Reshape para aplanar la salida y pasarla a la capa conv1d_transpose_52\n",
        "  transpuesta1 = tf.keras.layers.Conv1DTranspose(16, 3, padding='same', activation=func1)(encoder_input)\n",
        "  transpuesta1 = tf.keras.layers.UpSampling1D(2)(transpuesta1)\n",
        "\n",
        "  decoder_output = tf.keras.layers.Conv1DTranspose(1, 3, padding='same', activation=func2)(transpuesta1)\n",
        "\n",
        "  # Aplicamos dropout para intentar reducir el overfitting\n",
        "  decoder_output = tf.keras.layers.Dropout(0.05)(decoder_output)\n",
        "  decoder_output = tf.keras.layers.Reshape((window_size,))(decoder_output)\n",
        "  decoder_output = tf.keras.layers.Dense(window_size, activation=func2)(decoder_output)\n",
        "\n",
        "  return decoder_output\n",
        "\n",
        "# Definimos una red CNN simple\n",
        "def cnn_simple(input_layer):\n",
        "    x = tf.keras.layers.Reshape((window_size, 1))(input_layer)  # Redimensionamos la entrada para la CNN\n",
        "    x = tf.keras.layers.Conv1D(32, kernel_size=3, activation='linear', padding='same')(x)\n",
        "    x = tf.keras.layers.MaxPooling1D(pool_size=2)(x)\n",
        "    x = tf.keras.layers.Dropout(0.2)(x)\n",
        "\n",
        "    x = tf.keras.layers.Conv1D(64, kernel_size=3, activation='linear', padding='same')(x)\n",
        "    x = tf.keras.layers.MaxPooling1D(pool_size=2)(x)\n",
        "    x = tf.keras.layers.Dropout(0.2)(x)\n",
        "\n",
        "    x = tf.keras.layers.Flatten()(x)\n",
        "    return x\n",
        "\n",
        "\n",
        "# Definimos el autoencoder como el modelo que toma la entrada y devuelve la salida del decoder\n",
        "autoencoder = tf.keras.models.Model(entrada, decorder_function(encoder_function(entrada)))\n",
        "\n",
        "optimizer = tf.keras.optimizers.Adam(learning_rate=0.0001)\n",
        "\n",
        "autoencoder.compile( loss='mse',\n",
        "    optimizer=optimizer)\n",
        "\n",
        "#autoencoder.build((None, input_size))\n",
        "\n",
        "autoencoder.summary()\n",
        "\n",
        "# Adding Gaussian noise to the training data\n",
        "\n",
        "\n",
        "noise_train = np.random.normal(loc=0.0, scale=0.1, size=X_train.shape)\n",
        "noise_val = np.random.normal(loc=0.0, scale=0.1, size=X_val.shape)\n",
        "\n",
        "X_train_noisy = X_train + noise_train\n",
        "X_val_noisy = X_val + noise_val\n",
        "\n",
        "# Training the DSAE\n",
        "# -------------------------------------------------\n",
        "# Train the autoencoder\n",
        "history = autoencoder.fit(\n",
        "    x = X_train_noisy,\n",
        "    y = X_train_noisy,\n",
        "    batch_size=32,\n",
        "    epochs = 50,\n",
        "    validation_data = (X_val_noisy,X_val_noisy),\n",
        "    verbose=2\n",
        ")"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "J6sy6Y39pPAi",
        "outputId": "d5ebccf5-a918-4673-8230-63dfecfb5424"
      },
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "\u001b[1mModel: \"functional\"\u001b[0m\n"
            ],
            "text/html": [
              "<pre style=\"white-space:pre;overflow-x:auto;line-height:normal;font-family:Menlo,'DejaVu Sans Mono',consolas,'Courier New',monospace\"><span style=\"font-weight: bold\">Model: \"functional\"</span>\n",
              "</pre>\n"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━┓\n",
              "┃\u001b[1m \u001b[0m\u001b[1mLayer (type)                        \u001b[0m\u001b[1m \u001b[0m┃\u001b[1m \u001b[0m\u001b[1mOutput Shape               \u001b[0m\u001b[1m \u001b[0m┃\u001b[1m \u001b[0m\u001b[1m        Param #\u001b[0m\u001b[1m \u001b[0m┃\n",
              "┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━┩\n",
              "│ input_layer (\u001b[38;5;33mInputLayer\u001b[0m)             │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m128\u001b[0m)                 │               \u001b[38;5;34m0\u001b[0m │\n",
              "├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤\n",
              "│ reshape (\u001b[38;5;33mReshape\u001b[0m)                    │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m128\u001b[0m, \u001b[38;5;34m1\u001b[0m)              │               \u001b[38;5;34m0\u001b[0m │\n",
              "├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤\n",
              "│ conv1d (\u001b[38;5;33mConv1D\u001b[0m)                      │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m128\u001b[0m, \u001b[38;5;34m16\u001b[0m)             │              \u001b[38;5;34m64\u001b[0m │\n",
              "├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤\n",
              "│ max_pooling1d (\u001b[38;5;33mMaxPooling1D\u001b[0m)         │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m64\u001b[0m, \u001b[38;5;34m16\u001b[0m)              │               \u001b[38;5;34m0\u001b[0m │\n",
              "├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤\n",
              "│ dropout (\u001b[38;5;33mDropout\u001b[0m)                    │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m64\u001b[0m, \u001b[38;5;34m16\u001b[0m)              │               \u001b[38;5;34m0\u001b[0m │\n",
              "├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤\n",
              "│ bottleneck (\u001b[38;5;33mDense\u001b[0m)                   │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m64\u001b[0m, \u001b[38;5;34m128\u001b[0m)             │           \u001b[38;5;34m2,176\u001b[0m │\n",
              "├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤\n",
              "│ activity_regularization              │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m64\u001b[0m, \u001b[38;5;34m128\u001b[0m)             │               \u001b[38;5;34m0\u001b[0m │\n",
              "│ (\u001b[38;5;33mActivityRegularization\u001b[0m)             │                             │                 │\n",
              "├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤\n",
              "│ dropout_1 (\u001b[38;5;33mDropout\u001b[0m)                  │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m64\u001b[0m, \u001b[38;5;34m128\u001b[0m)             │               \u001b[38;5;34m0\u001b[0m │\n",
              "├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤\n",
              "│ conv1d_transpose (\u001b[38;5;33mConv1DTranspose\u001b[0m)   │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m64\u001b[0m, \u001b[38;5;34m16\u001b[0m)              │           \u001b[38;5;34m6,160\u001b[0m │\n",
              "├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤\n",
              "│ up_sampling1d (\u001b[38;5;33mUpSampling1D\u001b[0m)         │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m128\u001b[0m, \u001b[38;5;34m16\u001b[0m)             │               \u001b[38;5;34m0\u001b[0m │\n",
              "├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤\n",
              "│ conv1d_transpose_1 (\u001b[38;5;33mConv1DTranspose\u001b[0m) │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m128\u001b[0m, \u001b[38;5;34m1\u001b[0m)              │              \u001b[38;5;34m49\u001b[0m │\n",
              "├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤\n",
              "│ dropout_2 (\u001b[38;5;33mDropout\u001b[0m)                  │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m128\u001b[0m, \u001b[38;5;34m1\u001b[0m)              │               \u001b[38;5;34m0\u001b[0m │\n",
              "├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤\n",
              "│ reshape_1 (\u001b[38;5;33mReshape\u001b[0m)                  │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m128\u001b[0m)                 │               \u001b[38;5;34m0\u001b[0m │\n",
              "├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤\n",
              "│ dense (\u001b[38;5;33mDense\u001b[0m)                        │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m128\u001b[0m)                 │          \u001b[38;5;34m16,512\u001b[0m │\n",
              "└──────────────────────────────────────┴─────────────────────────────┴─────────────────┘\n"
            ],
            "text/html": [
              "<pre style=\"white-space:pre;overflow-x:auto;line-height:normal;font-family:Menlo,'DejaVu Sans Mono',consolas,'Courier New',monospace\">┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━┓\n",
