BCI_Hyperparameter_Grid_Search.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "SHUFFLE = True\n",
    "BATCH_SIZE = 20\n",
    "TEST_TRAIN = 0.2\n",
    "\n",
    "KERNEL_SIZE = (100, 50, 25, 12, 6)\n",
    "N_KERNELS = (10,20,40,80,160)\n",
    "POOL_SIZE = (5,2,5,2)\n",
    "\n",
    "\n",
    "DROPOUT = (0.1, 0.2, 0.3, 0.4, 0.5)\n",
    "DENSE_LEN = 10"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "import scipy.io as sio\n",
    "import numpy as np\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn import preprocessing\n",
    "# make sure you don't hog all the video memory\n",
    "import tensorflow as tf\n",
    "config = tf.ConfigProto()\n",
    "config.gpu_options.allow_growth = True\n",
    "sess = tf.Session(config=config)\n",
    "from keras import backend as K\n",
    "K.set_session(sess)\n",
    "###################################\n",
    "from keras.layers import Input, Dense, Conv1D, MaxPooling1D, UpSampling2D, Dropout, BatchNormalization, Flatten, ELU\n",
    "from keras.models import Model\n",
    "from keras.models import load_model\n",
    "from keras.callbacks import TensorBoard, EarlyStopping\n",
    "from keras import optimizers\n",
    "from keras.initializers import Orthogonal as orth\n",
    "from keras.callbacks import TensorBoard, EarlyStopping\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "paths = ['/home/amplifier/home/DATASETS/BCI_0101T_TD.mat',\n",
    "         '/home/amplifier/home/DATASETS/BCI_0202T_TD.mat',\n",
    "        '/home/amplifier/home/DATASETS/BCI_0303T_TD.mat',\n",
    "        '/home/amplifier/home/DATASETS/BCI_0403T_TD.mat',\n",
    "        '/home/amplifier/home/DATASETS/BCI_0503T_TD.mat',\n",
    "        '/home/amplifier/home/DATASETS/BCI_0603T_TD.mat',\n",
    "        '/home/amplifier/home/DATASETS/BCI_0703T_TD.mat',\n",
    "        '/home/amplifier/home/DATASETS/BCI_0801T_TD.mat']\n",
    "\n",
    "def load_file(file):\n",
    "    print(file)\n",
    "    mat_contents = sio.loadmat(file)\n",
    "    X = mat_contents['X']\n",
    "    Y = mat_contents['Z']\n",
    "\n",
    "    if X.shape[1]<X.shape[2]:\n",
    "        X = np.transpose(X,[0,2,1])\n",
    "\n",
    "    if Y.shape[1] > Y.shape[0]:\n",
    "        Y = Y.T\n",
    "\n",
    "    x_train, x_test, y_train, y_test = train_test_split(X, Y, test_size=TEST_TRAIN, shuffle=SHUFFLE)\n",
    "\n",
    "    # convert to float64 for numerical stability:\n",
    "    x_train = x_train.astype('float64')\n",
    "    y_train = y_train.astype('float64')\n",
    "    x_test = x_test.astype('float64')\n",
    "    y_test = y_test.astype('float64')\n",
    "\n",
    "\n",
    "    # one hot encode the labels:\n",
    "    onehot_encoder = preprocessing.OneHotEncoder(sparse=False)\n",
    "    y_train = onehot_encoder.fit_transform(y_train)\n",
    "    y_test = onehot_encoder.fit_transform(y_test)\n",
    "\n",
    "    # convert to float16 to save space:\n",
    "    x_train = x_train.astype('float16')\n",
    "    y_train = y_train.astype('float16')\n",
    "    x_test = x_test.astype('float16')\n",
    "    y_test = y_test.astype('float16')\n",
    "\n",
    "    leng = X.shape[1] # if you work in the FD, this is the height of the sample time-frequency image, othewise EEG channels\n",
    "    chan = X.shape[2] # if you work in the FD, this is the width of the sample time-frequency image, othewise time samples of EEG signal\n",
