Model_Multiplexing.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "SHUFFLE = False\n",
    "BATCH_SIZE = 20\n",
    "TEST_TRAIN = 0.25\n",
    "file = 'Merged456-197-289_ICA(-eyes)+AUDpreproc.mat, DS2=64Hz, FIR=2-30Hz, centnorm=1, step=2, win=2, TD, 1-93.mat' #0.80(EEGnet.v2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "# get the Dataset:\n",
    "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",
    "import pickle\n",
    "\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",
    "\n",
    "from keras.callbacks import EarlyStopping, ModelCheckpoint, TensorBoard\n",
    "from keras.models import load_model\n",
    "from EEGModels import EEGNet, ShallowConvNet, DeepConvNet\n",
    "\n",
    "\n",
    "def load_data(file, SHUFFLE, BATCH_SIZE, TEST_TRAIN):\n",
    "        \n",
    "    \n",
    "\n",
    "    path = '/home/amplifier/home/DATASETS/' + file\n",
    "    mat_contents = sio.loadmat(path)\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",
    "    print('Original data shape:', X.shape)\n",
    "    print('Original labels shape:', Y.shape)\n",
    "\n",
    "    # verify that the model REALLY finds a mapping between the input and the labels. If we get\n",
    "    # our accuracy by chance, then we should get the same accuracy on a permuted dataset:\n",
    "    # Y = np.random.permutation(Y)\n",
    "\n",
    "\n",
    "    # winsize = mat_contents['winsize']\n",
    "    # stepsize = mat_contents['stepsize']\n",
    "    # trial_len = mat_contents['trial_len']\n",
    "    # low_cutoff = mat_contents['low_cutoff']\n",
    "    # high_cutoff = mat_contents['high_cutoff']\n",
    "    # source = mat_contents['filein']\n",
    "\n",
    "    x_train, x_test, y_train, y_test = train_test_split(X, Y, test_size=TEST_TRAIN, shuffle=SHUFFLE)\n",
    "    print('Original data type:', x_train.dtype)\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",
    "    # 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",
    "    print('Normalized data type:', x_train.dtype)\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",
    "\n",
    "    print('test input shape', x_test.shape, \"Nomralized MEAN:\", np.mean(x_test), \"min\", np.min(x_test),\"max\", np.max(x_test))\n",
    "    print('train input shape', x_train.shape, \"Nomralized MEAN:\", np.mean(x_train), \"min\", np.min(x_train),\"max\", np.max(x_train))\n",
    "\n",
    "    print('test labels shape', y_test.shape, \"Nomralized MEAN:\", np.mean(y_test), \"min\", np.min(y_test),\"max\", np.max(y_test))\n",
    "    print('train labels shape', y_train.shape, \"Nomralized MEAN:\", np.mean(y_train), \"min\", np.min(y_train),\"max\", np.max(y_train))\n",
    "\n",
    "    # print('Window length', winsize)\n",
    "    # print('Step size:', stepsize)\n",
    "    # print('Trial length:', trial_len)\n",
    "    \n",
    "    return x_train, x_test, y_train, y_test\n",
    "\n",
    "def rightsize(x_train, x_test, y_train, y_test):\n",
    "    x_train = x_train.reshape(x_train.shape[0],1,x_train.shape[1],x_train.shape[2])\n",
    "    x_train = x_train.transpose(0,1,3,2)\n",
    "    x_test = x_test.reshape(x_test.shape[0],1,x_test.shape[1],x_test.shape[2])\n",
    "    x_test = x_test.transpose(0,1,3,2)\n",
    "#     print(x_train.shape,'\\n',\n",
    "#           x_test.shape,'\\n',\n",
    "#           y_train.shape,'\\n',\n",
    "#           y_test.shape)\n",