              "┃<span style=\"font-weight: bold\"> Layer (type)                         </span>┃<span style=\"font-weight: bold\"> Output Shape                </span>┃<span style=\"font-weight: bold\">         Param # </span>┃\n",
              "┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━┩\n",
              "│ input_layer (<span style=\"color: #0087ff; text-decoration-color: #0087ff\">InputLayer</span>)             │ (<span style=\"color: #00d7ff; text-decoration-color: #00d7ff\">None</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">128</span>)                 │               <span style=\"color: #00af00; text-decoration-color: #00af00\">0</span> │\n",
              "├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤\n",
              "│ reshape (<span style=\"color: #0087ff; text-decoration-color: #0087ff\">Reshape</span>)                    │ (<span style=\"color: #00d7ff; text-decoration-color: #00d7ff\">None</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">128</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">1</span>)              │               <span style=\"color: #00af00; text-decoration-color: #00af00\">0</span> │\n",
              "├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤\n",
              "│ conv1d (<span style=\"color: #0087ff; text-decoration-color: #0087ff\">Conv1D</span>)                      │ (<span style=\"color: #00d7ff; text-decoration-color: #00d7ff\">None</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">128</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">16</span>)             │              <span style=\"color: #00af00; text-decoration-color: #00af00\">64</span> │\n",
              "├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤\n",
              "│ max_pooling1d (<span style=\"color: #0087ff; text-decoration-color: #0087ff\">MaxPooling1D</span>)         │ (<span style=\"color: #00d7ff; text-decoration-color: #00d7ff\">None</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">64</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">16</span>)              │               <span style=\"color: #00af00; text-decoration-color: #00af00\">0</span> │\n",
              "├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤\n",
              "│ dropout (<span style=\"color: #0087ff; text-decoration-color: #0087ff\">Dropout</span>)                    │ (<span style=\"color: #00d7ff; text-decoration-color: #00d7ff\">None</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">64</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">16</span>)              │               <span style=\"color: #00af00; text-decoration-color: #00af00\">0</span> │\n",
              "├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤\n",
              "│ bottleneck (<span style=\"color: #0087ff; text-decoration-color: #0087ff\">Dense</span>)                   │ (<span style=\"color: #00d7ff; text-decoration-color: #00d7ff\">None</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">64</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">128</span>)             │           <span style=\"color: #00af00; text-decoration-color: #00af00\">2,176</span> │\n",
              "├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤\n",
              "│ activity_regularization              │ (<span style=\"color: #00d7ff; text-decoration-color: #00d7ff\">None</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">64</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">128</span>)             │               <span style=\"color: #00af00; text-decoration-color: #00af00\">0</span> │\n",
              "│ (<span style=\"color: #0087ff; text-decoration-color: #0087ff\">ActivityRegularization</span>)             │                             │                 │\n",
              "├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤\n",
              "│ dropout_1 (<span style=\"color: #0087ff; text-decoration-color: #0087ff\">Dropout</span>)                  │ (<span style=\"color: #00d7ff; text-decoration-color: #00d7ff\">None</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">64</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">128</span>)             │               <span style=\"color: #00af00; text-decoration-color: #00af00\">0</span> │\n",
              "├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤\n",
              "│ conv1d_transpose (<span style=\"color: #0087ff; text-decoration-color: #0087ff\">Conv1DTranspose</span>)   │ (<span style=\"color: #00d7ff; text-decoration-color: #00d7ff\">None</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">64</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">16</span>)              │           <span style=\"color: #00af00; text-decoration-color: #00af00\">6,160</span> │\n",
              "├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤\n",
              "│ up_sampling1d (<span style=\"color: #0087ff; text-decoration-color: #0087ff\">UpSampling1D</span>)         │ (<span style=\"color: #00d7ff; text-decoration-color: #00d7ff\">None</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">128</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">16</span>)             │               <span style=\"color: #00af00; text-decoration-color: #00af00\">0</span> │\n",
              "├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤\n",
              "│ conv1d_transpose_1 (<span style=\"color: #0087ff; text-decoration-color: #0087ff\">Conv1DTranspose</span>) │ (<span style=\"color: #00d7ff; text-decoration-color: #00d7ff\">None</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">128</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">1</span>)              │              <span style=\"color: #00af00; text-decoration-color: #00af00\">49</span> │\n",
              "├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤\n",
              "│ dropout_2 (<span style=\"color: #0087ff; text-decoration-color: #0087ff\">Dropout</span>)                  │ (<span style=\"color: #00d7ff; text-decoration-color: #00d7ff\">None</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">128</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">1</span>)              │               <span style=\"color: #00af00; text-decoration-color: #00af00\">0</span> │\n",
              "├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤\n",
              "│ reshape_1 (<span style=\"color: #0087ff; text-decoration-color: #0087ff\">Reshape</span>)                  │ (<span style=\"color: #00d7ff; text-decoration-color: #00d7ff\">None</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">128</span>)                 │               <span style=\"color: #00af00; text-decoration-color: #00af00\">0</span> │\n",
              "├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤\n",
              "│ dense (<span style=\"color: #0087ff; text-decoration-color: #0087ff\">Dense</span>)                        │ (<span style=\"color: #00d7ff; text-decoration-color: #00d7ff\">None</span>, <span style=\"color: #00af00; text-decoration-color: #00af00\">128</span>)                 │          <span style=\"color: #00af00; text-decoration-color: #00af00\">16,512</span> │\n",
              "└──────────────────────────────────────┴─────────────────────────────┴─────────────────┘\n",
              "</pre>\n"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "\u001b[1m Total params: \u001b[0m\u001b[38;5;34m24,961\u001b[0m (97.50 KB)\n"
            ],
            "text/html": [
              "<pre style=\"white-space:pre;overflow-x:auto;line-height:normal;font-family:Menlo,'DejaVu Sans Mono',consolas,'Courier New',monospace\"><span style=\"font-weight: bold\"> Total params: </span><span style=\"color: #00af00; text-decoration-color: #00af00\">24,961</span> (97.50 KB)\n",
              "</pre>\n"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "\u001b[1m Trainable params: \u001b[0m\u001b[38;5;34m24,961\u001b[0m (97.50 KB)\n"
            ],
            "text/html": [
              "<pre style=\"white-space:pre;overflow-x:auto;line-height:normal;font-family:Menlo,'DejaVu Sans Mono',consolas,'Courier New',monospace\"><span style=\"font-weight: bold\"> Trainable params: </span><span style=\"color: #00af00; text-decoration-color: #00af00\">24,961</span> (97.50 KB)\n",
              "</pre>\n"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "\u001b[1m Non-trainable params: \u001b[0m\u001b[38;5;34m0\u001b[0m (0.00 B)\n"
            ],
            "text/html": [
