instrument_classifier.ipynb

{
 "nbformat": 4,
 "nbformat_minor": 0,
 "metadata": {
  "colab": {
   "name": "instrument_classifier.ipynb",
   "provenance": [],
   "collapsed_sections": [
    "QQHIifuczbmZ",
    "YZ_6RL6azhrb"
   ],
   "machine_shape": "hm"
  },
  "kernelspec": {
   "name": "python3",
   "display_name": "Python 3"
  },
  "language_info": {
   "name": "python"
  },
  "accelerator": "GPU"
 },
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "l1HlSlfepr24",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "import ntpath\n",
    "import os\n",
    "import time\n",
    "\n",
    "import librosa\n",
    "import librosa.display\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import tensorflow as tf\n",
    "from keras import layers\n",
    "from tensorflow import keras"
   ]
  },
  {
   "cell_type": "markdown",
   "source": [
    "# Utility functions"
   ],
   "metadata": {
    "id": "sIGsUNqTzd9a",
    "pycharm": {
     "name": "#%% md\n"
    }
   }
  },
  {
   "cell_type": "code",
   "source": [
    "def convert_labels(labels):\n",
    "    instruments = ['cel', 'cla', 'flu', 'gac', 'gel', 'org', 'pia', 'sax', 'tru', 'vio', 'voi']\n",
    "    for index, instrument in enumerate(instruments):\n",
    "        labels = [index if label == instrument else label for label in labels]\n",
    "    return labels\n",
    "\n",
    "def reconvert_labels(labels):\n",
    "    instruments = ['cel', 'cla', 'flu', 'gac', 'gel', 'org', 'pia', 'sax', 'tru', 'vio', 'voi']\n",
    "    return np.asarray([instruments[label] for label in labels])\n",
    "\n",
    "def time_passed(t):\n",
    "    now = time.time()\n",
    "    print('Time passed: {} seconds.'.format(now - t))\n",
    "    return now"
   ],
   "metadata": {
    "id": "QU97IKDFxnv7",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "markdown",
   "source": [
    "# Read data"
   ],
   "metadata": {
    "id": "14ZsNSIg0AJO",
    "pycharm": {
     "name": "#%% md\n"
    }
   }
  },
  {
   "cell_type": "code",
   "source": [
    "t = time.time()\n",
    "\n",
    "X_train = np.load('./dataset/X_train_data.npy', allow_pickle=True)\n",
    "y_train = np.load('./dataset/y_train_data.npy', allow_pickle=True)\n",
    "print(y_train.shape[0])\n",
    "t = time_passed(t)\n",
    "print(X_train.shape)"
   ],
   "metadata": {
    "id": "MZTepoS8tCFH",
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "outputId": "9a829177-b22f-4d54-a09e-bfc1fc791dfb",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": [
      "20115\n",
      "Time passed: 7.749913215637207 seconds.\n",
      "(20115, 128, 87)\n"
     ]
    }
   ]
  },
  {
   "cell_type": "code",
   "source": [
    "X_test = np.load('./dataset/X_test_data.npy', allow_pickle=True)\n",
    "y_test = np.load('./dataset/y_test_data.npy', allow_pickle=True)\n",
    "print(y_test.shape[0])\n",
    "t = time_passed(t)\n",
    "print(X_test.shape)"
   ],
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "3zw2XOHUowmB",
    "outputId": "adc962a1-3d6d-4b1e-951a-983c96bed4fa",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": [
      "45661\n",
      "Time passed: 17.201735734939575 seconds.\n",
      "(45661, 128, 87)\n"
     ]
    }
   ]
  },
  {
   "cell_type": "code",
   "source": [
    "X_train = np.reshape(X_train, [-1, 128, 87, 1])\n",
    "X_test = np.reshape(X_test, [-1, 128, 87, 1])\n",
    "t = time_passed(t)"
   ],
   "metadata": {
    "id": "_kp7qhz_2gr9",
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "outputId": "530113ba-b359-4abd-be18-911f740aaea3",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": [
      "Time passed: 0.008810997009277344 seconds.\n"
     ]
    }
   ]
  },
  {
   "cell_type": "code",
   "source": [
    "X = np.concatenate((X_train, X_test), axis=0)\n",
    "y = np.concatenate((y_train, y_test), axis=0)"
   ],
   "metadata": {
    "id": "cT8uV37ZaWaE",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "source": [
    "print(X.shape)\n",
    "y.shape"
   ],
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "7Qos7G34bFGi",
    "outputId": "d8593940-29ad-4e60-94ca-68004d939d71",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": [
      "(65776, 128, 87, 1)\n"
     ]
    },
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": [
       "(65776, 11)"
      ]
     },
     "metadata": {},
     "execution_count": 8
    }
   ]
  },
  {
   "cell_type": "code",
   "source": [
    "randomize = np.arange(len(y))\n",
    "\n",
    "np.random.shuffle(randomize)\n",
    "\n",
    "X = X[randomize]\n",
    "y = y[randomize]"
   ],
   "metadata": {
    "id": "28HcXvH9rZhw",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "source": [
    "X_train = X[:30000]\n",
    "y_train = y[:30000]\n",
    "X_test = X[30000:]\n",
    "y_test = y[30000:]"
   ],
   "metadata": {
    "id": "KFlV4aKxAgVR",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "source": [
    "y_train_classes = np.argmax(y_train, axis=-1)\n",
    "y_train_classes = reconvert_labels(y_train_classes)\n",
    "unique, counts = np.unique(y_train_classes, return_counts=True)\n",
    "print(np.asarray((unique, counts)).T)"
   ],
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "f9vOJpwWid3W",
    "outputId": "31320a9a-5b12-4e8b-bec2-f05886cf1067",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": [
      "[['cel' '1360']\n",
      " ['cla' '1141']\n",
      " ['flu' '1672']\n",
      " ['gac' '4445']\n",
      " ['gel' '6907']\n",
      " ['org' '1741']\n",
      " ['pia' '6750']\n",
      " ['sax' '942']\n",
      " ['tru' '880']\n",
      " ['vio' '1223']\n",
      " ['voi' '2939']]\n"
     ]
    }
   ]
  },
  {
   "cell_type": "code",
   "source": [
    "y_test_classes = np.argmax(y_test, axis=-1)\n",
    "y_test_classes = reconvert_labels(y_test_classes)\n",