    "    return x_train, x_test, y_train, y_test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loading model 1 and weights...\n",
      "Original data shape: (450, 128, 60)\n",
      "Original labels shape: (450, 1)\n",
      "Original data type: float32\n",
      "Normalized data type: float16\n",
      "test input shape (113, 128, 60) Nomralized MEAN: -2.754e-05 min -4.816 max 5.66\n",
      "train input shape (337, 128, 60) Nomralized MEAN: -1.25e-05 min -16.14 max 19.11\n",
      "test labels shape (113, 2) Nomralized MEAN: 0.5 min 0.0 max 1.0\n",
      "train labels shape (337, 2) Nomralized MEAN: 0.5 min 0.0 max 1.0\n",
      "data loaded\n",
      "model and weights loaded\n",
      "m1_decoded_train.shape (337, 2)\n",
      "m1_decoded_test.shape (113, 2)\n",
      "\n",
      "===========================\n",
      "\n",
      "loading data...\n",
      "Original data shape: (450, 128, 60)\n",
      "Original labels shape: (450, 1)\n",
      "Original data type: float32\n",
      "Normalized data type: float16\n",
      "test input shape (113, 128, 60) Nomralized MEAN: -2.754e-05 min -4.816 max 5.66\n",
      "train input shape (337, 128, 60) Nomralized MEAN: -1.25e-05 min -16.14 max 19.11\n",
      "test labels shape (113, 2) Nomralized MEAN: 0.5 min 0.0 max 1.0\n",
      "train labels shape (337, 2) Nomralized MEAN: 0.5 min 0.0 max 1.0\n",
      "data loaded\n",
      "\n",
      "loading model 2 and weights...\n",
      "model and weights loaded\n",
      "m2_decoded_train.shape (337, 2)\n",
      "m2_decoded_test.shape (113, 2)\n",
      "\n",
      "===========================\n",
      "\n",
      "loading data for model 3...\n",
      "data loaded\n",
      "\n",
      "loading model 3 and weights...\n",
      "model and weights loaded\n",
      "m3_decoded_train.shape (337, 2)\n",
      "m3_decoded_test.shape (113, 2)\n",
      "\n",
      "===========================\n",
      "\n"
     ]
    }
   ],
   "source": [
    "test_labels = []\n",
    "print('loading model 1 and weights...')\n",
    "x_train, x_test, y_train, y_test = load_data(file, SHUFFLE, BATCH_SIZE, TEST_TRAIN)\n",
    "x_train, x_test, y_train, y_test = rightsize(x_train, x_test, y_train, y_test)\n",
    "test_labels.append(y_test) # keep the labels in the first dataset (in fact the data samples should represent the same EEG sections)\n",
    "print('data loaded')\n",
    "model = load_model('/home/amplifier/home/NEW_DL/models/EEGnet_NEWEST.h5')\n",
    "model.load_weights('/home/amplifier/home/NEW_DL/weights/EEGnet_NEWEST_wts.h5')\n",
    "print('model and weights loaded')\n",
    "\n",
    "m1_decoded_train = model.predict(x_train)\n",
    "m1_decoded_test = model.predict(x_test)\n",
    "print('m1_decoded_train.shape', m1_decoded_train.shape)\n",
    "print('m1_decoded_test.shape', m1_decoded_test.shape)\n",
    "print('\\n===========================\\n')\n",
    "\n",
    "print('loading data...')\n",
    "x_train, x_test, y_train, y_test = load_data(file, SHUFFLE, BATCH_SIZE, TEST_TRAIN)\n",
    "test_labels.append(y_test)\n",
    "print('data loaded\\n')\n",
    "\n",
    "print('loading model 2 and weights...')\n",
    "model = load_model('/home/amplifier/home/NEW_DL/models/EEGnet_CNN_Dense_Binary_Classifier_TD_1D.h5')\n",
    "model.load_weights('/home/amplifier/home/NEW_DL/weights/EEGnet_CNN_Dense_Binary_Classifier_TD_1D_wts.h5')\n",
    "print('model and weights loaded')\n",