              "<pre style=\"white-space:pre;overflow-x:auto;line-height:normal;font-family:Menlo,'DejaVu Sans Mono',consolas,'Courier New',monospace\"><span style=\"font-weight: bold\"> Non-trainable params: </span><span style=\"color: #00af00; text-decoration-color: #00af00\">0</span> (0.00 B)\n",
              "</pre>\n"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/50\n",
            "375/375 - 8s - 20ms/step - loss: 12.8668 - val_loss: 5.2848\n",
            "Epoch 2/50\n",
            "375/375 - 1s - 2ms/step - loss: 3.1868 - val_loss: 1.9182\n",
            "Epoch 3/50\n",
            "375/375 - 1s - 2ms/step - loss: 1.5176 - val_loss: 1.2190\n",
            "Epoch 4/50\n",
            "375/375 - 1s - 2ms/step - loss: 1.0823 - val_loss: 0.9804\n",
            "Epoch 5/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.8743 - val_loss: 0.8059\n",
            "Epoch 6/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.7288 - val_loss: 0.6731\n",
            "Epoch 7/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.6215 - val_loss: 0.5777\n",
            "Epoch 8/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.5424 - val_loss: 0.5010\n",
            "Epoch 9/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.4804 - val_loss: 0.4367\n",
            "Epoch 10/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.4286 - val_loss: 0.3854\n",
            "Epoch 11/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.3889 - val_loss: 0.3443\n",
            "Epoch 12/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.3527 - val_loss: 0.3095\n",
            "Epoch 13/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.3245 - val_loss: 0.2819\n",
            "Epoch 14/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.3017 - val_loss: 0.2601\n",
            "Epoch 15/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.2802 - val_loss: 0.2400\n",
            "Epoch 16/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.2617 - val_loss: 0.2231\n",
            "Epoch 17/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.2463 - val_loss: 0.2076\n",
            "Epoch 18/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.2316 - val_loss: 0.1947\n",
            "Epoch 19/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.2193 - val_loss: 0.1825\n",
            "Epoch 20/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.2092 - val_loss: 0.1733\n",
            "Epoch 21/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.2003 - val_loss: 0.1645\n",
            "Epoch 22/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.1921 - val_loss: 0.1579\n",
            "Epoch 23/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.1824 - val_loss: 0.1503\n",
            "Epoch 24/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.1760 - val_loss: 0.1451\n",
            "Epoch 25/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.1691 - val_loss: 0.1407\n",
            "Epoch 26/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.1650 - val_loss: 0.1319\n",
            "Epoch 27/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.1586 - val_loss: 0.1298\n",
            "Epoch 28/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.1547 - val_loss: 0.1239\n",
            "Epoch 29/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.1505 - val_loss: 0.1202\n",
            "Epoch 30/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.1468 - val_loss: 0.1170\n",
            "Epoch 31/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.1424 - val_loss: 0.1146\n",
            "Epoch 32/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.1394 - val_loss: 0.1098\n",
            "Epoch 33/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.1361 - val_loss: 0.1065\n",
            "Epoch 34/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.1333 - val_loss: 0.1064\n",
            "Epoch 35/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.1300 - val_loss: 0.1031\n",
            "Epoch 36/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.1260 - val_loss: 0.1009\n",
            "Epoch 37/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.1239 - val_loss: 0.1006\n",
            "Epoch 38/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.1214 - val_loss: 0.0936\n",
            "Epoch 39/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.1191 - val_loss: 0.0992\n",
            "Epoch 40/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.1170 - val_loss: 0.0884\n",
            "Epoch 41/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.1161 - val_loss: 0.0876\n",
            "Epoch 42/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.1122 - val_loss: 0.0875\n",
            "Epoch 43/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.1119 - val_loss: 0.0844\n",
            "Epoch 44/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.1099 - val_loss: 0.0829\n",
            "Epoch 45/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.1068 - val_loss: 0.0828\n",
            "Epoch 46/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.1056 - val_loss: 0.0812\n",
            "Epoch 47/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.1038 - val_loss: 0.0770\n",
            "Epoch 48/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.1028 - val_loss: 0.0770\n",
            "Epoch 49/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.1012 - val_loss: 0.0788\n",
            "Epoch 50/50\n",
            "375/375 - 1s - 2ms/step - loss: 0.1002 - val_loss: 0.0723\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import matplotlib.pyplot as plt\n",
        "\n",
        "# Plotting\n",
        "plt.plot(history.history['loss'], label='Training Loss')\n",
        "plt.plot(history.history['val_loss'], label='Validation Loss')\n",
        "plt.title('Training and Validation Loss')\n",
        "plt.ylabel('Loss')\n",
        "plt.xlabel('Epoch')\n",
        "plt.legend()\n",
        "plt.show()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 472
        },
        "id": "8go9v8scpX-F",
        "outputId": "1f1512ba-a330-4f66-db5f-9636839cabf5"
      },
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Classifier"
      ],
      "metadata": {
        "id": "HtmvDpE3YPO5"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "encode = tf.keras.models.Model(inputs=autoencoder.input,\n",
        "                               outputs=autoencoder.get_layer('bottleneck').output)\n",
        "encode.output"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "euoinKCHJxMD",
        "outputId": "2ee4dc64-c0af-421f-b199-ae5660e9e393"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<KerasTensor shape=(None, 64, 128), dtype=float32, sparse=False, name=keras_tensor_19>"
            ]
          },
          "metadata": {},
          "execution_count": 22
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import numpy as np\n",
        "import tensorflow as tf\n",
        "from tensorflow.keras.models import Model\n",
        "from tensorflow.keras.layers import LSTM, Reshape, Conv1D, MaxPooling1D, LeakyReLU, Flatten, Dense, Dropout, Input, BatchNormalization, Concatenate\n",
        "from tensorflow.keras.optimizers import Adam\n",
        "from tensorflow.keras.callbacks import EarlyStopping, ReduceLROnPlateau\n",
        "\n",
        "num_classes = 3\n",
        "\n",
        "# Definir la LSTM\n",
        "def lstm_function(input_layer):\n",
        "    # Reshape para que sea compatible con LSTM\n",
        "    lstm_input = Reshape((window_size, 1))(input_layer)\n",
        "\n",
        "    # Capa LSTM con 50 unidades\n",
        "    lstm_output = LSTM(50, return_sequences=False)(lstm_input)\n",
        "    return lstm_output\n",
        "\n",
        "def fc(enco):\n",
        "    # Capa Nro-1\n",
        "   # x = Conv1D(32, kernel_size=3, activation='linear', padding='valid')(enco)\n",
        "\n",
        "    x = enco\n",
        "    # Capa Nro-7\n",
        "    x = Flatten()(x)\n",
        "\n",
        "\n",
        "    # Capa Nro-9\n",
        "    #x = Dense(num_classes, activation='softmax')(x)\n",
        "    return x\n",
        "\n",
        "encode = tf.keras.models.Model(inputs=autoencoder.input,\n",
        "                               outputs=autoencoder.get_layer('bottleneck').output)\n",
        "\n",
        "# Pasar la misma entrada por la red CNN simple\n",
        "cnn_output = cnn_simple(entrada)\n",
        "\n",
        "lstm_output = lstm_function(entrada)\n",
        "\n",
        "\n",
        "# Concatenar las salidas del encoder y de la CNN\n",