    "unique, counts = np.unique(y_test_classes, return_counts=True)\n",
    "print(np.asarray((unique, counts)).T)"
   ],
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "E4BEFQz6k47N",
    "outputId": "38244433-aae4-47f3-d021-76527ee17a27",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": [
      "[['cel' '1648']\n",
      " ['cla' '1319']\n",
      " ['flu' '2023']\n",
      " ['gac' '5272']\n",
      " ['gel' '8183']\n",
      " ['org' '2212']\n",
      " ['pia' '8131']\n",
      " ['sax' '1063']\n",
      " ['tru' '1031']\n",
      " ['vio' '1406']\n",
      " ['voi' '3488']]\n"
     ]
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": [
    "# CNN Network"
   ],
   "metadata": {
    "id": "YZ_6RL6azhrb",
    "pycharm": {
     "name": "#%% md\n"
    }
   }
  },
  {
   "cell_type": "code",
   "source": [
    "def create_model(input_shape, output_shape):\n",
    "    model = keras.Sequential()\n",
    "  \n",
    "    model.add(layers.Conv2D(32, kernel_size=(3, 3), activation='relu', padding='same', input_shape=input_shape))\n",
    "    model.add(layers.MaxPooling2D(pool_size=(3, 3)))\n",
    "    model.add(layers.Dropout(0.25))\n",
    "\n",
    "    model.add(layers.Conv2D(64, kernel_size=(3, 3), activation='relu', padding='same'))\n",
    "    model.add(layers.MaxPooling2D(pool_size=(3, 3)))\n",
    "    model.add(layers.Dropout(0.25))\n",
    "\n",
    "    model.add(layers.Conv2D(128, kernel_size=(3, 3), activation='relu', padding='same'))\n",
    "    model.add(layers.MaxPooling2D(pool_size=(3, 3)))\n",
    "    model.add(layers.Dropout(0.25))\n",
    "\n",
    "    model.add(layers.Conv2D(256, kernel_size=(3, 3), activation='relu', padding='same'))\n",
    "    model.add(layers.MaxPooling2D(pool_size=(3, 3)))\n",
    "    model.add(layers.Dropout(0.25))\n",
    "\n",
    "    model.add(layers.Flatten())\n",
    "    model.add(layers.Dense(512, activation='relu'))\n",
    "    model.add(layers.Dropout(0.25))\n",
    "    model.add(layers.Dense(512, activation='relu'))\n",
    "    model.add(layers.Dropout(0.5))\n",
    "    model.add(layers.Dense(output_shape, activation='softmax'))\n",
    "\n",
    "    return model"
   ],
   "metadata": {
    "id": "mzUOtFnjxlRX",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "source": [
    "model = create_model((128, 87, 1), 11)\n",
    "model.summary()\n",
    "t = time_passed(t)"
   ],
   "metadata": {
    "id": "XbIrl_u12krq",
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "outputId": "6be38ce7-d663-4bb5-de8e-b76fdddf2461",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " conv2d (Conv2D)             (None, 128, 87, 32)       320       \n",
      "                                                                 \n",
      " max_pooling2d (MaxPooling2D  (None, 42, 29, 32)       0         \n",
      " )                                                               \n",
      "                                                                 \n",
      " dropout (Dropout)           (None, 42, 29, 32)        0         \n",
      "                                                                 \n",
      " conv2d_1 (Conv2D)           (None, 42, 29, 64)        18496     \n",
      "                                                                 \n",
      " max_pooling2d_1 (MaxPooling  (None, 14, 9, 64)        0         \n",
      " 2D)                                                             \n",
      "                                                                 \n",
      " dropout_1 (Dropout)         (None, 14, 9, 64)         0         \n",
      "                                                                 \n",
      " conv2d_2 (Conv2D)           (None, 14, 9, 128)        73856     \n",
      "                                                                 \n",
      " max_pooling2d_2 (MaxPooling  (None, 4, 3, 128)        0         \n",
      " 2D)                                                             \n",
      "                                                                 \n",
      " dropout_2 (Dropout)         (None, 4, 3, 128)         0         \n",
      "                                                                 \n",
      " conv2d_3 (Conv2D)           (None, 4, 3, 256)         295168    \n",
      "                                                                 \n",
      " max_pooling2d_3 (MaxPooling  (None, 1, 1, 256)        0         \n",
      " 2D)                                                             \n",
      "                                                                 \n",
      " dropout_3 (Dropout)         (None, 1, 1, 256)         0         \n",
      "                                                                 \n",
      " flatten (Flatten)           (None, 256)               0         \n",
      "                                                                 \n",
      " dense (Dense)               (None, 512)               131584    \n",
      "                                                                 \n",
      " dropout_4 (Dropout)         (None, 512)               0         \n",
      "                                                                 \n",
      " dense_1 (Dense)             (None, 512)               262656    \n",
      "                                                                 \n",
      " dropout_5 (Dropout)         (None, 512)               0         \n",
      "                                                                 \n",
      " dense_2 (Dense)             (None, 11)                5643      \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 787,723\n",
      "Trainable params: 787,723\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "Time passed: 6.729798078536987 seconds.\n"
     ]
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": [
