3_GTzan_3f_CNN_2D_75w-Load-110250_frozen_stft.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true,
    "scrolled": true
   },
   "source": [
    "# End to end 2D CNN for GTzan music classification\n",
    "\n",
    "\n",
    "WINDOWED Version\n",
    "\n",
    "Adapted by AL Koerich\n",
    "\n",
    "- Modified by KM Koerich\n",
    "\n",
    "To GTzan 3-fold\n",
    "\n",
    "11 December 2018\n",
    "\n",
    "- Last Update: 08 August 2019"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import matplotlib\n",
    "import os, sys\n",
    "import soundfile as sf\n",
    "import gc\n",
    "\n",
    "from sklearn.utils import shuffle\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.preprocessing import normalize\n",
    "from sklearn.preprocessing import scale\n",
    "from keras import regularizers\n",
    "\n",
    "import os, sys\n",
    "from keras.utils import np_utils\n",
    "from keras.models import Model\n",
    "from keras.layers import Conv1D, Dense, MaxPool1D, Flatten\n",
    "from keras.callbacks import TensorBoard\n",
    "from keras.utils import np_utils, to_categorical\n",
    "from keras import optimizers\n",
    "from keras.layers.normalization import BatchNormalization\n",
    "from keras.layers.core import Dropout\n",
    "import keras.initializers as init\n",
    "from keras.utils.training_utils import multi_gpu_model\n",
    "\n",
    "os.environ[\"CUDA_VISIBLE_DEVICES\"]=\"0\"\n",
    "import tensorflow as tf\n",
    "config = tf.ConfigProto( )\n",
    "config.gpu_options.allow_growth = True\n",
    "sess   = tf.Session(config=config)\n",
    "import keras.backend.tensorflow_backend as tf_bkend\n",
    "tf_bkend.set_session(sess)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "#controling_Hyper parameters\n",
    "\n",
    "nb_classes = 10\n",
    "nb_epoch   = 150\n",
    "frame_size = 110250"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "fold1-2\n"
     ]
    }
   ],
   "source": [
    "#Indicate folds\n",
    "train_fold = [1, 2]\n",
    "test_fold = 3\n",
    "\n",
    "str_train_fold = \"fold\"+str(train_fold[0])+\"-\"+str(train_fold[1])\n",
    "print(str_train_fold)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train = np.load( \"folds_stft/2_GTzan_Xs_train_\"+str_train_fold+\"_110250_75_frozen_stft.npy\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "Y_train = np.load( \"folds_stft/2_GTzan_Ys_train_\"+str_train_fold+\"_110250_75_frozen_stft.npy\" )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0.0, 1.0, (14051, 10))"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Y_train.min(), Y_train.max(), Y_train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Adapt 1D data to 2D CNN\n",
    "X_train = np.squeeze(X_train)\n",
    "X_train = np.expand_dims(X_train, axis = 3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(14051, 257, 431, 1)"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "13"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gc.collect()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "f = X_train.shape[1]\n",
    "g = X_train.shape[2]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "def model_generator_GTzannet2D_1a():\n",
    "    \n",
    "    from keras.layers      import Input, Dense, Conv2D, AveragePooling1D, LeakyReLU, MaxPool2D, Flatten\n",
    "    from keras.layers.core import Dropout\n",
    "    from keras.models      import Model\n",
    "    from keras             import initializers, optimizers, regularizers\n",
    "    from keras.callbacks   import ModelCheckpoint\n",
    "    from keras.utils       import multi_gpu_model\n",
    "    \n",
    "    from keras.layers.normalization import BatchNormalization\n",
    "        \n",
    "    import keras.initializers as init\n",
    "\n",
    "    from kapre.utils          import Normalization2D\n",
    "    from kapre.augmentation   import AdditiveNoise    \n",
    "    \n",
    "    sr = 22050\n",
    "    \n",
    "    inp   = Input(shape = (f, g, 1)) \n",
    "    #----------------------\n",
    "    conv1  = Conv2D(filters = 32, kernel_size = (3, 3), activation = 'relu')(inp)\n",
    "    norm1  = BatchNormalization()(conv1)\n",
    "    #----------------------\n",
    "    conv2  = Conv2D(filters = 32, kernel_size = (3, 3) )(norm1)\n",
    "    act2   = LeakyReLU(alpha = 0.2)(conv2)\n",
    "    pool2  = MaxPool2D(pool_size = 2, strides = 2)(act2)\n",
    "    drop2  = Dropout(0.05)(pool2)\n",
    "    #----------------------\n",
    "    conv3  = Conv2D(filters = 64, kernel_size = (3, 3) )(drop2)\n",
    "    act3   = LeakyReLU(alpha = 0.2)(conv3)\n",
    "    #----------------------\n",
    "    conv4  = Conv2D(filters = 64, kernel_size = (3, 3) )(act3)\n",
    "    act4   = LeakyReLU(alpha = 0.2)(conv4)\n",
    "    pool4  = MaxPool2D(pool_size = 4, strides = 2)(act4)\n",
    "    #----------------------\n",
    "    flat   = Flatten()(pool4)  \n",
    "    #----------------------    \n",
    "    dense3 = Dense(512, activation='relu', kernel_initializer = initializers.glorot_uniform(seed = 0))(flat)\n",
    "    #dense3 = Dense(1024, activation='relu', kernel_initializer = initializers.glorot_uniform(seed = 0))(flat)\n",
    "    drop3  = Dropout(0.80)(dense3)    \n",