        "concatenated_output = Concatenate()([fc(encode.output), cnn_output, lstm_output])\n",
        "\n",
        "#full_model = tf.keras.models.Model(entrada, fc(encode.output))\n",
        "\n",
        "# Capa densa con activación softmax para clasificación en 3 clases\n",
        "output_layer = Dense(3, activation='softmax')(concatenated_output)\n",
        "\n",
        "# Crear el modelo completo\n",
        "full_model = Model(inputs=entrada, outputs=output_layer)\n",
        "encode = tf.keras.models.Model(entrada, encode.output)\n",
        "\n",
        "# Freeze the layers in the encoder model\n",
        "for layer in encode.layers:\n",
        "    layer.trainable = False\n",
        "\n",
        "optimizer = tf.keras.optimizers.Adam(learning_rate=0.00001)\n",
        "\n",
        "full_model.compile(optimizer=optimizer, loss='categorical_crossentropy', metrics=['accuracy'])\n",
        "\n",
        "X_train_encoded = full_model.predict(X_train_noisy)\n",
        "X_val_encoded = full_model.predict(X_val_noisy)\n",
        "\n",
        "reduce_lr = tf.keras.callbacks.ReduceLROnPlateau(\n",
        "    monitor='val_loss',\n",
        "    factor=0.5,  # Reduce la tasa de aprendizaje a la mitad\n",
        "    patience=5,  # Espera 5 épocas antes de reducir la tasa de aprendizaje\n",
        "    min_lr=1e-6  # No reduce la tasa de aprendizaje por debajo de este valor\n",
        ")\n",
        "\n",
        "\n",
        "# Add Early Stopping\n",
        "early_stopping = tf.keras.callbacks.EarlyStopping(\n",
        "    monitor='val_loss',  # The performance metric to monitor\n",
        "    patience=10,         # Number of epochs with no improvement after which training will be stopped\n",
        "    restore_best_weights=True  # Whether to restore model weights from the epoch with the best value of the monitored quantity\n",
        ")\n",
        "\n",
        "batch_size = 16  # O prueba con 16, 128, etc.\n",
        "\n",
        "# Train the classifier with early stopping\n",
        "classifier_history = full_model.fit(\n",
        "    X_train,\n",
        "    y_train,\n",
        "    validation_data=(X_val, y_val),\n",
        "    epochs=120,\n",
        "    batch_size=batch_size,\n",
        "    callbacks=[early_stopping, reduce_lr]  # Añadir ReduceLROnPlateau\n",
        ")\n",
        "\n",
        "# Evaluate the model\n",
        "loss, accuracy = full_model.evaluate(X_test, y_test)\n",
        "print(f'Loss: {loss}, Accuracy: {accuracy}')\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ScGraPTdYViK",
        "outputId": "9a3b71ab-81d0-42e9-92e2-3a7bc959b66a"
      },
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\u001b[1m375/375\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m2s\u001b[0m 3ms/step\n",
            "\u001b[1m47/47\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 7ms/step\n",
            "Epoch 1/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m10s\u001b[0m 8ms/step - accuracy: 0.3433 - loss: 1.0964 - val_accuracy: 0.3767 - val_loss: 1.0407 - learning_rate: 1.0000e-05\n",
            "Epoch 2/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 7ms/step - accuracy: 0.3596 - loss: 1.0216 - val_accuracy: 0.5880 - val_loss: 0.9691 - learning_rate: 1.0000e-05\n",
            "Epoch 3/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.5693 - loss: 0.9083 - val_accuracy: 0.6287 - val_loss: 0.8004 - learning_rate: 1.0000e-05\n",
            "Epoch 4/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 7ms/step - accuracy: 0.6243 - loss: 0.7769 - val_accuracy: 0.7053 - val_loss: 0.7287 - learning_rate: 1.0000e-05\n",
            "Epoch 5/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.6868 - loss: 0.7074 - val_accuracy: 0.7033 - val_loss: 0.6958 - learning_rate: 1.0000e-05\n",
            "Epoch 6/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.7031 - loss: 0.6752 - val_accuracy: 0.7080 - val_loss: 0.6747 - learning_rate: 1.0000e-05\n",
            "Epoch 7/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.7151 - loss: 0.6531 - val_accuracy: 0.7187 - val_loss: 0.6564 - learning_rate: 1.0000e-05\n",
            "Epoch 8/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.7276 - loss: 0.6364 - val_accuracy: 0.7220 - val_loss: 0.6359 - learning_rate: 1.0000e-05\n",
            "Epoch 9/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.7391 - loss: 0.6222 - val_accuracy: 0.7267 - val_loss: 0.6233 - learning_rate: 1.0000e-05\n",
            "Epoch 10/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.7470 - loss: 0.6089 - val_accuracy: 0.7333 - val_loss: 0.6110 - learning_rate: 1.0000e-05\n",
            "Epoch 11/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.7530 - loss: 0.5962 - val_accuracy: 0.7447 - val_loss: 0.5968 - learning_rate: 1.0000e-05\n",
            "Epoch 12/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.7548 - loss: 0.5864 - val_accuracy: 0.7513 - val_loss: 0.5859 - learning_rate: 1.0000e-05\n",
            "Epoch 13/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.7634 - loss: 0.5752 - val_accuracy: 0.7527 - val_loss: 0.5764 - learning_rate: 1.0000e-05\n",
            "Epoch 14/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.7585 - loss: 0.5657 - val_accuracy: 0.7553 - val_loss: 0.5674 - learning_rate: 1.0000e-05\n",
            "Epoch 15/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.7691 - loss: 0.5556 - val_accuracy: 0.7620 - val_loss: 0.5593 - learning_rate: 1.0000e-05\n",
            "Epoch 16/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.7749 - loss: 0.5471 - val_accuracy: 0.7680 - val_loss: 0.5506 - learning_rate: 1.0000e-05\n",
            "Epoch 17/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.7736 - loss: 0.5420 - val_accuracy: 0.7700 - val_loss: 0.5408 - learning_rate: 1.0000e-05\n",
            "Epoch 18/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.7718 - loss: 0.5347 - val_accuracy: 0.7800 - val_loss: 0.5336 - learning_rate: 1.0000e-05\n",
            "Epoch 19/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.7787 - loss: 0.5244 - val_accuracy: 0.7847 - val_loss: 0.5249 - learning_rate: 1.0000e-05\n",
            "Epoch 20/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.7853 - loss: 0.5158 - val_accuracy: 0.7900 - val_loss: 0.5169 - learning_rate: 1.0000e-05\n",
            "Epoch 21/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.7888 - loss: 0.5074 - val_accuracy: 0.7927 - val_loss: 0.5105 - learning_rate: 1.0000e-05\n",
            "Epoch 22/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.7950 - loss: 0.5004 - val_accuracy: 0.7993 - val_loss: 0.5048 - learning_rate: 1.0000e-05\n",
            "Epoch 23/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8002 - loss: 0.4926 - val_accuracy: 0.7980 - val_loss: 0.4996 - learning_rate: 1.0000e-05\n",
            "Epoch 24/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8045 - loss: 0.4879 - val_accuracy: 0.8027 - val_loss: 0.4949 - learning_rate: 1.0000e-05\n",
            "Epoch 25/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8063 - loss: 0.4814 - val_accuracy: 0.8033 - val_loss: 0.4907 - learning_rate: 1.0000e-05\n",
            "Epoch 26/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8087 - loss: 0.4740 - val_accuracy: 0.8033 - val_loss: 0.4866 - learning_rate: 1.0000e-05\n",
            "Epoch 27/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8118 - loss: 0.4696 - val_accuracy: 0.8067 - val_loss: 0.4828 - learning_rate: 1.0000e-05\n",
            "Epoch 28/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8151 - loss: 0.4638 - val_accuracy: 0.8087 - val_loss: 0.4794 - learning_rate: 1.0000e-05\n",
            "Epoch 29/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8197 - loss: 0.4591 - val_accuracy: 0.8080 - val_loss: 0.4760 - learning_rate: 1.0000e-05\n",
            "Epoch 30/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8165 - loss: 0.4560 - val_accuracy: 0.8140 - val_loss: 0.4727 - learning_rate: 1.0000e-05\n",
            "Epoch 31/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8209 - loss: 0.4511 - val_accuracy: 0.8147 - val_loss: 0.4698 - learning_rate: 1.0000e-05\n",
            "Epoch 32/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8258 - loss: 0.4451 - val_accuracy: 0.8160 - val_loss: 0.4671 - learning_rate: 1.0000e-05\n",
            "Epoch 33/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8258 - loss: 0.4421 - val_accuracy: 0.8173 - val_loss: 0.4642 - learning_rate: 1.0000e-05\n",