    "# Train model"
   ],
   "metadata": {
    "id": "YWT3n009yEas",
    "pycharm": {
     "name": "#%% md\n"
    }
   }
  },
  {
   "cell_type": "code",
   "source": [
    "opt = keras.optimizers.Adam(learning_rate=0.0001, beta_1=0.9, beta_2=0.999, epsilon=None, decay=0.0, amsgrad=False)\n",
    "\n",
    "model.compile(loss='categorical_crossentropy',\n",
    "              optimizer=opt,\n",
    "              metrics=['accuracy'])\n",
    "\n",
    "history = model.fit(X_train, y_train,\n",
    "                    epochs=200,\n",
    "                    validation_data=(X_test, y_test))\n",
    "t = time_passed(t)"
   ],
   "metadata": {
    "id": "9r8v0aJmtGzl",
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "outputId": "d46e66dc-375c-4c40-ec6f-d59aaf561737",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": [
      "Epoch 1/200\n",
      "938/938 [==============================] - 22s 12ms/step - loss: 2.4024 - accuracy: 0.2264 - val_loss: 2.2671 - val_accuracy: 0.2273\n",
      "Epoch 2/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 2.0257 - accuracy: 0.3182 - val_loss: 2.1867 - val_accuracy: 0.3346\n",
      "Epoch 3/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 1.9726 - accuracy: 0.3299 - val_loss: 2.1757 - val_accuracy: 0.3168\n",
      "Epoch 4/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 1.9403 - accuracy: 0.3430 - val_loss: 2.1013 - val_accuracy: 0.3028\n",
      "Epoch 5/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 1.9084 - accuracy: 0.3540 - val_loss: 2.0715 - val_accuracy: 0.2842\n",
      "Epoch 6/200\n",
      "938/938 [==============================] - 10s 11ms/step - loss: 1.8771 - accuracy: 0.3688 - val_loss: 2.0130 - val_accuracy: 0.3581\n",
      "Epoch 7/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 1.8501 - accuracy: 0.3785 - val_loss: 1.9950 - val_accuracy: 0.3188\n",
      "Epoch 8/200\n",
      "938/938 [==============================] - 10s 11ms/step - loss: 1.8245 - accuracy: 0.3925 - val_loss: 1.9745 - val_accuracy: 0.3430\n",
      "Epoch 9/200\n",
      "938/938 [==============================] - 10s 11ms/step - loss: 1.7876 - accuracy: 0.4075 - val_loss: 1.8843 - val_accuracy: 0.3926\n",
      "Epoch 10/200\n",
      "938/938 [==============================] - 10s 11ms/step - loss: 1.7453 - accuracy: 0.4233 - val_loss: 1.8165 - val_accuracy: 0.4263\n",
      "Epoch 11/200\n",
      "938/938 [==============================] - 10s 11ms/step - loss: 1.6939 - accuracy: 0.4408 - val_loss: 1.7557 - val_accuracy: 0.4401\n",
      "Epoch 12/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 1.6483 - accuracy: 0.4533 - val_loss: 1.6643 - val_accuracy: 0.4717\n",
      "Epoch 13/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 1.5917 - accuracy: 0.4743 - val_loss: 1.5757 - val_accuracy: 0.4978\n",
      "Epoch 14/200\n",
      "938/938 [==============================] - 10s 11ms/step - loss: 1.5420 - accuracy: 0.4929 - val_loss: 1.5571 - val_accuracy: 0.5093\n",
      "Epoch 15/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 1.5056 - accuracy: 0.5021 - val_loss: 1.4959 - val_accuracy: 0.5315\n",
      "Epoch 16/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 1.4751 - accuracy: 0.5130 - val_loss: 1.4768 - val_accuracy: 0.5325\n",
      "Epoch 17/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 1.4408 - accuracy: 0.5254 - val_loss: 1.4007 - val_accuracy: 0.5510\n",
      "Epoch 18/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 1.4086 - accuracy: 0.5356 - val_loss: 1.4135 - val_accuracy: 0.5575\n",
      "Epoch 19/200\n",
      "938/938 [==============================] - 10s 11ms/step - loss: 1.3749 - accuracy: 0.5457 - val_loss: 1.3534 - val_accuracy: 0.5846\n",
      "Epoch 20/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 1.3572 - accuracy: 0.5539 - val_loss: 1.3126 - val_accuracy: 0.5933\n",
      "Epoch 21/200\n",
      "938/938 [==============================] - 10s 11ms/step - loss: 1.3200 - accuracy: 0.5640 - val_loss: 1.3277 - val_accuracy: 0.5782\n",
      "Epoch 22/200\n",
      "938/938 [==============================] - 10s 11ms/step - loss: 1.3075 - accuracy: 0.5701 - val_loss: 1.2582 - val_accuracy: 0.6011\n",
      "Epoch 23/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 1.2790 - accuracy: 0.5753 - val_loss: 1.2381 - val_accuracy: 0.6153\n",
      "Epoch 24/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 1.2532 - accuracy: 0.5867 - val_loss: 1.2086 - val_accuracy: 0.6163\n",
      "Epoch 25/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 1.2278 - accuracy: 0.5945 - val_loss: 1.1442 - val_accuracy: 0.6296\n",
      "Epoch 26/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 1.2111 - accuracy: 0.5981 - val_loss: 1.1664 - val_accuracy: 0.6334\n",
      "Epoch 27/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 1.1800 - accuracy: 0.6095 - val_loss: 1.1381 - val_accuracy: 0.6374\n",
      "Epoch 28/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 1.1681 - accuracy: 0.6167 - val_loss: 1.1344 - val_accuracy: 0.6425\n",
      "Epoch 29/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 1.1569 - accuracy: 0.6172 - val_loss: 1.1052 - val_accuracy: 0.6424\n",
      "Epoch 30/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 1.1323 - accuracy: 0.6241 - val_loss: 1.1242 - val_accuracy: 0.6438\n",
      "Epoch 31/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 1.1159 - accuracy: 0.6290 - val_loss: 1.0453 - val_accuracy: 0.6627\n",
      "Epoch 32/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 1.0962 - accuracy: 0.6384 - val_loss: 1.0434 - val_accuracy: 0.6682\n",
      "Epoch 33/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 1.0858 - accuracy: 0.6400 - val_loss: 1.0218 - val_accuracy: 0.6812\n",