    "\n",
    "    if len(X.shape)==3:\n",
    "        streams = 1 # this is EEG channels if you work with frequency domain, in the TD streams = 1\n",
    "    if len(X.shape)==4:\n",
    "        streams = X.shape[3] # this is EEG channels if you work with frequency domain, in the TD streams = 1\n",
    "    \n",
    "    return x_train, x_test, y_train, y_test, leng, chan, streams\n",
    "\n",
    "\n",
    "\n",
    "def conv_net(leng):\n",
    "    input_img = Input(shape=(leng, chan))  # adapt this if using `channels_first` image\n",
    "    input_norm = BatchNormalization(axis=1)(input_img)\n",
    "    e1 = Conv1D(N_KERNELS[0], KERNEL_SIZE[0], activation=ELU(alpha=1.0), padding='same',\n",
    "                                    kernel_initializer = orth(gain=1.0, seed=None))(input_norm)\n",
    "#     e2 = MaxPooling2D(POOL_SIZE, padding='same')(e1)\n",
    "    e2 = MaxPooling1D(pool_size=POOL_SIZE[0], strides=None, padding='valid')(e1)\n",
    "    e2b = BatchNormalization(axis=1)(e2)\n",
    "    e2a = Dropout(DROPOUT[0])(e2b)\n",
    "    e3 = Conv1D(N_KERNELS[1], KERNEL_SIZE[1], activation=ELU(alpha=1.0), padding='same',\n",
    "                                    kernel_initializer = orth(gain=1.0, seed=None))(e2a)\n",
    "#     e4 = MaxPooling2D(POOL_SIZE, padding='same')(e3)\n",
    "    e4 = MaxPooling1D(pool_size=POOL_SIZE[1], strides=None, padding='valid')(e3)\n",
    "    e4b = BatchNormalization(axis=1)(e4)\n",
    "    e4a = Dropout(DROPOUT[1])(e4b)\n",
    "    e5 = Conv1D(N_KERNELS[2], KERNEL_SIZE[2], activation=ELU(alpha=1.0), padding='same',\n",
    "                                    kernel_initializer = orth(gain=1.0, seed=None))(e4a)\n",
    "#     e6 = MaxPooling2D(POOL_SIZE, padding='same')(e5)\n",
    "    e6 = MaxPooling1D(pool_size=POOL_SIZE[2], strides=None, padding='valid')(e5)\n",
    "    e6b = BatchNormalization(axis=1)(e6)\n",
    "    e6a = Dropout(DROPOUT[2])(e6b)\n",
    "    e7 = Conv1D(N_KERNELS[3], KERNEL_SIZE[3], activation=ELU(alpha=1.0), padding='same',\n",
    "                                    kernel_initializer = orth(gain=1.0, seed=None))(e6a)\n",
    "#     e8 = MaxPooling2D(POOL_SIZE, padding='same')(e7)\n",
    "    e8 = MaxPooling1D(pool_size=POOL_SIZE[3], strides=None, padding='valid')(e7)\n",
    "    e9 = Dropout(DROPOUT[3])(e8)\n",
    "    e9b = Conv1D(N_KERNELS[4], KERNEL_SIZE[4], activation=ELU(alpha=1.0), padding='same',\n",
    "                                    kernel_initializer = orth(gain=1.0, seed=None))(e9)\n",
    "    e10 = Flatten()(e9b)\n",
    "    e10b = BatchNormalization(axis=1)(e10)\n",
    "    e11 = Dense(DENSE_LEN, activation=ELU(alpha=1.0),\n",
    "                                     kernel_initializer = orth(gain=1.0, seed=None))(e10b)\n",
    "    e12 = Dropout(DROPOUT[4])(e11)\n",
    "    e12b = BatchNormalization(axis=1)(e12)\n",
    "    e13 = Dense(2, activation='softmax')(e12b)\n",
    "\n",
    "    return Model(input_img, e13)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "/home/amplifier/home/DATASETS/BCI_0101T_TD.mat\n",
      "/home/amplifier/home/DATASETS/BCI_0101T_TD.mat trial 0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/opt/conda/lib/python3.6/site-packages/keras/activations.py:115: UserWarning: Do not pass a layer instance (such as ELU) as the activation argument of another layer. Instead, advanced activation layers should be used just like any other layer in a model.\n",