    "    #----------------------\n",
    "    dense4 = Dense(nb_classes, activation='softmax')(drop3)\n",
    "    #----------------------\n",
    "\n",
    "    model  = Model(inp, dense4)\n",
    "\n",
    "    model.compile(loss = 'categorical_crossentropy',\n",
    "                  optimizer = optimizers.Adadelta(lr = 1.0, rho = 0.95, epsilon = 1e-08, decay = 0.0),\n",
    "                  metrics = ['accuracy'] )\n",
    "    \n",
    "    model.summary()\n",
    "    \n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_1 (InputLayer)         (None, 257, 431, 1)       0         \n",
      "_________________________________________________________________\n",
      "conv2d_1 (Conv2D)            (None, 255, 429, 32)      320       \n",
      "_________________________________________________________________\n",
      "batch_normalization_1 (Batch (None, 255, 429, 32)      128       \n",
      "_________________________________________________________________\n",
      "conv2d_2 (Conv2D)            (None, 253, 427, 32)      9248      \n",
      "_________________________________________________________________\n",
      "leaky_re_lu_1 (LeakyReLU)    (None, 253, 427, 32)      0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_1 (MaxPooling2 (None, 126, 213, 32)      0         \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 126, 213, 32)      0         \n",
      "_________________________________________________________________\n",
      "conv2d_3 (Conv2D)            (None, 124, 211, 64)      18496     \n",
      "_________________________________________________________________\n",
      "leaky_re_lu_2 (LeakyReLU)    (None, 124, 211, 64)      0         \n",
      "_________________________________________________________________\n",
      "conv2d_4 (Conv2D)            (None, 122, 209, 64)      36928     \n",
      "_________________________________________________________________\n",
      "leaky_re_lu_3 (LeakyReLU)    (None, 122, 209, 64)      0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_2 (MaxPooling2 (None, 60, 103, 64)       0         \n",
      "_________________________________________________________________\n",
      "flatten_1 (Flatten)          (None, 395520)            0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 512)               202506752 \n",
      "_________________________________________________________________\n",
      "dropout_2 (Dropout)          (None, 512)               0         \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 10)                5130      \n",
      "=================================================================\n",
      "Total params: 202,577,002\n",
      "Trainable params: 202,576,938\n",
      "Non-trainable params: 64\n",
      "_________________________________________________________________\n",
      "weights/weights_3_GTzan_3f_fold1-2_20p_110250_75_frozen_stft.best.hdf5\n"
     ]
    }
   ],
   "source": [
    "\n",
    "import time\n",
    "from keras.callbacks import ModelCheckpoint\n",
    "from livelossplot    import PlotLossesKeras\n",
    "from keras           import optimizers\n",
    "hist = []\n",
    "\n",
    "#ale = model_generator_GTzannet2D_1a()\n",
    "model = model_generator_GTzannet2D_1a()\n",
    "#model = multi_gpu_model(ale, gpus=2)\n",
    "\n",
    "#checkpoints\n",
    "str0 = \"weights/\"\n",
    "str1 = \"weights_3_GTzan_3f_\"+str_train_fold+\"_20p_110250_75_frozen_stft\" \n",
    "str2 = \".best.hdf5\" \n",
    "filepath = str0+str1+str2\n",
    "print(filepath)\n",
    "\n",
    "checkpoint = ModelCheckpoint( filepath, monitor = 'val_acc', verbose = 1, save_best_only = True, mode = 'max' ) \n",
    "\n",
    "callbacks_list = [checkpoint, PlotLossesKeras()]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x576 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Log-loss (cost function):\n",
      "training   (min:    0.080, max:    2.529, cur:    0.093)\n",
      "validation (min:    0.850, max:   12.650, cur:    1.629)\n",
      "\n",
      "Accuracy:\n",
      "training   (min:    0.240, max:    0.979, cur:    0.978)\n",
      "validation (min:    0.104, max:    0.797, cur:    0.464)\n",
      "Epoch 77/150\n",
      " 9030/11240 [=======================>......] - ETA: 19s - loss: 0.0644 - acc: 0.9796"
     ]
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-13-a07123c43e51>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      8\u001b[0m                       \u001b[0mshuffle\u001b[0m    \u001b[0;34m=\u001b[0m \u001b[0;32mTrue\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      9\u001b[0m                       \u001b[0mcallbacks\u001b[0m  \u001b[0;34m=\u001b[0m \u001b[0mcallbacks_list\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 10\u001b[0;31m                       \u001b[0mvalidation_split\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;36m0.2\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     11\u001b[0m                      ))\n",