            "Epoch 34/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8298 - loss: 0.4374 - val_accuracy: 0.8193 - val_loss: 0.4613 - learning_rate: 1.0000e-05\n",
            "Epoch 35/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8326 - loss: 0.4350 - val_accuracy: 0.8200 - val_loss: 0.4584 - learning_rate: 1.0000e-05\n",
            "Epoch 36/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8334 - loss: 0.4283 - val_accuracy: 0.8233 - val_loss: 0.4555 - learning_rate: 1.0000e-05\n",
            "Epoch 37/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8345 - loss: 0.4283 - val_accuracy: 0.8267 - val_loss: 0.4525 - learning_rate: 1.0000e-05\n",
            "Epoch 38/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8375 - loss: 0.4223 - val_accuracy: 0.8300 - val_loss: 0.4490 - learning_rate: 1.0000e-05\n",
            "Epoch 39/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8365 - loss: 0.4200 - val_accuracy: 0.8333 - val_loss: 0.4463 - learning_rate: 1.0000e-05\n",
            "Epoch 40/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8400 - loss: 0.4133 - val_accuracy: 0.8367 - val_loss: 0.4418 - learning_rate: 1.0000e-05\n",
            "Epoch 41/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8417 - loss: 0.4083 - val_accuracy: 0.8413 - val_loss: 0.4382 - learning_rate: 1.0000e-05\n",
            "Epoch 42/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8483 - loss: 0.4032 - val_accuracy: 0.8447 - val_loss: 0.4323 - learning_rate: 1.0000e-05\n",
            "Epoch 43/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8510 - loss: 0.3955 - val_accuracy: 0.8487 - val_loss: 0.4268 - learning_rate: 1.0000e-05\n",
            "Epoch 44/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8546 - loss: 0.3895 - val_accuracy: 0.8520 - val_loss: 0.4212 - learning_rate: 1.0000e-05\n",
            "Epoch 45/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8597 - loss: 0.3815 - val_accuracy: 0.8573 - val_loss: 0.4163 - learning_rate: 1.0000e-05\n",
            "Epoch 46/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8602 - loss: 0.3796 - val_accuracy: 0.8633 - val_loss: 0.4092 - learning_rate: 1.0000e-05\n",
            "Epoch 47/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8690 - loss: 0.3678 - val_accuracy: 0.8667 - val_loss: 0.4046 - learning_rate: 1.0000e-05\n",
            "Epoch 48/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8701 - loss: 0.3630 - val_accuracy: 0.8687 - val_loss: 0.4015 - learning_rate: 1.0000e-05\n",
            "Epoch 49/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8782 - loss: 0.3531 - val_accuracy: 0.8693 - val_loss: 0.3962 - learning_rate: 1.0000e-05\n",
            "Epoch 50/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8794 - loss: 0.3481 - val_accuracy: 0.8700 - val_loss: 0.3910 - learning_rate: 1.0000e-05\n",
            "Epoch 51/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8811 - loss: 0.3472 - val_accuracy: 0.8733 - val_loss: 0.3886 - learning_rate: 1.0000e-05\n",
            "Epoch 52/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8822 - loss: 0.3435 - val_accuracy: 0.8767 - val_loss: 0.3870 - learning_rate: 1.0000e-05\n",
            "Epoch 53/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8834 - loss: 0.3402 - val_accuracy: 0.8793 - val_loss: 0.3834 - learning_rate: 1.0000e-05\n",
            "Epoch 54/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8883 - loss: 0.3355 - val_accuracy: 0.8813 - val_loss: 0.3804 - learning_rate: 1.0000e-05\n",
            "Epoch 55/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8861 - loss: 0.3331 - val_accuracy: 0.8813 - val_loss: 0.3802 - learning_rate: 1.0000e-05\n",
            "Epoch 56/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8847 - loss: 0.3338 - val_accuracy: 0.8827 - val_loss: 0.3780 - learning_rate: 1.0000e-05\n",
            "Epoch 57/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8862 - loss: 0.3266 - val_accuracy: 0.8847 - val_loss: 0.3753 - learning_rate: 1.0000e-05\n",
            "Epoch 58/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8916 - loss: 0.3262 - val_accuracy: 0.8860 - val_loss: 0.3745 - learning_rate: 1.0000e-05\n",
            "Epoch 59/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8928 - loss: 0.3230 - val_accuracy: 0.8867 - val_loss: 0.3734 - learning_rate: 1.0000e-05\n",
            "Epoch 60/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8906 - loss: 0.3240 - val_accuracy: 0.8867 - val_loss: 0.3712 - learning_rate: 1.0000e-05\n",
            "Epoch 61/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8914 - loss: 0.3235 - val_accuracy: 0.8873 - val_loss: 0.3704 - learning_rate: 1.0000e-05\n",
            "Epoch 62/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8948 - loss: 0.3192 - val_accuracy: 0.8880 - val_loss: 0.3695 - learning_rate: 1.0000e-05\n",
            "Epoch 63/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8938 - loss: 0.3139 - val_accuracy: 0.8887 - val_loss: 0.3686 - learning_rate: 1.0000e-05\n",
            "Epoch 64/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8943 - loss: 0.3183 - val_accuracy: 0.8893 - val_loss: 0.3685 - learning_rate: 1.0000e-05\n",
            "Epoch 65/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8957 - loss: 0.3176 - val_accuracy: 0.8907 - val_loss: 0.3662 - learning_rate: 1.0000e-05\n",
            "Epoch 66/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8969 - loss: 0.3126 - val_accuracy: 0.8913 - val_loss: 0.3656 - learning_rate: 1.0000e-05\n",
            "Epoch 67/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8915 - loss: 0.3130 - val_accuracy: 0.8893 - val_loss: 0.3623 - learning_rate: 1.0000e-05\n",
            "Epoch 68/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8959 - loss: 0.3073 - val_accuracy: 0.8907 - val_loss: 0.3646 - learning_rate: 1.0000e-05\n",
            "Epoch 69/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8955 - loss: 0.3092 - val_accuracy: 0.8947 - val_loss: 0.3641 - learning_rate: 1.0000e-05\n",
            "Epoch 70/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8937 - loss: 0.3097 - val_accuracy: 0.8947 - val_loss: 0.3621 - learning_rate: 1.0000e-05\n",
            "Epoch 71/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8963 - loss: 0.3063 - val_accuracy: 0.8933 - val_loss: 0.3623 - learning_rate: 1.0000e-05\n",
            "Epoch 72/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8974 - loss: 0.3045 - val_accuracy: 0.8947 - val_loss: 0.3583 - learning_rate: 1.0000e-05\n",
            "Epoch 73/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8993 - loss: 0.3032 - val_accuracy: 0.8947 - val_loss: 0.3580 - learning_rate: 1.0000e-05\n",
            "Epoch 74/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9016 - loss: 0.3024 - val_accuracy: 0.8920 - val_loss: 0.3559 - learning_rate: 1.0000e-05\n",
            "Epoch 75/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8978 - loss: 0.3041 - val_accuracy: 0.8967 - val_loss: 0.3585 - learning_rate: 1.0000e-05\n",
            "Epoch 76/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8996 - loss: 0.2995 - val_accuracy: 0.8960 - val_loss: 0.3579 - learning_rate: 1.0000e-05\n",
            "Epoch 77/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8960 - loss: 0.3074 - val_accuracy: 0.8960 - val_loss: 0.3549 - learning_rate: 1.0000e-05\n",
            "Epoch 78/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9005 - loss: 0.2956 - val_accuracy: 0.8973 - val_loss: 0.3566 - learning_rate: 1.0000e-05\n",
            "Epoch 79/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9011 - loss: 0.2994 - val_accuracy: 0.8987 - val_loss: 0.3527 - learning_rate: 1.0000e-05\n",
            "Epoch 80/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9017 - loss: 0.2993 - val_accuracy: 0.8993 - val_loss: 0.3536 - learning_rate: 1.0000e-05\n",
            "Epoch 81/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9022 - loss: 0.2972 - val_accuracy: 0.9000 - val_loss: 0.3523 - learning_rate: 1.0000e-05\n",
            "Epoch 82/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9012 - loss: 0.2974 - val_accuracy: 0.8993 - val_loss: 0.3513 - learning_rate: 1.0000e-05\n",
            "Epoch 83/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8976 - loss: 0.2953 - val_accuracy: 0.8980 - val_loss: 0.3497 - learning_rate: 1.0000e-05\n",