      "Epoch 34/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 1.0677 - accuracy: 0.6448 - val_loss: 1.0147 - val_accuracy: 0.6768\n",
      "Epoch 35/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 1.0527 - accuracy: 0.6516 - val_loss: 1.0073 - val_accuracy: 0.6775\n",
      "Epoch 36/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 1.0344 - accuracy: 0.6556 - val_loss: 0.9938 - val_accuracy: 0.6811\n",
      "Epoch 37/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 1.0190 - accuracy: 0.6634 - val_loss: 0.9683 - val_accuracy: 0.6930\n",
      "Epoch 38/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 1.0077 - accuracy: 0.6668 - val_loss: 0.9350 - val_accuracy: 0.6987\n",
      "Epoch 39/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.9927 - accuracy: 0.6729 - val_loss: 0.9572 - val_accuracy: 0.6923\n",
      "Epoch 40/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.9761 - accuracy: 0.6813 - val_loss: 0.9378 - val_accuracy: 0.6942\n",
      "Epoch 41/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.9678 - accuracy: 0.6796 - val_loss: 0.9042 - val_accuracy: 0.7056\n",
      "Epoch 42/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.9579 - accuracy: 0.6817 - val_loss: 0.9211 - val_accuracy: 0.7057\n",
      "Epoch 43/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.9430 - accuracy: 0.6886 - val_loss: 0.8854 - val_accuracy: 0.7173\n",
      "Epoch 44/200\n",
      "938/938 [==============================] - 10s 11ms/step - loss: 0.9322 - accuracy: 0.6931 - val_loss: 0.8709 - val_accuracy: 0.7176\n",
      "Epoch 45/200\n",
      "938/938 [==============================] - 10s 11ms/step - loss: 0.9228 - accuracy: 0.6947 - val_loss: 0.8766 - val_accuracy: 0.7174\n",
      "Epoch 46/200\n",
      "938/938 [==============================] - 10s 11ms/step - loss: 0.9128 - accuracy: 0.6978 - val_loss: 0.8781 - val_accuracy: 0.7160\n",
      "Epoch 47/200\n",
      "938/938 [==============================] - 10s 11ms/step - loss: 0.8951 - accuracy: 0.7031 - val_loss: 0.8593 - val_accuracy: 0.7197\n",
      "Epoch 48/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.8899 - accuracy: 0.7056 - val_loss: 0.8531 - val_accuracy: 0.7172\n",
      "Epoch 49/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.8784 - accuracy: 0.7084 - val_loss: 0.8404 - val_accuracy: 0.7297\n",
      "Epoch 50/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.8670 - accuracy: 0.7117 - val_loss: 0.8340 - val_accuracy: 0.7266\n",
      "Epoch 51/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.8581 - accuracy: 0.7152 - val_loss: 0.8247 - val_accuracy: 0.7298\n",
      "Epoch 52/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.8475 - accuracy: 0.7196 - val_loss: 0.8007 - val_accuracy: 0.7395\n",
      "Epoch 53/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.8425 - accuracy: 0.7177 - val_loss: 0.8042 - val_accuracy: 0.7351\n",
      "Epoch 54/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.8322 - accuracy: 0.7258 - val_loss: 0.7948 - val_accuracy: 0.7455\n",
      "Epoch 55/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.8245 - accuracy: 0.7242 - val_loss: 0.7801 - val_accuracy: 0.7443\n",
      "Epoch 56/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.8136 - accuracy: 0.7301 - val_loss: 0.7834 - val_accuracy: 0.7442\n",
      "Epoch 57/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.7938 - accuracy: 0.7348 - val_loss: 0.7774 - val_accuracy: 0.7467\n",
      "Epoch 58/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.7899 - accuracy: 0.7373 - val_loss: 0.7944 - val_accuracy: 0.7404\n",
      "Epoch 59/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.7886 - accuracy: 0.7341 - val_loss: 0.7521 - val_accuracy: 0.7510\n",
      "Epoch 60/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.7800 - accuracy: 0.7435 - val_loss: 0.7554 - val_accuracy: 0.7524\n",
      "Epoch 61/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.7740 - accuracy: 0.7419 - val_loss: 0.7560 - val_accuracy: 0.7558\n",
      "Epoch 62/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.7605 - accuracy: 0.7464 - val_loss: 0.7453 - val_accuracy: 0.7550\n",
      "Epoch 63/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.7580 - accuracy: 0.7485 - val_loss: 0.7390 - val_accuracy: 0.7601\n",
      "Epoch 64/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.7492 - accuracy: 0.7518 - val_loss: 0.7278 - val_accuracy: 0.7620\n",
      "Epoch 65/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.7422 - accuracy: 0.7526 - val_loss: 0.7260 - val_accuracy: 0.7644\n",
      "Epoch 66/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.7355 - accuracy: 0.7531 - val_loss: 0.7118 - val_accuracy: 0.7680\n",
      "Epoch 67/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.7182 - accuracy: 0.7599 - val_loss: 0.7187 - val_accuracy: 0.7629\n",
      "Epoch 68/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.7137 - accuracy: 0.7612 - val_loss: 0.7160 - val_accuracy: 0.7669\n",
      "Epoch 69/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.7146 - accuracy: 0.7620 - val_loss: 0.7203 - val_accuracy: 0.7648\n",
      "Epoch 70/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.7024 - accuracy: 0.7666 - val_loss: 0.7039 - val_accuracy: 0.7714\n",
      "Epoch 71/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.7005 - accuracy: 0.7639 - val_loss: 0.7033 - val_accuracy: 0.7691\n",
      "Epoch 72/200\n",
      "938/938 [==============================] - 10s 11ms/step - loss: 0.6899 - accuracy: 0.7670 - val_loss: 0.6936 - val_accuracy: 0.7728\n",
      "Epoch 73/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.6875 - accuracy: 0.7692 - val_loss: 0.6914 - val_accuracy: 0.7722\n",