      "  identifier=identifier.__class__.__name__))\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy on the test data: 0.666666666667\n",
      "Accuracy on the training data: 1.0\n",
      "/home/amplifier/home/DATASETS/BCI_0101T_TD.mat\n",
      "/home/amplifier/home/DATASETS/BCI_0101T_TD.mat trial 1\n",
      "Accuracy on the test data: 0.708333333333\n",
      "Accuracy on the training data: 1.0\n",
      "/home/amplifier/home/DATASETS/BCI_0101T_TD.mat\n",
      "/home/amplifier/home/DATASETS/BCI_0101T_TD.mat trial 2\n",
      "Accuracy on the test data: 0.791666666667\n",
      "Accuracy on the training data: 1.0\n",
      "/home/amplifier/home/DATASETS/BCI_0101T_TD.mat\n",
      "/home/amplifier/home/DATASETS/BCI_0101T_TD.mat trial 3\n",
      "Accuracy on the test data: 0.708333333333\n",
      "Accuracy on the training data: 1.0\n",
      "/home/amplifier/home/DATASETS/BCI_0101T_TD.mat\n",
      "/home/amplifier/home/DATASETS/BCI_0101T_TD.mat trial 4\n",
      "Accuracy on the test data: 0.875\n",
      "Accuracy on the training data: 1.0\n",
      "/home/amplifier/home/DATASETS/BCI_0202T_TD.mat\n",
      "/home/amplifier/home/DATASETS/BCI_0202T_TD.mat trial 0\n",
      "Accuracy on the test data: 0.625\n",
      "Accuracy on the training data: 1.0\n",
      "/home/amplifier/home/DATASETS/BCI_0202T_TD.mat\n",
      "/home/amplifier/home/DATASETS/BCI_0202T_TD.mat trial 1\n",
      "Accuracy on the test data: 0.666666666667\n",
      "Accuracy on the training data: 1.0\n",
      "/home/amplifier/home/DATASETS/BCI_0202T_TD.mat\n",
      "/home/amplifier/home/DATASETS/BCI_0202T_TD.mat trial 2\n",
      "Accuracy on the test data: 0.916666666667\n",
      "Accuracy on the training data: 1.0\n",
      "/home/amplifier/home/DATASETS/BCI_0202T_TD.mat\n",
      "/home/amplifier/home/DATASETS/BCI_0202T_TD.mat trial 3\n",
      "Accuracy on the test data: 0.708333333333\n",
      "Accuracy on the training data: 1.0\n",
      "/home/amplifier/home/DATASETS/BCI_0202T_TD.mat\n",
      "/home/amplifier/home/DATASETS/BCI_0202T_TD.mat trial 4\n",
      "Accuracy on the test data: 0.541666666667\n",
      "Accuracy on the training data: 1.0\n",
      "/home/amplifier/home/DATASETS/BCI_0303T_TD.mat\n",
      "/home/amplifier/home/DATASETS/BCI_0303T_TD.mat trial 0\n",
      "Accuracy on the test data: 0.8125\n",
      "Accuracy on the training data: 1.0\n",
      "/home/amplifier/home/DATASETS/BCI_0303T_TD.mat\n",
      "/home/amplifier/home/DATASETS/BCI_0303T_TD.mat trial 1\n",
      "Accuracy on the test data: 0.78125\n",
      "Accuracy on the training data: 1.0\n",
      "/home/amplifier/home/DATASETS/BCI_0303T_TD.mat\n",
      "/home/amplifier/home/DATASETS/BCI_0303T_TD.mat trial 2\n",
      "Accuracy on the test data: 0.78125\n",
      "Accuracy on the training data: 1.0\n",
      "/home/amplifier/home/DATASETS/BCI_0303T_TD.mat\n",
      "/home/amplifier/home/DATASETS/BCI_0303T_TD.mat trial 3\n",
      "Accuracy on the test data: 0.75\n",
      "Accuracy on the training data: 1.0\n",
      "/home/amplifier/home/DATASETS/BCI_0303T_TD.mat\n",
      "/home/amplifier/home/DATASETS/BCI_0303T_TD.mat trial 4\n",
      "Accuracy on the test data: 0.84375\n",