            "Epoch 84/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9007 - loss: 0.2954 - val_accuracy: 0.9007 - val_loss: 0.3511 - learning_rate: 1.0000e-05\n",
            "Epoch 85/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9022 - loss: 0.2949 - val_accuracy: 0.9000 - val_loss: 0.3492 - learning_rate: 1.0000e-05\n",
            "Epoch 86/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9009 - loss: 0.2958 - val_accuracy: 0.9013 - val_loss: 0.3489 - learning_rate: 1.0000e-05\n",
            "Epoch 87/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9017 - loss: 0.2944 - val_accuracy: 0.9033 - val_loss: 0.3493 - learning_rate: 1.0000e-05\n",
            "Epoch 88/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.8987 - loss: 0.2957 - val_accuracy: 0.9040 - val_loss: 0.3464 - learning_rate: 1.0000e-05\n",
            "Epoch 89/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9027 - loss: 0.2891 - val_accuracy: 0.9027 - val_loss: 0.3474 - learning_rate: 1.0000e-05\n",
            "Epoch 90/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9052 - loss: 0.2874 - val_accuracy: 0.9020 - val_loss: 0.3452 - learning_rate: 1.0000e-05\n",
            "Epoch 91/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9018 - loss: 0.2924 - val_accuracy: 0.9040 - val_loss: 0.3464 - learning_rate: 1.0000e-05\n",
            "Epoch 92/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9046 - loss: 0.2893 - val_accuracy: 0.9013 - val_loss: 0.3434 - learning_rate: 1.0000e-05\n",
            "Epoch 93/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9019 - loss: 0.2881 - val_accuracy: 0.9027 - val_loss: 0.3448 - learning_rate: 1.0000e-05\n",
            "Epoch 94/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9030 - loss: 0.2845 - val_accuracy: 0.9000 - val_loss: 0.3429 - learning_rate: 1.0000e-05\n",
            "Epoch 95/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9042 - loss: 0.2860 - val_accuracy: 0.9053 - val_loss: 0.3452 - learning_rate: 1.0000e-05\n",
            "Epoch 96/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9026 - loss: 0.2867 - val_accuracy: 0.9040 - val_loss: 0.3438 - learning_rate: 1.0000e-05\n",
            "Epoch 97/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9051 - loss: 0.2835 - val_accuracy: 0.9000 - val_loss: 0.3415 - learning_rate: 1.0000e-05\n",
            "Epoch 98/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9030 - loss: 0.2850 - val_accuracy: 0.9040 - val_loss: 0.3428 - learning_rate: 1.0000e-05\n",
            "Epoch 99/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9026 - loss: 0.2869 - val_accuracy: 0.9047 - val_loss: 0.3421 - learning_rate: 1.0000e-05\n",
            "Epoch 100/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9080 - loss: 0.2819 - val_accuracy: 0.9047 - val_loss: 0.3421 - learning_rate: 1.0000e-05\n",
            "Epoch 101/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9060 - loss: 0.2840 - val_accuracy: 0.9040 - val_loss: 0.3406 - learning_rate: 1.0000e-05\n",
            "Epoch 102/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9053 - loss: 0.2838 - val_accuracy: 0.9047 - val_loss: 0.3414 - learning_rate: 1.0000e-05\n",
            "Epoch 103/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9074 - loss: 0.2832 - val_accuracy: 0.9020 - val_loss: 0.3414 - learning_rate: 1.0000e-05\n",
            "Epoch 104/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9059 - loss: 0.2844 - val_accuracy: 0.9027 - val_loss: 0.3393 - learning_rate: 1.0000e-05\n",
            "Epoch 105/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9044 - loss: 0.2836 - val_accuracy: 0.9033 - val_loss: 0.3404 - learning_rate: 1.0000e-05\n",
            "Epoch 106/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9045 - loss: 0.2849 - val_accuracy: 0.9047 - val_loss: 0.3380 - learning_rate: 1.0000e-05\n",
            "Epoch 107/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9041 - loss: 0.2825 - val_accuracy: 0.9040 - val_loss: 0.3378 - learning_rate: 1.0000e-05\n",
            "Epoch 108/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9063 - loss: 0.2833 - val_accuracy: 0.9033 - val_loss: 0.3371 - learning_rate: 1.0000e-05\n",
            "Epoch 109/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9064 - loss: 0.2811 - val_accuracy: 0.9033 - val_loss: 0.3367 - learning_rate: 1.0000e-05\n",
            "Epoch 110/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9059 - loss: 0.2813 - val_accuracy: 0.9020 - val_loss: 0.3341 - learning_rate: 1.0000e-05\n",
            "Epoch 111/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9082 - loss: 0.2769 - val_accuracy: 0.9040 - val_loss: 0.3377 - learning_rate: 1.0000e-05\n",
            "Epoch 112/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9079 - loss: 0.2761 - val_accuracy: 0.9040 - val_loss: 0.3353 - learning_rate: 1.0000e-05\n",
            "Epoch 113/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9060 - loss: 0.2798 - val_accuracy: 0.9033 - val_loss: 0.3392 - learning_rate: 1.0000e-05\n",
            "Epoch 114/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9080 - loss: 0.2779 - val_accuracy: 0.9047 - val_loss: 0.3386 - learning_rate: 1.0000e-05\n",
            "Epoch 115/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9051 - loss: 0.2797 - val_accuracy: 0.9020 - val_loss: 0.3327 - learning_rate: 1.0000e-05\n",
            "Epoch 116/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9078 - loss: 0.2790 - val_accuracy: 0.9033 - val_loss: 0.3352 - learning_rate: 1.0000e-05\n",
            "Epoch 117/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9097 - loss: 0.2796 - val_accuracy: 0.9027 - val_loss: 0.3357 - learning_rate: 1.0000e-05\n",
            "Epoch 118/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9068 - loss: 0.2782 - val_accuracy: 0.9027 - val_loss: 0.3363 - learning_rate: 1.0000e-05\n",
            "Epoch 119/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9071 - loss: 0.2778 - val_accuracy: 0.9027 - val_loss: 0.3349 - learning_rate: 1.0000e-05\n",
            "Epoch 120/120\n",
            "\u001b[1m750/750\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m6s\u001b[0m 8ms/step - accuracy: 0.9077 - loss: 0.2768 - val_accuracy: 0.9047 - val_loss: 0.3366 - learning_rate: 1.0000e-05\n",
            "\u001b[1m47/47\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step - accuracy: 0.9251 - loss: 0.2672\n",
            "Loss: 0.28628766536712646, Accuracy: 0.9173333048820496\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "\n",
        "# Balanceo de clases"
      ],
      "metadata": {
        "id": "3u8F0OR9hqFE"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "import numpy as np\n",
        "import tensorflow as tf\n",
        "from tensorflow.keras.models import Model\n",
        "from tensorflow.keras.layers import Conv1D, MaxPooling1D, LeakyReLU, Flatten, Dense, Dropout, Input\n",
        "from tensorflow.keras.optimizers import Adam\n",
        "from tensorflow.keras.callbacks import EarlyStopping, ReduceLROnPlateau\n",
        "from sklearn.utils import resample\n",
        "from imblearn.over_sampling import SMOTE\n",
        "\n",
        "# Balancear las clases utilizando SMOTE\n",
        "def balance_classes(X, y):\n",
        "    smote = SMOTE()\n",
        "    X_resampled, y_resampled = smote.fit_resample(X, y)\n",
        "    return X_resampled, y_resampled\n",
        "\n",
        "# Crear un modelo de clasificador mejorado\n",
        "num_classes = 3\n",
        "\n",
        "def create_classifier(inputs):\n",
        "\n",
        "    x = Conv1D(128, kernel_size=3, activation='linear', padding='same')(inputs)\n",
        "    x = LeakyReLU(alpha=0.01)(x)\n",
        "    x = MaxPooling1D(pool_size=2, strides=2, padding='same')(x)\n",
        "    x = Dropout(0.5)(x)\n",
        "    x = Flatten()(x)\n",
        "    x = Dense(100, activation='linear')(x)\n",
        "    x = LeakyReLU(alpha=0.01)(x)\n",
        "    x = Dropout(0.5)(x)\n",
        "    x = Dense(50, activation='linear')(x)\n",
        "    x = LeakyReLU(alpha=0.01)(x)\n",
        "    x = Dropout(0.5)(x)\n",
        "    outputs = Dense(num_classes, activation='softmax')(x)\n",
        "    return outputs\n",
        "\n",
        "encode = tf.keras.models.Model(inputs=autoencoder.input,\n",
        "                               outputs=autoencoder.get_layer('bottleneck').output)\n",
        "\n",
        "classifier = tf.keras.models.Model(entrada, create_classifier(encode.output))\n",