      "Epoch 74/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.6830 - accuracy: 0.7707 - val_loss: 0.7127 - val_accuracy: 0.7672\n",
      "Epoch 75/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.6760 - accuracy: 0.7734 - val_loss: 0.6887 - val_accuracy: 0.7782\n",
      "Epoch 76/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.6756 - accuracy: 0.7745 - val_loss: 0.7065 - val_accuracy: 0.7694\n",
      "Epoch 77/200\n",
      "938/938 [==============================] - 10s 11ms/step - loss: 0.6673 - accuracy: 0.7779 - val_loss: 0.6692 - val_accuracy: 0.7815\n",
      "Epoch 78/200\n",
      "938/938 [==============================] - 10s 11ms/step - loss: 0.6557 - accuracy: 0.7784 - val_loss: 0.6687 - val_accuracy: 0.7819\n",
      "Epoch 79/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.6557 - accuracy: 0.7783 - val_loss: 0.6607 - val_accuracy: 0.7838\n",
      "Epoch 80/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.6462 - accuracy: 0.7852 - val_loss: 0.6619 - val_accuracy: 0.7847\n",
      "Epoch 81/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.6508 - accuracy: 0.7833 - val_loss: 0.6697 - val_accuracy: 0.7790\n",
      "Epoch 82/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.6455 - accuracy: 0.7825 - val_loss: 0.6693 - val_accuracy: 0.7825\n",
      "Epoch 83/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.6241 - accuracy: 0.7904 - val_loss: 0.6539 - val_accuracy: 0.7861\n",
      "Epoch 84/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.6282 - accuracy: 0.7889 - val_loss: 0.6765 - val_accuracy: 0.7774\n",
      "Epoch 85/200\n",
      "938/938 [==============================] - 10s 11ms/step - loss: 0.6220 - accuracy: 0.7904 - val_loss: 0.6563 - val_accuracy: 0.7852\n",
      "Epoch 86/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.6200 - accuracy: 0.7938 - val_loss: 0.6539 - val_accuracy: 0.7866\n",
      "Epoch 87/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.6124 - accuracy: 0.7944 - val_loss: 0.7029 - val_accuracy: 0.7718\n",
      "Epoch 88/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.6108 - accuracy: 0.7954 - val_loss: 0.6410 - val_accuracy: 0.7879\n",
      "Epoch 89/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.6043 - accuracy: 0.7988 - val_loss: 0.6345 - val_accuracy: 0.7928\n",
      "Epoch 90/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.5914 - accuracy: 0.8010 - val_loss: 0.6562 - val_accuracy: 0.7887\n",
      "Epoch 91/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.6008 - accuracy: 0.7980 - val_loss: 0.6281 - val_accuracy: 0.7942\n",
      "Epoch 92/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.5938 - accuracy: 0.8019 - val_loss: 0.6665 - val_accuracy: 0.7814\n",
      "Epoch 93/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.5940 - accuracy: 0.8004 - val_loss: 0.6391 - val_accuracy: 0.7928\n",
      "Epoch 94/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.5905 - accuracy: 0.8034 - val_loss: 0.6392 - val_accuracy: 0.7900\n",
      "Epoch 95/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.5759 - accuracy: 0.8071 - val_loss: 0.6219 - val_accuracy: 0.7958\n",
      "Epoch 96/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.5664 - accuracy: 0.8083 - val_loss: 0.6431 - val_accuracy: 0.7898\n",
      "Epoch 97/200\n",
      "938/938 [==============================] - 10s 11ms/step - loss: 0.5624 - accuracy: 0.8095 - val_loss: 0.6376 - val_accuracy: 0.7912\n",
      "Epoch 98/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.5701 - accuracy: 0.8072 - val_loss: 0.6199 - val_accuracy: 0.7968\n",
      "Epoch 99/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.5606 - accuracy: 0.8105 - val_loss: 0.6411 - val_accuracy: 0.7895\n",
      "Epoch 100/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.5555 - accuracy: 0.8111 - val_loss: 0.6177 - val_accuracy: 0.7988\n",
      "Epoch 101/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.5543 - accuracy: 0.8137 - val_loss: 0.6324 - val_accuracy: 0.7947\n",
      "Epoch 102/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.5607 - accuracy: 0.8115 - val_loss: 0.6047 - val_accuracy: 0.8017\n",
      "Epoch 103/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.5437 - accuracy: 0.8162 - val_loss: 0.6322 - val_accuracy: 0.7898\n",
      "Epoch 104/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.5445 - accuracy: 0.8142 - val_loss: 0.6511 - val_accuracy: 0.7901\n",
      "Epoch 105/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.5437 - accuracy: 0.8181 - val_loss: 0.6192 - val_accuracy: 0.8006\n",
      "Epoch 106/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.5407 - accuracy: 0.8194 - val_loss: 0.6282 - val_accuracy: 0.7948\n",
      "Epoch 107/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.5300 - accuracy: 0.8209 - val_loss: 0.6220 - val_accuracy: 0.7964\n",
      "Epoch 108/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.5320 - accuracy: 0.8188 - val_loss: 0.6346 - val_accuracy: 0.7934\n",
      "Epoch 109/200\n",
      "938/938 [==============================] - 10s 11ms/step - loss: 0.5198 - accuracy: 0.8241 - val_loss: 0.6289 - val_accuracy: 0.7951\n",
      "Epoch 110/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.5212 - accuracy: 0.8235 - val_loss: 0.6243 - val_accuracy: 0.7977\n",
      "Epoch 111/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.5260 - accuracy: 0.8222 - val_loss: 0.5958 - val_accuracy: 0.8045\n",
      "Epoch 112/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.5201 - accuracy: 0.8230 - val_loss: 0.6065 - val_accuracy: 0.8024\n",
      "Epoch 113/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.5065 - accuracy: 0.8263 - val_loss: 0.6304 - val_accuracy: 0.7937\n",
      "Epoch 114/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.5124 - accuracy: 0.8250 - val_loss: 0.6120 - val_accuracy: 0.8006\n",