      "\u001b[0;32m/usr/local/lib/python3.5/dist-packages/keras/engine/training.py\u001b[0m in \u001b[0;36mfit\u001b[0;34m(self, x, y, batch_size, epochs, verbose, callbacks, validation_split, validation_data, shuffle, class_weight, sample_weight, initial_epoch, steps_per_epoch, validation_steps, **kwargs)\u001b[0m\n\u001b[1;32m   1703\u001b[0m                               \u001b[0minitial_epoch\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0minitial_epoch\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1704\u001b[0m                               \u001b[0msteps_per_epoch\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0msteps_per_epoch\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1705\u001b[0;31m                               validation_steps=validation_steps)\n\u001b[0m\u001b[1;32m   1706\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1707\u001b[0m     def evaluate(self, x=None, y=None,\n",
      "\u001b[0;32m/usr/local/lib/python3.5/dist-packages/keras/engine/training.py\u001b[0m in \u001b[0;36m_fit_loop\u001b[0;34m(self, f, ins, out_labels, batch_size, epochs, verbose, callbacks, val_f, val_ins, shuffle, callback_metrics, initial_epoch, steps_per_epoch, validation_steps)\u001b[0m\n\u001b[1;32m   1233\u001b[0m                         \u001b[0mins_batch\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mi\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mins_batch\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mi\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtoarray\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1234\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1235\u001b[0;31m                     \u001b[0mouts\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mf\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mins_batch\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1236\u001b[0m                     \u001b[0;32mif\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0misinstance\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mouts\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlist\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1237\u001b[0m                         \u001b[0mouts\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0mouts\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/usr/local/lib/python3.5/dist-packages/keras/backend/tensorflow_backend.py\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, inputs)\u001b[0m\n\u001b[1;32m   2476\u001b[0m         \u001b[0msession\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mget_session\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2477\u001b[0m         updated = session.run(fetches=fetches, feed_dict=feed_dict,\n\u001b[0;32m-> 2478\u001b[0;31m                               **self.session_kwargs)\n\u001b[0m\u001b[1;32m   2479\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mupdated\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0mlen\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0moutputs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2480\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/usr/local/lib/python3.5/dist-packages/tensorflow/python/client/session.py\u001b[0m in \u001b[0;36mrun\u001b[0;34m(self, fetches, feed_dict, options, run_metadata)\u001b[0m\n\u001b[1;32m    903\u001b[0m     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    904\u001b[0m       result = self._run(None, fetches, feed_dict, options_ptr,\n\u001b[0;32m--> 905\u001b[0;31m                          run_metadata_ptr)\n\u001b[0m\u001b[1;32m    906\u001b[0m       \u001b[0;32mif\u001b[0m \u001b[0mrun_metadata\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    907\u001b[0m         \u001b[0mproto_data\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtf_session\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mTF_GetBuffer\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mrun_metadata_ptr\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/usr/local/lib/python3.5/dist-packages/tensorflow/python/client/session.py\u001b[0m in \u001b[0;36m_run\u001b[0;34m(self, handle, fetches, feed_dict, options, run_metadata)\u001b[0m\n\u001b[1;32m   1138\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mfinal_fetches\u001b[0m \u001b[0;32mor\u001b[0m \u001b[0mfinal_targets\u001b[0m \u001b[0;32mor\u001b[0m \u001b[0;34m(\u001b[0m\u001b[0mhandle\u001b[0m \u001b[0;32mand\u001b[0m \u001b[0mfeed_dict_tensor\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1139\u001b[0m       results = self._do_run(handle, final_targets, final_fetches,\n\u001b[0;32m-> 1140\u001b[0;31m                              feed_dict_tensor, options, run_metadata)\n\u001b[0m\u001b[1;32m   1141\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1142\u001b[0m       \u001b[0mresults\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/usr/local/lib/python3.5/dist-packages/tensorflow/python/client/session.py\u001b[0m in \u001b[0;36m_do_run\u001b[0;34m(self, handle, target_list, fetch_list, feed_dict, options, run_metadata)\u001b[0m\n\u001b[1;32m   1319\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mhandle\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1320\u001b[0m       return self._do_call(_run_fn, feeds, fetches, targets, options,\n\u001b[0;32m-> 1321\u001b[0;31m                            run_metadata)\n\u001b[0m\u001b[1;32m   