minimal matlab/cnn_eeg.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# zabiralnazi@yahoo.com"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# cnn benchmark"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Anaconda3\\lib\\site-packages\\h5py\\__init__.py:34: FutureWarning: Conversion of the second argument of issubdtype from `float` to `np.floating` is deprecated. In future, it will be treated as `np.float64 == np.dtype(float).type`.\n",
      "  from ._conv import register_converters as _register_converters\n"
     ]
    }
   ],
   "source": [
    "from scipy import io, signal\n",
    "import matplotlib.pyplot as plt\n",
    "import dtcwt\n",
    "import numpy as np\n",
    "import itertools\n",
    "import pywt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(5, 1024, 270)\n",
      "(270, 1)\n"
     ]
    }
   ],
   "source": [
    "# Load DWT-denoised time-series data\n",
    "SubA = io.loadmat('SubA_5chan_3LRF.mat')\n",
    "X = SubA['EEGDATA']\n",
    "y = SubA['LABELS']\n",
    "print(X.shape)\n",
    "print(y.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(270, 1024, 5)\n"
     ]
    }
   ],
   "source": [
    "X = np.swapaxes(X,0,2)\n",
    "print(X.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "y[y==1] = 0\n",
    "y[y==2] = 1\n",
    "y[y==3] = 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.utils import to_categorical\n",
    "\n",
    "y = to_categorical(y,3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[0. 1. 0.]\n",
      " [1. 0. 0.]\n",
      " [1. 0. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]]\n"
     ]
    }
   ],
   "source": [
    "print(y[0:5])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from keras.models import Sequential\n",
    "from keras.layers import Reshape, Conv1D, MaxPooling1D, GlobalAveragePooling1D, Dropout, Dense"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "temporal_dimension = X.shape[1]\n",
    "num_channels = X.shape[2]\n",
    "num_classes = 3"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "input_shape = temporal_dimension"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv1d_21 (Conv1D)           (None, 1015, 100)         5100      \n",
      "_________________________________________________________________\n",
      "conv1d_22 (Conv1D)           (None, 1006, 100)         100100    \n",
      "_________________________________________________________________\n",
      "max_pooling1d_6 (MaxPooling1 (None, 335, 100)          0         \n",
      "_________________________________________________________________\n",
      "conv1d_23 (Conv1D)           (None, 326, 160)          160160    \n",
      "_________________________________________________________________\n",
      "conv1d_24 (Conv1D)           (None, 317, 160)          256160    \n",
      "_________________________________________________________________\n",
      "global_average_pooling1d_6 ( (None, 160)               0         \n",
      "_________________________________________________________________\n",
      "dropout_5 (Dropout)          (None, 160)               0         \n",
      "_________________________________________________________________\n",
      "dense_7 (Dense)              (None, 3)                 483       \n",
      "=================================================================\n",
      "Total params: 522,003\n",
      "Trainable params: 522,003\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "model = Sequential()\n",
    "model.add(Conv1D(100, 10, activation='relu', input_shape=(temporal_dimension, num_channels)))\n",
    "model.add(Conv1D(100, 10, activation='relu'))\n",
    "model.add(MaxPooling1D(3))\n",
    "model.add(Conv1D(160, 10, activation='relu'))\n",
    "model.add(Conv1D(160, 10, activation='relu'))\n",
    "model.add(GlobalAveragePooling1D())\n",
    "#model.add(Dense(100, activation='relu'))\n",
    "model.add(Dropout(0.1))\n",
    "model.add(Dense(num_classes, activation='sigmoid'))\n",
    "print(model.summary())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 243 samples, validate on 27 samples\n",
      "Epoch 1/5\n",
      "243/243 [==============================] - 10s 41ms/step - loss: 0.6929 - acc: 0.5391 - val_loss: 0.6910 - val_acc: 0.6667\n",
      "Epoch 2/5\n",
      "243/243 [==============================] - 9s 37ms/step - loss: 0.6900 - acc: 0.6667 - val_loss: 0.6839 - val_acc: 0.6667\n",
      "Epoch 3/5\n",
      "243/243 [==============================] - 11s 45ms/step - loss: 0.6811 - acc: 0.6667 - val_loss: 0.6665 - val_acc: 0.6667\n",
      "Epoch 4/5\n",
      "243/243 [==============================] - 11s 44ms/step - loss: 0.6620 - acc: 0.6667 - val_loss: 0.6389 - val_acc: 0.6667\n",
      "Epoch 5/5\n",
      "243/243 [==============================] - 11s 46ms/step - loss: 0.6409 - acc: 0.6667 - val_loss: 0.6481 - val_acc: 0.6667\n"
     ]
    }
   ],
   "source": [
    "model.compile(loss='binary_crossentropy',\n",
    "                optimizer='adam', metrics=['accuracy'])\n",
    "\n",
    "BATCH_SIZE = 128\n",
    "EPOCHS = 5\n",
    "\n",
    "history = model.fit(X,\n",
    "                      y,\n",
    "                      batch_size=BATCH_SIZE,\n",
    "                      epochs=EPOCHS,\n",
    "                      validation_split=0.1,\n",
    "                      verbose=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# reference"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# https://blog.goodaudience.com/introduction-to-1d-convolutional-neural-networks-in-keras-for-time-sequences-3a7ff801a2cf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "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": "3.6.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}