      "Epoch 115/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.5147 - accuracy: 0.8278 - val_loss: 0.6070 - val_accuracy: 0.8003\n",
      "Epoch 116/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.5063 - accuracy: 0.8267 - val_loss: 0.6159 - val_accuracy: 0.7997\n",
      "Epoch 117/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.5044 - accuracy: 0.8284 - val_loss: 0.6421 - val_accuracy: 0.7920\n",
      "Epoch 118/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.5046 - accuracy: 0.8305 - val_loss: 0.5952 - val_accuracy: 0.8059\n",
      "Epoch 119/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4929 - accuracy: 0.8356 - val_loss: 0.5902 - val_accuracy: 0.8064\n",
      "Epoch 120/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4902 - accuracy: 0.8342 - val_loss: 0.5888 - val_accuracy: 0.8074\n",
      "Epoch 121/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4850 - accuracy: 0.8357 - val_loss: 0.5927 - val_accuracy: 0.8070\n",
      "Epoch 122/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.5010 - accuracy: 0.8307 - val_loss: 0.5948 - val_accuracy: 0.8048\n",
      "Epoch 123/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4862 - accuracy: 0.8373 - val_loss: 0.5964 - val_accuracy: 0.8072\n",
      "Epoch 124/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4888 - accuracy: 0.8348 - val_loss: 0.6094 - val_accuracy: 0.8029\n",
      "Epoch 125/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4792 - accuracy: 0.8365 - val_loss: 0.6386 - val_accuracy: 0.7924\n",
      "Epoch 126/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4761 - accuracy: 0.8392 - val_loss: 0.6094 - val_accuracy: 0.8016\n",
      "Epoch 127/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4687 - accuracy: 0.8394 - val_loss: 0.5788 - val_accuracy: 0.8117\n",
      "Epoch 128/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4739 - accuracy: 0.8407 - val_loss: 0.6070 - val_accuracy: 0.8046\n",
      "Epoch 129/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4646 - accuracy: 0.8399 - val_loss: 0.5919 - val_accuracy: 0.8054\n",
      "Epoch 130/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4618 - accuracy: 0.8434 - val_loss: 0.5931 - val_accuracy: 0.8075\n",
      "Epoch 131/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4703 - accuracy: 0.8402 - val_loss: 0.5840 - val_accuracy: 0.8100\n",
      "Epoch 132/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4620 - accuracy: 0.8423 - val_loss: 0.5795 - val_accuracy: 0.8119\n",
      "Epoch 133/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4647 - accuracy: 0.8423 - val_loss: 0.5833 - val_accuracy: 0.8100\n",
      "Epoch 134/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4562 - accuracy: 0.8427 - val_loss: 0.5988 - val_accuracy: 0.8085\n",
      "Epoch 135/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4560 - accuracy: 0.8438 - val_loss: 0.5908 - val_accuracy: 0.8084\n",
      "Epoch 136/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4551 - accuracy: 0.8448 - val_loss: 0.5838 - val_accuracy: 0.8113\n",
      "Epoch 137/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4510 - accuracy: 0.8458 - val_loss: 0.5719 - val_accuracy: 0.8136\n",
      "Epoch 138/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4545 - accuracy: 0.8450 - val_loss: 0.6020 - val_accuracy: 0.8072\n",
      "Epoch 139/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4431 - accuracy: 0.8489 - val_loss: 0.5905 - val_accuracy: 0.8091\n",
      "Epoch 140/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4477 - accuracy: 0.8473 - val_loss: 0.5966 - val_accuracy: 0.8084\n",
      "Epoch 141/200\n",
      "938/938 [==============================] - 10s 11ms/step - loss: 0.4370 - accuracy: 0.8502 - val_loss: 0.5717 - val_accuracy: 0.8146\n",
      "Epoch 142/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4486 - accuracy: 0.8484 - val_loss: 0.5904 - val_accuracy: 0.8086\n",
      "Epoch 143/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4413 - accuracy: 0.8505 - val_loss: 0.5883 - val_accuracy: 0.8109\n",
      "Epoch 144/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4363 - accuracy: 0.8527 - val_loss: 0.5845 - val_accuracy: 0.8115\n",
      "Epoch 145/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4420 - accuracy: 0.8489 - val_loss: 0.5825 - val_accuracy: 0.8133\n",
      "Epoch 146/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4351 - accuracy: 0.8493 - val_loss: 0.5776 - val_accuracy: 0.8129\n",
      "Epoch 147/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4370 - accuracy: 0.8507 - val_loss: 0.5791 - val_accuracy: 0.8122\n",
      "Epoch 148/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4283 - accuracy: 0.8531 - val_loss: 0.5801 - val_accuracy: 0.8138\n",
      "Epoch 149/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4273 - accuracy: 0.8531 - val_loss: 0.5912 - val_accuracy: 0.8097\n",
      "Epoch 150/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4229 - accuracy: 0.8559 - val_loss: 0.5806 - val_accuracy: 0.8130\n",
      "Epoch 151/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4324 - accuracy: 0.8541 - val_loss: 0.6183 - val_accuracy: 0.8017\n",
      "Epoch 152/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4226 - accuracy: 0.8545 - val_loss: 0.5585 - val_accuracy: 0.8187\n",
      "Epoch 153/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4221 - accuracy: 0.8559 - val_loss: 0.5894 - val_accuracy: 0.8106\n",
      "Epoch 154/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4268 - accuracy: 0.8538 - val_loss: 0.5786 - val_accuracy: 0.8123\n",
      "Epoch 155/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4134 - accuracy: 0.8601 - val_loss: 0.6023 - val_accuracy: 0.8074\n",