1322\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1323\u001b[0m       \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_do_call\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0m_prun_fn\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mhandle\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfeeds\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfetches\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/usr/local/lib/python3.5/dist-packages/tensorflow/python/client/session.py\u001b[0m in \u001b[0;36m_do_call\u001b[0;34m(self, fn, *args)\u001b[0m\n\u001b[1;32m   1325\u001b[0m   \u001b[0;32mdef\u001b[0m \u001b[0m_do_call\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfn\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1326\u001b[0m     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1327\u001b[0;31m       \u001b[0;32mreturn\u001b[0m \u001b[0mfn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1328\u001b[0m     \u001b[0;32mexcept\u001b[0m \u001b[0merrors\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mOpError\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1329\u001b[0m       \u001b[0mmessage\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcompat\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mas_text\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0me\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmessage\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/usr/local/lib/python3.5/dist-packages/tensorflow/python/client/session.py\u001b[0m in \u001b[0;36m_run_fn\u001b[0;34m(feed_dict, fetch_list, target_list, options, run_metadata)\u001b[0m\n\u001b[1;32m   1310\u001b[0m       \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_extend_graph\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1311\u001b[0m       return self._call_tf_sessionrun(\n\u001b[0;32m-> 1312\u001b[0;31m           options, feed_dict, fetch_list, target_list, run_metadata)\n\u001b[0m\u001b[1;32m   1313\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1314\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0m_prun_fn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mhandle\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfeed_dict\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfetch_list\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/usr/local/lib/python3.5/dist-packages/tensorflow/python/client/session.py\u001b[0m in \u001b[0;36m_call_tf_sessionrun\u001b[0;34m(self, options, feed_dict, fetch_list, target_list, run_metadata)\u001b[0m\n\u001b[1;32m   1418\u001b[0m         return tf_session.TF_Run(\n\u001b[1;32m   1419\u001b[0m             \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_session\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0moptions\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfeed_dict\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfetch_list\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtarget_list\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1420\u001b[0;31m             status, run_metadata)\n\u001b[0m\u001b[1;32m   1421\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1422\u001b[0m   \u001b[0;32mdef\u001b[0m \u001b[0m_call_tf_sessionprun\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mhandle\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfeed_dict\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfetch_list\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "#fitting the model\n",
    "gc.collect()\n",
    "batch_size = 35\n",
    "hist.append(model.fit(X_train, Y_train,\n",
    "                      batch_size = batch_size, \n",
    "                      epochs     = nb_epoch,\n",
    "                      verbose    = 1,\n",
    "                      shuffle    = True,\n",
    "                      callbacks  = callbacks_list,\n",
    "                      validation_split = 0.2\n",
    "                     ))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "filepath"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "https://www.music-ir.org/nema_out/mirex2017/results/act/latin_report/accperfold.html\n",
    "\n",
    "MIREX 2017:\n",
    "\n",
    "\n",
    "\n",
    "Summary Results    [top]\n",
    "Algorithm\n",
    "Classification Accuracy\n",
    "Normalised Classification Accuracy\n",
    "LPNKK1\n",
    "0.7586\n",
    "0.7571\n",
    "PLNPH1\n",
    "0.6619\n",
    "0.6573\n",
    "XLJ1\n",
    "0.6148\n",
    "0.6079\n",
    "LPNKK3\n",
    "0.7347\n",
    "0.7324\n",
    "LPNKK2\n",
    "0.6511\n",
    "0.6458\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "Fold\n",
    "LPNKK1\n",
    "LPNKK2\n",
    "LPNKK3\n",
    "PLNPH1\n",
    "XLJ1\n",
    "1\n",
    "0.7186\n",
    "0.6041\n",
    "0.6860\n",
    "0.5795\n",
    "0.5453\n",
    "0\n",
    "0.7627\n",
    "0.6426\n",
    "0.7444\n",
    "0.6772\n",
    "0.5998\n",
    "2\n",
    "0.8055\n",
    "0.7194\n",
    "0.7872\n",
    "0.7518\n",
    "0.7183\n",
    "\n",
    "\n",
    "/Users/akoerich/Dropbox/Mendeley/pdf/Aytar, Vondrick, Torralba - 2016.pdf\n",
    "\n",
    "/Users/akoerich/HOME/ETS/Etudiants/Sajjad Abdoli/DGA1032/DGA1032-Dissertation_abdoli.pdf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}