        "encode = tf.keras.models.Model(entrada, encode.output)\n",
        "\n",
        "# Freeze the layers in the encoder model\n",
        "for layer in encode.layers:\n",
        "    layer.trainable = False\n",
        "\n",
        "optimizer = tf.keras.optimizers.Adam(learning_rate=0.00001)\n",
        "\n",
        "classifier.compile(optimizer=optimizer, loss='categorical_crossentropy', metrics=['accuracy'])\n",
        "\n",
        "# Asume que `X_train`, `y_train`, `X_val`, `y_val`, `X_test` y `y_test` ya están definidos\n",
        "# Sobremuestrear la clase INTERICTAL utilizando SMOTE\n",
        "X_train_resampled, y_train_resampled = balance_classes(X_train, y_train)\n",
        "\n",
        "# Callbacks\n",
        "early_stopping = EarlyStopping(\n",
        "    monitor='val_loss',  # The performance metric to monitor\n",
        "    patience=10,         # Number of epochs with no improvement after which training will be stopped\n",
        "    restore_best_weights=True  # Whether to restore model weights from the epoch with the best value of the monitored quantity\n",
        ")\n",
        "\n",
        "reduce_lr = ReduceLROnPlateau(\n",
        "    monitor='val_loss',\n",
        "    factor=0.5,\n",
        "    patience=5,\n",
        "    min_lr=1e-6\n",
        ")\n",
        "\n",
        "# Train the classifier with early stopping and learning rate reduction\n",
        "classifier_history = classifier.fit(\n",
        "    X_train_resampled,\n",
        "    y_train_resampled,\n",
        "    validation_data=(X_val, y_val),\n",
        "    epochs=50,\n",
        "    batch_size=32,\n",
        "    callbacks=[early_stopping]  # Add early stopping and learning rate reduction\n",
        ")\n",
        "\n",
        "# Evaluate the model\n",
        "loss, accuracy = classifier.evaluate(X_test, y_test)\n",
        "print(f'Loss: {loss}, Accuracy: {accuracy}')\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "5SE8y-Qahwsq",
        "outputId": "452e7498-9ba5-403a-9151-9886126c7a9e"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.10/dist-packages/keras/src/layers/activations/leaky_relu.py:41: UserWarning: Argument `alpha` is deprecated. Use `negative_slope` instead.\n",
            "  warnings.warn(\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m10s\u001b[0m 13ms/step - accuracy: 0.4057 - loss: 1.0978 - val_accuracy: 0.5693 - val_loss: 1.0950\n",
            "Epoch 2/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.5170 - loss: 1.0921 - val_accuracy: 0.5680 - val_loss: 1.0869\n",
            "Epoch 3/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.5236 - loss: 1.0802 - val_accuracy: 0.5627 - val_loss: 1.0713\n",
            "Epoch 4/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.5264 - loss: 1.0583 - val_accuracy: 0.5680 - val_loss: 1.0446\n",
            "Epoch 5/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.5531 - loss: 1.0240 - val_accuracy: 0.5707 - val_loss: 1.0070\n",
            "Epoch 6/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.5673 - loss: 0.9799 - val_accuracy: 0.5700 - val_loss: 0.9646\n",
            "Epoch 7/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.5812 - loss: 0.9353 - val_accuracy: 0.5747 - val_loss: 0.9208\n",
            "Epoch 8/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.5857 - loss: 0.8927 - val_accuracy: 0.6493 - val_loss: 0.8829\n",
            "Epoch 9/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.5936 - loss: 0.8546 - val_accuracy: 0.6040 - val_loss: 0.8486\n",
            "Epoch 10/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.5950 - loss: 0.8238 - val_accuracy: 0.5913 - val_loss: 0.8195\n",
            "Epoch 11/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.5968 - loss: 0.7993 - val_accuracy: 0.5907 - val_loss: 0.7989\n",
            "Epoch 12/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6072 - loss: 0.7717 - val_accuracy: 0.5840 - val_loss: 0.7797\n",
            "Epoch 13/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6079 - loss: 0.7571 - val_accuracy: 0.5720 - val_loss: 0.7677\n",
            "Epoch 14/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6098 - loss: 0.7408 - val_accuracy: 0.5793 - val_loss: 0.7573\n",
            "Epoch 15/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6105 - loss: 0.7310 - val_accuracy: 0.5807 - val_loss: 0.7508\n",
            "Epoch 16/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6152 - loss: 0.7233 - val_accuracy: 0.5700 - val_loss: 0.7446\n",
            "Epoch 17/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6108 - loss: 0.7156 - val_accuracy: 0.5687 - val_loss: 0.7394\n",
            "Epoch 18/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6170 - loss: 0.7105 - val_accuracy: 0.5693 - val_loss: 0.7353\n",
            "Epoch 19/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6184 - loss: 0.7060 - val_accuracy: 0.5700 - val_loss: 0.7335\n",
            "Epoch 20/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6138 - loss: 0.6978 - val_accuracy: 0.5687 - val_loss: 0.7304\n",
            "Epoch 21/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6104 - loss: 0.6964 - val_accuracy: 0.5707 - val_loss: 0.7302\n",
            "Epoch 22/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6170 - loss: 0.6938 - val_accuracy: 0.5693 - val_loss: 0.7284\n",
            "Epoch 23/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6181 - loss: 0.6895 - val_accuracy: 0.5700 - val_loss: 0.7262\n",
            "Epoch 24/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6213 - loss: 0.6878 - val_accuracy: 0.5680 - val_loss: 0.7242\n",
            "Epoch 25/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6246 - loss: 0.6828 - val_accuracy: 0.5700 - val_loss: 0.7244\n",
            "Epoch 26/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6171 - loss: 0.6860 - val_accuracy: 0.5680 - val_loss: 0.7226\n",
            "Epoch 27/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6255 - loss: 0.6775 - val_accuracy: 0.5693 - val_loss: 0.7224\n",
            "Epoch 28/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6229 - loss: 0.6795 - val_accuracy: 0.5680 - val_loss: 0.7212\n",
            "Epoch 29/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6297 - loss: 0.6779 - val_accuracy: 0.5687 - val_loss: 0.7207\n",
            "Epoch 30/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6303 - loss: 0.6705 - val_accuracy: 0.5660 - val_loss: 0.7195\n",
            "Epoch 31/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6253 - loss: 0.6725 - val_accuracy: 0.5647 - val_loss: 0.7185\n",
            "Epoch 32/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6265 - loss: 0.6732 - val_accuracy: 0.5653 - val_loss: 0.7180\n",
            "Epoch 33/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6231 - loss: 0.6726 - val_accuracy: 0.5667 - val_loss: 0.7173\n",
            "Epoch 34/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6257 - loss: 0.6678 - val_accuracy: 0.5660 - val_loss: 0.7170\n",
            "Epoch 35/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6214 - loss: 0.6727 - val_accuracy: 0.5667 - val_loss: 0.7171\n",
            "Epoch 36/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6394 - loss: 0.6628 - val_accuracy: 0.5640 - val_loss: 0.7158\n",
            "Epoch 37/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6236 - loss: 0.6706 - val_accuracy: 0.5653 - val_loss: 0.7140\n",
            "Epoch 38/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6291 - loss: 0.6643 - val_accuracy: 0.5667 - val_loss: 0.7147\n",
            "Epoch 39/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6394 - loss: 0.6617 - val_accuracy: 0.5693 - val_loss: 0.7136\n",
            "Epoch 40/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6281 - loss: 0.6631 - val_accuracy: 0.5640 - val_loss: 0.7129\n",
            "Epoch 41/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6349 - loss: 0.6576 - val_accuracy: 0.5667 - val_loss: 0.7118\n",
            "Epoch 42/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6255 - loss: 0.6639 - val_accuracy: 0.5707 - val_loss: 0.7148\n",
            "Epoch 43/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6282 - loss: 0.6613 - val_accuracy: 0.5673 - val_loss: 0.7119\n",
            "Epoch 44/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6282 - loss: 0.6588 - val_accuracy: 0.5667 - val_loss: 0.7095\n",