      "Epoch 156/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4160 - accuracy: 0.8594 - val_loss: 0.6025 - val_accuracy: 0.8079\n",
      "Epoch 157/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4142 - accuracy: 0.8600 - val_loss: 0.5697 - val_accuracy: 0.8170\n",
      "Epoch 158/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4143 - accuracy: 0.8591 - val_loss: 0.5745 - val_accuracy: 0.8168\n",
      "Epoch 159/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4157 - accuracy: 0.8580 - val_loss: 0.5698 - val_accuracy: 0.8174\n",
      "Epoch 160/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4092 - accuracy: 0.8615 - val_loss: 0.5917 - val_accuracy: 0.8088\n",
      "Epoch 161/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4083 - accuracy: 0.8604 - val_loss: 0.5514 - val_accuracy: 0.8224\n",
      "Epoch 162/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4059 - accuracy: 0.8623 - val_loss: 0.5663 - val_accuracy: 0.8180\n",
      "Epoch 163/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4088 - accuracy: 0.8591 - val_loss: 0.5652 - val_accuracy: 0.8199\n",
      "Epoch 164/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4014 - accuracy: 0.8626 - val_loss: 0.5794 - val_accuracy: 0.8133\n",
      "Epoch 165/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4060 - accuracy: 0.8619 - val_loss: 0.5618 - val_accuracy: 0.8186\n",
      "Epoch 166/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.4022 - accuracy: 0.8628 - val_loss: 0.5821 - val_accuracy: 0.8126\n",
      "Epoch 167/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.3947 - accuracy: 0.8643 - val_loss: 0.5636 - val_accuracy: 0.8178\n",
      "Epoch 168/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.3964 - accuracy: 0.8643 - val_loss: 0.5650 - val_accuracy: 0.8172\n",
      "Epoch 169/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.3978 - accuracy: 0.8644 - val_loss: 0.5547 - val_accuracy: 0.8210\n",
      "Epoch 170/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.3923 - accuracy: 0.8667 - val_loss: 0.5814 - val_accuracy: 0.8141\n",
      "Epoch 171/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.3930 - accuracy: 0.8665 - val_loss: 0.5721 - val_accuracy: 0.8155\n",
      "Epoch 172/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.3899 - accuracy: 0.8668 - val_loss: 0.5606 - val_accuracy: 0.8190\n",
      "Epoch 173/200\n",
      "938/938 [==============================] - 10s 11ms/step - loss: 0.3934 - accuracy: 0.8660 - val_loss: 0.5806 - val_accuracy: 0.8162\n",
      "Epoch 174/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.3932 - accuracy: 0.8667 - val_loss: 0.5910 - val_accuracy: 0.8114\n",
      "Epoch 175/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.3888 - accuracy: 0.8694 - val_loss: 0.5638 - val_accuracy: 0.8190\n",
      "Epoch 176/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.3905 - accuracy: 0.8672 - val_loss: 0.5760 - val_accuracy: 0.8166\n",
      "Epoch 177/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.3858 - accuracy: 0.8688 - val_loss: 0.5677 - val_accuracy: 0.8202\n",
      "Epoch 178/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.3814 - accuracy: 0.8671 - val_loss: 0.5670 - val_accuracy: 0.8200\n",
      "Epoch 179/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.3757 - accuracy: 0.8715 - val_loss: 0.5551 - val_accuracy: 0.8218\n",
      "Epoch 180/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.3842 - accuracy: 0.8684 - val_loss: 0.5685 - val_accuracy: 0.8193\n",
      "Epoch 181/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.3809 - accuracy: 0.8705 - val_loss: 0.5550 - val_accuracy: 0.8229\n",
      "Epoch 182/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.3886 - accuracy: 0.8699 - val_loss: 0.5657 - val_accuracy: 0.8191\n",
      "Epoch 183/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.3786 - accuracy: 0.8685 - val_loss: 0.5576 - val_accuracy: 0.8233\n",
      "Epoch 184/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.3788 - accuracy: 0.8695 - val_loss: 0.5686 - val_accuracy: 0.8179\n",
      "Epoch 185/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.3746 - accuracy: 0.8722 - val_loss: 0.5712 - val_accuracy: 0.8199\n",
      "Epoch 186/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.3733 - accuracy: 0.8740 - val_loss: 0.5644 - val_accuracy: 0.8207\n",
      "Epoch 187/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.3779 - accuracy: 0.8703 - val_loss: 0.5549 - val_accuracy: 0.8228\n",
      "Epoch 188/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.3630 - accuracy: 0.8745 - val_loss: 0.5681 - val_accuracy: 0.8199\n",
      "Epoch 189/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.3715 - accuracy: 0.8742 - val_loss: 0.5658 - val_accuracy: 0.8213\n",
      "Epoch 190/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.3699 - accuracy: 0.8748 - val_loss: 0.5787 - val_accuracy: 0.8137\n",
      "Epoch 191/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.3642 - accuracy: 0.8751 - val_loss: 0.5613 - val_accuracy: 0.8240\n",
      "Epoch 192/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.3743 - accuracy: 0.8739 - val_loss: 0.5877 - val_accuracy: 0.8129\n",
      "Epoch 193/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.3656 - accuracy: 0.8761 - val_loss: 0.5723 - val_accuracy: 0.8185\n",
      "Epoch 194/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.3625 - accuracy: 0.8742 - val_loss: 0.5612 - val_accuracy: 0.8213\n",
      "Epoch 195/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.3664 - accuracy: 0.8746 - val_loss: 0.5593 - val_accuracy: 0.8231\n",
      "Epoch 196/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.3609 - accuracy: 0.8773 - val_loss: 0.5584 - val_accuracy: 0.8230\n",