      "Accuracy on the training data: 1.0\n",
      "/home/amplifier/home/DATASETS/BCI_0403T_TD.mat\n",
      "/home/amplifier/home/DATASETS/BCI_0403T_TD.mat trial 0\n",
      "Accuracy on the test data: 0.96875\n",
      "Accuracy on the training data: 1.0\n",
      "/home/amplifier/home/DATASETS/BCI_0403T_TD.mat\n",
      "/home/amplifier/home/DATASETS/BCI_0403T_TD.mat trial 1\n",
      "Accuracy on the test data: 1.0\n",
      "Accuracy on the training data: 1.0\n",
      "/home/amplifier/home/DATASETS/BCI_0403T_TD.mat\n",
      "/home/amplifier/home/DATASETS/BCI_0403T_TD.mat trial 2\n",
      "Accuracy on the test data: 1.0\n",
      "Accuracy on the training data: 1.0\n",
      "/home/amplifier/home/DATASETS/BCI_0403T_TD.mat\n",
      "/home/amplifier/home/DATASETS/BCI_0403T_TD.mat trial 3\n",
      "Accuracy on the test data: 0.96875\n",
      "Accuracy on the training data: 1.0\n",
      "/home/amplifier/home/DATASETS/BCI_0403T_TD.mat\n",
      "/home/amplifier/home/DATASETS/BCI_0403T_TD.mat trial 4\n",
      "Accuracy on the test data: 0.96875\n",
      "Accuracy on the training data: 1.0\n",
      "/home/amplifier/home/DATASETS/BCI_0503T_TD.mat\n",
      "/home/amplifier/home/DATASETS/BCI_0503T_TD.mat trial 0\n",
      "Accuracy on the test data: 0.625\n",
      "Accuracy on the training data: 1.0\n",
      "/home/amplifier/home/DATASETS/BCI_0503T_TD.mat\n",
      "/home/amplifier/home/DATASETS/BCI_0503T_TD.mat trial 1\n",
      "Accuracy on the test data: 0.59375\n",
      "Accuracy on the training data: 1.0\n",
      "/home/amplifier/home/DATASETS/BCI_0503T_TD.mat\n",
      "/home/amplifier/home/DATASETS/BCI_0503T_TD.mat trial 2\n",
      "Accuracy on the test data: 0.59375\n",
      "Accuracy on the training data: 1.0\n",
      "/home/amplifier/home/DATASETS/BCI_0503T_TD.mat\n",
      "/home/amplifier/home/DATASETS/BCI_0503T_TD.mat trial 3\n",
      "Accuracy on the test data: 0.40625\n",
      "Accuracy on the training data: 1.0\n",
      "/home/amplifier/home/DATASETS/BCI_0503T_TD.mat\n",
      "/home/amplifier/home/DATASETS/BCI_0503T_TD.mat trial 4\n",
      "Accuracy on the test data: 0.46875\n",
      "Accuracy on the training data: 1.0\n",
      "/home/amplifier/home/DATASETS/BCI_0603T_TD.mat\n",
      "/home/amplifier/home/DATASETS/BCI_0603T_TD.mat trial 0\n",
      "Accuracy on the test data: 0.6875\n",
      "Accuracy on the training data: 1.0\n",
      "/home/amplifier/home/DATASETS/BCI_0603T_TD.mat\n",
      "/home/amplifier/home/DATASETS/BCI_0603T_TD.mat trial 1\n",
      "Accuracy on the test data: 0.71875\n",
      "Accuracy on the training data: 1.0\n",
      "/home/amplifier/home/DATASETS/BCI_0603T_TD.mat\n",
      "/home/amplifier/home/DATASETS/BCI_0603T_TD.mat trial 2\n",
      "Accuracy on the test data: 0.75\n",
      "Accuracy on the training data: 1.0\n",
      "/home/amplifier/home/DATASETS/BCI_0603T_TD.mat\n",
      "/home/amplifier/home/DATASETS/BCI_0603T_TD.mat trial 3\n",
      "Accuracy on the test data: 0.625\n",
      "Accuracy on the training data: 1.0\n",
      "/home/amplifier/home/DATASETS/BCI_0603T_TD.mat\n",
      "/home/amplifier/home/DATASETS/BCI_0603T_TD.mat trial 4\n",
      "Accuracy on the test data: 0.71875\n",
      "Accuracy on the training data: 1.0\n",
      "/home/amplifier/home/DATASETS/BCI_0703T_TD.mat\n",