            "Epoch 45/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6242 - loss: 0.6612 - val_accuracy: 0.5700 - val_loss: 0.7109\n",
            "Epoch 46/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6381 - loss: 0.6563 - val_accuracy: 0.5707 - val_loss: 0.7103\n",
            "Epoch 47/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6318 - loss: 0.6546 - val_accuracy: 0.5713 - val_loss: 0.7101\n",
            "Epoch 48/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6340 - loss: 0.6492 - val_accuracy: 0.5793 - val_loss: 0.7101\n",
            "Epoch 49/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6338 - loss: 0.6563 - val_accuracy: 0.5680 - val_loss: 0.7069\n",
            "Epoch 50/50\n",
            "\u001b[1m453/453\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.6320 - loss: 0.6556 - val_accuracy: 0.5733 - val_loss: 0.7082\n",
            "\u001b[1m47/47\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.5924 - loss: 0.6568\n",
            "Loss: 0.6817027926445007, Accuracy: 0.5873333215713501\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Grafica de perdida y precision del clasificador"
      ],
      "metadata": {
        "id": "dUfsGxQLvduT"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "import matplotlib.pyplot as plt\n",
        "\n",
        "# Plotting validation accuracy and validation loss on the same graph\n",
        "plt.figure(figsize=(10, 5))\n",
        "epochs = range(1, len(classifier_history.history['val_loss']) + 1)\n",
        "\n",
        "plt.plot(epochs, classifier_history.history['loss'], label='Train')\n",
        "plt.plot(epochs, classifier_history.history['val_loss'], label='Validation')\n",
        "plt.title('Validation Loss')\n",
        "plt.xlabel('Epochs')\n",
        "plt.legend()\n",
        "\n",
        "plt.show()\n",
        "\n",
        "# Plotting training accuracy and training loss on the same graph\n",
        "plt.figure(figsize=(10, 5))\n",
        "\n",
        "plt.plot(epochs, classifier_history.history['accuracy'], label='Training')\n",
        "plt.plot(epochs, classifier_history.history['val_accuracy'], label='Validation')\n",
        "plt.title('Training Accuracy')\n",
        "plt.xlabel('Epochs')\n",
        "plt.legend()\n",
        "\n",
        "plt.show()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 957
        },
        "id": "bDLjptgUva4A",
        "outputId": "1e8cf42e-56b0-4f4b-b591-e306cbb952b8"
      },
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 1000x500 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 1000x500 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Metricas"
      ],
      "metadata": {
        "id": "l5LlXQHzvmaL"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.metrics import confusion_matrix, classification_report\n",
        "import seaborn as sns\n",
        "import pandas as pd\n",
        "n_classes = 3\n",
        "\n",
        "label_names = ['NORMAL', 'INTERICTAL', 'ICTAL']\n",
        "#encode = tf.keras.models.Model(entrada, encode)\n",
        "\n",
        "#encoded_test = encode.predict(X_test)\n",
        "\n",
        "# Get the predictions\n",
        "\n",
        "y_pred_prob = full_model.predict(X_test)  # Realizar las predicciones\n",
        "y_pred_classes = np.argmax(y_pred_prob, axis=1)  # Obtener la clase predicha\n",
        "y_true = np.argmax(y_test, axis=1)  # Convertir las etiquetas verdaderas a clases\n",
        "\n",
        "# Calculate sensitivity, specificity, and F1 score\n",
        "# Function to calculate specificity\n",
        "def calculate_specificity(y_true, y_pred, labels):\n",
        "    cm = confusion_matrix(y_true, y_pred, labels=labels)\n",
        "    specificity = []\n",
        "    for i, label in enumerate(labels):\n",
        "        true_negatives = np.delete(np.delete(cm, i, axis=0), i, axis=1).sum()\n",
        "        false_positives = np.delete(cm[i, :], i).sum()\n",
        "        specificity.append(true_negatives / (true_negatives + false_positives))\n",
        "    return specificity\n",
        "\n",
        "# Get the predictions\n",
        "y_pred = np.argmax(y_pred_prob, axis=1)\n",
        "y_true = np.argmax(y_test, axis=1)\n",
        "\n",
        "# Calculate confusion matrix\n",
        "cm = confusion_matrix(y_true, y_pred)\n",
        "\n",
        "# Calculate sensitivity, specificity, and F1 score\n",
        "report = classification_report(y_true, y_pred, output_dict=True)\n",
        "sensitivity = [report[str(i)]['recall'] for i in range(n_classes)]\n",
        "f1_scores = [report[str(i)]['f1-score'] for i in range(n_classes)]\n",
        "specificity = calculate_specificity(y_true, y_pred, labels=[0, 1, 2])\n",
        "\n",
        "# Calculate precision for each class\n",
        "precision = []\n",
        "for i, label in enumerate(label_names):\n",
        "    tp = cm[i, i]  # True positives\n",
        "    fp = np.delete(cm[i, :], i).sum()  # False positives\n",
        "    precision.append(tp / (tp + fp))\n",
        "\n",
        "# Plot confusion matrix for each class\n",
        "plt.figure(figsize=(8, 6))\n",
        "sns.heatmap(cm, annot=True, fmt='d', cmap='Blues', xticklabels=label_names, yticklabels=label_names)\n",
        "plt.title('Confusion Matrix')\n",
        "plt.xlabel('Predicted Labels')\n",
        "plt.ylabel('True Labels')\n",
        "plt.show()\n",
        "\n",
        "# Create a dataframe for sensitivity, specificity, and F1 score\n",
        "metrics_df = pd.DataFrame({\n",
        "    'Class': label_names,\n",
        "    'Precision': precision,\n",
        "    'Sensitivity': sensitivity,\n",
        "    'Specificity': specificity,\n",
        "    'F1 Score': f1_scores\n",
        "})\n",
        "\n",
        "# Display the dataframe as a table\n",
        "print(metrics_df)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 651
        },
        "id": "g16c8AmIviBR",
        "outputId": "fd4ef421-fa74-49cb-a9ee-59aa606a0450"
      },
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\u001b[1m47/47\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 4ms/step\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 800x600 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "        Class  Precision  Sensitivity  Specificity  F1 Score\n",
            "0      NORMAL   0.946429     0.946429     0.961494  0.926132\n",
            "1  INTERICTAL   0.887395     0.887395     0.927568  0.902564\n",
            "2       ICTAL   0.916955     0.916955     0.980296  0.928196\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "#Resultados\n",
        "\n",
        "##con lr 0.0001 y el decoder con funcion de activacion linear:\n",
        "\n",
        "\n",
        "\n",
        "```\n",
        "Class  Precision  Sensitivity  Specificity  F1 Score\n",
        "0      NORMAL   0.969156     0.969156     0.977778  0.946868\n",
        "1  INTERICTAL   0.910924     0.910924     0.942949  0.929674\n",
        "2       ICTAL   0.955017     0.955017     0.989309  0.963351\n",
        "```\n",
        "\n",
        "## **Balanceo de Clases y Aumento de Datos** con funcion de activacion linear:\n",
        "\n",
        "```\n",
        " Class  Precision  Sensitivity  Specificity  F1 Score\n",
        "0      NORMAL   0.990260     0.990260     0.992481  0.925645\n",
        "1  INTERICTAL   0.818487     0.818487     0.892644  0.894399\n",
        "2       ICTAL   0.975779     0.975779     0.994147  0.951096\n",
        "```\n",
        "\n",
        "## **Balanceo de Clases y Aumento de Datos** con funcion de activacion relu menos la ultima:\n",
        "\n",
        "\n",
        "\n",
        "```\n",
        "        Class  Precision  Sensitivity  Specificity  F1 Score\n",
        "0      NORMAL   0.980519     0.980519     0.985330  0.930663\n",
        "1  INTERICTAL   0.842017     0.842017     0.904954  0.905967\n",
        "2       ICTAL   0.982699     0.982699     0.995809  0.953020\n",
        "```\n",
        "## con mejora para la clase 1\n",
        "\n",
        "```\n",
        "Class  Precision  Sensitivity  Specificity  F1 Score\n",
        "0      NORMAL   0.961039     0.961039     0.972093  0.942675\n",
        "1  INTERICTAL   0.884034     0.884034     0.926907  0.913988\n",
        "2       ICTAL   0.968858     0.968858     0.992475  0.944351\n",
        "```\n",
        "\n",
        "\n",
        "\n",
        "\n"
      ],
      "metadata": {
        "id": "nxpY_3gyjUij"
      }
    }
  ]
}