      "Epoch 197/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.3621 - accuracy: 0.8765 - val_loss: 0.5824 - val_accuracy: 0.8150\n",
      "Epoch 198/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.3575 - accuracy: 0.8773 - val_loss: 0.5632 - val_accuracy: 0.8210\n",
      "Epoch 199/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.3525 - accuracy: 0.8802 - val_loss: 0.5849 - val_accuracy: 0.8169\n",
      "Epoch 200/200\n",
      "938/938 [==============================] - 10s 10ms/step - loss: 0.3532 - accuracy: 0.8795 - val_loss: 0.5821 - val_accuracy: 0.8169\n",
      "Time passed: 1966.411502122879 seconds.\n"
     ]
    }
   ]
  },
  {
   "cell_type": "code",
   "source": [
    "# Plot training & validation accuracy values\n",
    "plt.plot(history.history['accuracy'])\n",
    "plt.plot(history.history['val_accuracy'])\n",
    "plt.title('Model accuracy')\n",
    "plt.ylabel('Accuracy')\n",
    "plt.xlabel('Epoch')\n",
    "plt.legend(['Train', 'Valid'], loc='upper left')\n",
    "plt.savefig('{}_accuracy.png'.format(time.strftime(\"%Y%m%d-%H%M%S\")))\n",
    "plt.show()\n",
    "\n",
    "t = time_passed(t)\n"
   ],
   "metadata": {
    "id": "c1r_xyMkfRoP",
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 312
    },
    "outputId": "4c681bc6-2afe-478b-b497-9cecc660c117",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": [
    {
     "output_type": "display_data",
     "data": {
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ],
      "image/png": "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\n"
     },
     "metadata": {
      "needs_background": "light"
     }
    },
    {
     "output_type": "stream",
     "name": "stdout",
     "text": [
      "Time passed: 0.22568941116333008 seconds.\n"
     ]
    }
   ]
  },
  {
   "cell_type": "code",
   "source": [
    "model.save('./model')\n",
    "print('Saved trained model')\n",
    "\n",
    "t = time_passed(t)"
   ],
   "metadata": {
    "id": "wk_UsuvP2wZM",
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "outputId": "2f92bd79-1d45-49ff-e887-a0b6d45916b4",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": [
      "INFO:tensorflow:Assets written to: /drive/My Drive/dataset/model/assets\n",
      "Saved trained model\n",
      "Time passed: 2.12528920173645 seconds.\n"
     ]
    }
   ]
  },
  {
   "cell_type": "code",
   "source": [
    "model = keras.models.load_model('./model')"
   ],
   "metadata": {
    "id": "wvCuCGKvh6ht",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "source": [
    "scores = model.evaluate(X_test, y_test, verbose=1)\n",
    "print('Test loss:', scores[0])\n",
    "print('Test accuracy:', scores[1])\n",
    "t = time_passed(t)"
   ],
   "metadata": {
    "id": "6do7cxXW2rac",
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "outputId": "f9b10caf-f2f9-41ab-b1df-1f1c3d39b4aa",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": [
      "1118/1118 [==============================] - 4s 3ms/step - loss: 0.5821 - accuracy: 0.8169\n",
      "Test loss: 0.5820885896682739\n",
      "Test accuracy: 0.8168883919715881\n",
      "Time passed: 6.066837787628174 seconds.\n"
     ]
    }
   ]
  },
  {
   "cell_type": "code",
   "source": [
    "y_pred = model.predict(X_test)"
   ],
   "metadata": {
    "id": "RuP5hD7zDXLL",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "source": [
    "y_pred_classes = np.argmax(y_pred, axis=-1)\n",
    "y_pred_classes = reconvert_labels(y_pred_classes)\n",
    "y_pred_classes"
   ],
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "XX6lqB_SEQPZ",
    "outputId": "7db81f2f-c6d3-4bdd-dc91-985120c75a5b",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": [
       "array(['pia', 'pia', 'voi', ..., 'pia', 'sax', 'gel'], dtype='<U3')"
      ]
     },
     "metadata": {},
     "execution_count": 21
    }
   ]
  },
  {
   "cell_type": "code",
   "source": [
    "y_test_classes = np.argmax(y_test, axis=-1)\n",
    "y_test_classes = reconvert_labels(y_test_classes)\n",
    "y_test_classes"
   ],
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "KEPxIPUeFi8-",
    "outputId": "75b7851c-b019-49fb-93be-6658d0f11218",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": [
       "array(['pia', 'pia', 'voi', ..., 'gel', 'pia', 'org'], dtype='<U3')"
      ]
     },
     "metadata": {},
     "execution_count": 22
    }
   ]
  },
  {
   "cell_type": "code",
   "source": [
    "from sklearn.metrics import confusion_matrix\n",
    "instruments = [\"cel\", \"cla\", \"flu\", \"gac\", \"gel\", \"org\", \"pia\", \"sax\", \"tru\", \"vio\", \"voi\"]\n",
    "cf_matrix = confusion_matrix(y_test_classes, y_pred_classes, labels=instruments)"
   ],
   "metadata": {
    "id": "sOGHufgtFOTE",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "source": [
    "import seaborn as sns\n",
    "plt.figure(figsize = (10,7))\n",
    "sns.color_palette(\"flare\", as_cmap=True)\n",
    "plt.tick_params(axis='both', which='major', labelsize=10, labelbottom = False, bottom=False, top = False, labeltop=True)\n",
    "sns.heatmap(cf_matrix, annot=True, fmt='d', xticklabels=instruments, yticklabels=instruments, cmap=\"flare\")"
   ],
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 445
    },
    "id": "99m6SNbfHEYd",
    "outputId": "8bc3b7a4-8d6b-4fa1-ba6d-c8f7565d9761",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "execution_count": null,
   "outputs": [
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7f631d296990>"
      ]
     },
     "metadata": {},
     "execution_count": 24
    },
    {
     "output_type": "display_data",
     "data": {
      "text/plain": [
       "<Figure size 720x504 with 2 Axes>"
      ],
      "image/png": "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\n"
     },
     "metadata": {
      "needs_background": "light"
     }
    }
   ]
  }
 ]
}