    "# let's predict a TRAINING sammple:\n",
    "m2_decoded_train = model.predict(x_train)\n",
    "m2_decoded_test = model.predict(x_test)\n",
    "print('m2_decoded_train.shape', m2_decoded_train.shape)\n",
    "print('m2_decoded_test.shape', m2_decoded_test.shape)\n",
    "print('\\n===========================\\n')\n",
    "\n",
    "\n",
    "print('loading data for model 3...')\n",
    "with open('/home/amplifier/home/DATASETS/IMG_net_end_2end_data.pkl', 'rb') as f:  # Python 3: open(..., 'rb')\n",
    "    x_train, x_test, y_train, y_test = pickle.load(f)\n",
    "test_labels.append(y_test)\n",
    "print('data loaded\\n')\n",
    "print('loading model 3 and weights...')\n",
    "model = load_model('/home/amplifier/home/NEW_DL/models/IMG_net_2.h5')\n",
    "model.load_weights('/home/amplifier/home/NEW_DL/weights/IMG_net_2_wts.h5')\n",
    "print('model and weights loaded')\n",
    "# let's predict a TRAINING sammple:\n",
    "m3_decoded_train = model.predict(x_train)\n",
    "m3_decoded_test = model.predict(x_test)\n",
    "print('m3_decoded_train.shape', m3_decoded_train.shape)\n",
    "print('m3_decoded_test.shape', m3_decoded_test.shape)\n",
    "print('\\n===========================\\n')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(113, 2, 3)\n",
      "(113, 2)\n"
     ]
    }
   ],
   "source": [
    "# check if the labels are the same in all the datasets:\n",
    "if not np.array_equal(test_labels[0],test_labels[1]) or not np.array_equal(test_labels[0],test_labels[2]) or not np.array_equal(test_labels[1],test_labels[2]):\n",
    "    print('LABELS ARE NOT SAME!!!')\n",
    "            \n",
    "m1_decoded_test_hot = np.zeros_like(m1_decoded_test)\n",
    "for i in range(len(m1_decoded_test)):\n",
    "    m1_decoded_test_hot[i,0]=1 if m1_decoded_test[i,0]>=0.5 else 0\n",
    "    m1_decoded_test_hot[i,1]=1 if m1_decoded_test[i,1]>0.5 else 0\n",
    "\n",
    "m2_decoded_test_hot = np.zeros_like(m2_decoded_test)\n",
    "for i in range(len(m2_decoded_test)):\n",
    "    m2_decoded_test_hot[i,0]=1 if m2_decoded_test[i,0]>=0.5 else 0\n",
    "    m2_decoded_test_hot[i,1]=1 if m2_decoded_test[i,1]>0.5 else 0\n",
    "\n",
    "m3_decoded_test_hot = np.zeros_like(m3_decoded_test)\n",
    "for i in range(len(m3_decoded_test)):\n",
    "    m3_decoded_test_hot[i,0]=1 if m3_decoded_test[i,0]>=0.5 else 0\n",
    "    m3_decoded_test_hot[i,1]=1 if m3_decoded_test[i,1]>0.5 else 0\n",
    "    \n",
    "# compute the voted prediction:\n",
    "voting = np.stack([m1_decoded_test_hot, m2_decoded_test_hot, m3_decoded_test_hot], axis=2)\n",
    "print(voting.shape)\n",
    "voting = np.round(np.mean(voting, axis=2),0)\n",
    "print(voting.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "model 1 accuracy: 0.7964601769911505 \n",
      " model 2 accuracy: 0.7345132743362832 \n",
      " model 3 accuracy: 0.7256637168141593 \n",
      " voting accuracy: 0.8141592920353983\n"
     ]
    }
   ],
   "source": [
    "# calculate accuracy of the models and voting accuracy on the test set:\n",
    "hits_m1 = (m1_decoded_test_hot==test_labels[0])[:,0]*1\n",
    "hits_m2 = (m2_decoded_test_hot==test_labels[0])[:,0]*1\n",
    "hits_m3 = (m3_decoded_test_hot==test_labels[0])[:,0]*1\n",
    "hits_voting = (voting==test_labels[0])[:,0]*1\n",
    "\n",
    "\n",
    "print('model 1 accuracy:', np.mean(hits_m1), '\\n',\n",
    "      'model 2 accuracy:', np.mean(hits_m2), '\\n',\n",