      "/home/amplifier/home/DATASETS/BCI_0703T_TD.mat trial 0\n",
      "Accuracy on the test data: 0.71875\n",
      "Accuracy on the training data: 1.0\n",
      "/home/amplifier/home/DATASETS/BCI_0703T_TD.mat\n",
      "/home/amplifier/home/DATASETS/BCI_0703T_TD.mat trial 1\n",
      "Accuracy on the test data: 0.78125\n",
      "Accuracy on the training data: 1.0\n",
      "/home/amplifier/home/DATASETS/BCI_0703T_TD.mat\n",
      "/home/amplifier/home/DATASETS/BCI_0703T_TD.mat trial 2\n",
      "Accuracy on the test data: 0.75\n",
      "Accuracy on the training data: 1.0\n",
      "/home/amplifier/home/DATASETS/BCI_0703T_TD.mat\n",
      "/home/amplifier/home/DATASETS/BCI_0703T_TD.mat trial 3\n",
      "Accuracy on the test data: 0.71875\n",
      "Accuracy on the training data: 1.0\n",
      "/home/amplifier/home/DATASETS/BCI_0703T_TD.mat\n",
      "/home/amplifier/home/DATASETS/BCI_0703T_TD.mat trial 4\n",
      "Accuracy on the test data: 0.65625\n",
      "Accuracy on the training data: 1.0\n",
      "/home/amplifier/home/DATASETS/BCI_0801T_TD.mat\n",
      "/home/amplifier/home/DATASETS/BCI_0801T_TD.mat trial 0\n",
      "Accuracy on the test data: 0.5625\n",
      "Accuracy on the training data: 1.0\n",
      "/home/amplifier/home/DATASETS/BCI_0801T_TD.mat\n",
      "/home/amplifier/home/DATASETS/BCI_0801T_TD.mat trial 1\n",
      "Accuracy on the test data: 0.5625\n",
      "Accuracy on the training data: 1.0\n",
      "/home/amplifier/home/DATASETS/BCI_0801T_TD.mat\n",
      "/home/amplifier/home/DATASETS/BCI_0801T_TD.mat trial 2\n",
      "Accuracy on the test data: 0.5625\n",
      "Accuracy on the training data: 1.0\n",
      "/home/amplifier/home/DATASETS/BCI_0801T_TD.mat\n",
      "/home/amplifier/home/DATASETS/BCI_0801T_TD.mat trial 3\n",
      "Accuracy on the test data: 0.5625\n",
      "Accuracy on the training data: 1.0\n",
      "/home/amplifier/home/DATASETS/BCI_0801T_TD.mat\n",
      "/home/amplifier/home/DATASETS/BCI_0801T_TD.mat trial 4\n",
      "Accuracy on the test data: 0.5625\n",
      "Accuracy on the training data: 1.0\n"
     ]
    }
   ],
   "source": [
    "VAL_ACC = np.zeros([len(paths),5])\n",
    "\n",
    "for file in range(len(paths)):\n",
    "    for iter in range(5):\n",
    "        x_train, x_test, y_train, y_test, leng, chan, streams = load_file(paths[file])\n",
    "        print(paths[file], 'trial', iter)\n",
    "        \n",
    "        x = Input(shape=(chan, leng, streams))\n",
    "        model = conv_net(leng)\n",
    "        model.compile(loss='binary_crossentropy', optimizer='adadelta', metrics=['accuracy'])\n",
    "        early_stopping = EarlyStopping(monitor='val_loss', patience=100, mode='min')\n",
    "\n",
    "        train_history = model.fit(x_train, y_train,\n",
    "                    epochs=1000,\n",
    "                    verbose = 0,\n",
    "                    batch_size=BATCH_SIZE,\n",
    "                    shuffle=True,\n",
    "                    validation_data=(x_test, y_test),\n",
    "                    callbacks=[TensorBoard(log_dir='/home/amplifier/home/CNN_stim'), early_stopping])\n",
    "        decoded_train = model.predict(x_train)\n",
    "        decoded_test = model.predict(x_test)\n",
    "\n",
    "    #     loss = train_history.history['loss']\n",
    "    #     val_loss = train_history.history['val_loss']\n",
    "    #     plt.plot(loss)\n",
    "    #     plt.plot(val_loss)\n",
    "    #     plt.legend(['loss', 'val_loss'])\n",
    "    #     plt.title('AdaDelta')\n",
    "    #     plt.ylabel('loss')\n",
    "    #     plt.xlabel('epoch')\n",
    "    #     plt.show()\n",
    "\n",
    "        cor_test = []\n",
    "        L = x_test.shape\n",
    "        for sample_no in range(x_test.shape[0]):\n",
    "            pred = model.predict(np.expand_dims(x_test[sample_no],0))\n",
    "            if ((pred[0][0]<pred[0][1]) and (y_test[sample_no,0] < y_test[sample_no,1]) or\n",
    "               ((pred[0][0]>pred[0][1]) & (y_test[sample_no,0] > y_test[sample_no,1]))):\n",
    "                cor_test.append(1)\n",
    "            else:\n",
    "                cor_test.append(0)\n",
    "\n",
    "        cor_train = []\n",
    "        L = x_train.shape\n",
    "        for sample_no in range(x_train.shape[0]):\n",
    "            pred = model.predict(np.expand_dims(x_train[sample_no],0))\n",
    "            if ((pred[0][0]<pred[0][1]) and (y_train[sample_no,0] < y_train[sample_no,1]) or\n",
    "               ((pred[0][0]>pred[0][1]) and (y_train[sample_no,0] > y_train[sample_no,1]))):\n",
    "                cor_train.append(1)\n",
    "            else:\n",
    "                cor_train.append(0)\n",
    "\n",
    "        print('Accuracy on the test data:', np.mean(cor_test))\n",
    "        print('Accuracy on the training data:', np.mean(cor_train))\n",
    "        VAL_ACC[file,iter] = np.mean(cor_test)\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fed8773f588>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from scipy import stats\n",
    "\n",
    "SE = stats.sem(VAL_ACC[0:8,:], axis=1)\n",
    "M = np.mean(VAL_ACC[0:8,:], axis=1)\n",
    "\n",
    "x = list(range(1,9))\n",
    "plt.plot(x, M, 'k-')\n",
    "plt.fill_between(x, M-SE, M+SE)\n",
    "plt.xticks(x)\n",
    "plt.xlabel('DATASETS')\n",
    "plt.ylabel('VAL_ACC')\n",
    "plt.show()"
   ]
  }
 ],
 "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": "3.6.3"
  },
  "latex_envs": {
   "LaTeX_envs_menu_present": true,
   "autoclose": false,
   "autocomplete": true,
   "bibliofile": "biblio.bib",
   "cite_by": "apalike",
   "current_citInitial": 1,
   "eqLabelWithNumbers": true,
   "eqNumInitial": 1,
   "hotkeys": {
    "equation": "Ctrl-E",
    "itemize": "Ctrl-I"
   },
   "labels_anchors": false,
   "latex_user_defs": false,
   "report_style_numbering": false,
   "user_envs_cfg": false
  },
  "nbTranslate": {
   "displayLangs": [
    "*"
   ],
   "hotkey": "alt-t",
   "langInMainMenu": true,
   "sourceLang": "en",
   "targetLang": "fr",
   "useGoogleTranslate": true
  },
  "varInspector": {
   "cols": {
    "lenName": 16,
    "lenType": 16,
    "lenVar": 40
   },
   "kernels_config": {
    "python": {
     "delete_cmd_postfix": "",
     "delete_cmd_prefix": "del ",
     "library": "var_list.py",
     "varRefreshCmd": "print(var_dic_list())"
    },
    "r": {
     "delete_cmd_postfix": ") ",
     "delete_cmd_prefix": "rm(",
     "library": "var_list.r",
     "varRefreshCmd": "cat(var_dic_list()) "
    }
   },
   "types_to_exclude": [
    "module",
    "function",
    "builtin_function_or_method",
    "instance",
    "_Feature"
   ],
   "window_display": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}