    "      'model 3 accuracy:', np.mean(hits_m3), '\\n',\n",
    "      'voting accuracy:', np.mean(hits_voting))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[1. 0.] [1. 0.] [1. 0.]       |   | [0. 1.]\n",
      "[0. 1.] [1. 0.] [1. 0.] *     |   | [0. 1.]\n",
      "[0. 1.] [1. 0.] [0. 1.] *   * | * | [0. 1.]\n",
      "[0. 1.] [0. 1.] [1. 0.] * *   | * | [0. 1.]\n",
      "[1. 0.] [1. 0.] [1. 0.]       |   | [0. 1.]\n",
      "[1. 0.] [1. 0.] [1. 0.]       |   | [0. 1.]\n",
      "[1. 0.] [1. 0.] [1. 0.]       |   | [0. 1.]\n",
      "[0. 1.] [1. 0.] [0. 1.] *   * | * | [0. 1.]\n",
      "[1. 0.] [1. 0.] [0. 1.] * *   | * | [1. 0.]\n",
      "[1. 0.] [1. 0.] [1. 0.] * * * | * | [1. 0.]\n",
      "[1. 0.] [1. 0.] [1. 0.] * * * | * | [1. 0.]\n",
      "[1. 0.] [1. 0.] [1. 0.] * * * | * | [1. 0.]\n",
      "[1. 0.] [1. 0.] [1. 0.] * * * | * | [1. 0.]\n",
      "[1. 0.] [1. 0.] [1. 0.] * * * | * | [1. 0.]\n",
      "[1. 0.] [0. 1.] [1. 0.] *   * | * | [1. 0.]\n",
      "[1. 0.] [1. 0.] [1. 0.] * * * | * | [1. 0.]\n",
      "[1. 0.] [1. 0.] [1. 0.] * * * | * | [1. 0.]\n",
      "[1. 0.] [0. 1.] [1. 0.] *   * | * | [1. 0.]\n",
      "[1. 0.] [1. 0.] [1. 0.] * * * | * | [1. 0.]\n",
      "[1. 0.] [1. 0.] [1. 0.] * * * | * | [1. 0.]\n",
      "[1. 0.] [1. 0.] [1. 0.] * * * | * | [1. 0.]\n",
      "[1. 0.] [1. 0.] [1. 0.] * * * | * | [1. 0.]\n",
      "[1. 0.] [1. 0.] [1. 0.] * * * | * | [1. 0.]\n",
      "[1. 0.] [1. 0.] [1. 0.]       |   | [0. 1.]\n",
      "[0. 1.] [0. 1.] [1. 0.] * *   | * | [0. 1.]\n",
      "[0. 1.] [0. 1.] [0. 1.] * * * | * | [0. 1.]\n",
      "[0. 1.] [0. 1.] [0. 1.] * * * | * | [0. 1.]\n",
      "[0. 1.] [1. 0.] [1. 0.] *     |   | [0. 1.]\n",
      "[0. 1.] [1. 0.] [0. 1.] *   * | * | [0. 1.]\n",
      "[0. 1.] [1. 0.] [0. 1.] *   * | * | [0. 1.]\n",
      "[0. 1.] [1. 0.] [0. 1.] *   * | * | [0. 1.]\n",
      "[0. 1.] [0. 1.] [0. 1.] * * * | * | [0. 1.]\n",
      "[1. 0.] [0. 1.] [0. 1.]   * * | * | [0. 1.]\n",
      "[0. 1.] [0. 1.] [0. 1.] * * * | * | [0. 1.]\n",
      "[0. 1.] [1. 0.] [1. 0.] *     |   | [0. 1.]\n",
      "[0. 1.] [1. 0.] [0. 1.] *   * | * | [0. 1.]\n",
      "[0. 1.] [0. 1.] [0. 1.] * * * | * | [0. 1.]\n",
      "[0. 1.] [1. 0.] [0. 1.] *   * | * | [0. 1.]\n",
      "[0. 1.] [0. 1.] [1. 0.]     * |   | [1. 0.]\n",
      "[0. 1.] [1. 0.] [0. 1.]   *   |   | [1. 0.]\n",
      "[0. 1.] [1. 0.] [1. 0.]   * * | * | [1. 0.]\n",
      "[0. 1.] [1. 0.] [0. 1.]   *   |   | [1. 0.]\n",
      "[0. 1.] [1. 0.] [1. 0.]   * * | * | [1. 0.]\n",
      "[0. 1.] [1. 0.] [0. 1.]   *   |   | [1. 0.]\n",
      "[1. 0.] [1. 0.] [1. 0.] * * * | * | [1. 0.]\n",
      "[1. 0.] [1. 0.] [1. 0.] * * * | * | [1. 0.]\n",
      "[0. 1.] [1. 0.] [1. 0.]   * * | * | [1. 0.]\n",
      "[1. 0.] [1. 0.] [1. 0.] * * * | * | [1. 0.]\n",
      "[0. 1.] [1. 0.] [0. 1.]   *   |   | [1. 0.]\n",
      "[1. 0.] [1. 0.] [1. 0.] * * * | * | [1. 0.]\n",
      "[0. 1.] [1. 0.] [0. 1.]   *   |   | [1. 0.]\n",
      "[0. 1.] [1. 0.] [0. 1.]   *   |   | [1. 0.]\n",
      "[1. 0.] [1. 0.] [1. 0.] * * * | * | [1. 0.]\n",
      "[0. 1.] [1. 0.] [1. 0.] *     |   | [0. 1.]\n",
      "[0. 1.] [0. 1.] [0. 1.] * * * | * | [0. 1.]\n",
      "[0. 1.] [1. 0.] [0. 1.] *   * | * | [0. 1.]\n",
      "[1. 0.] [0. 1.] [0. 1.]   * * | * | [0. 1.]\n",
      "[0. 1.] [0. 1.] [1. 0.] * *   | * | [0. 1.]\n",
      "[0. 1.] [0. 1.] [1. 0.] * *   | * | [0. 1.]\n",
      "[0. 1.] [0. 1.] [0. 1.] * * * | * | [0. 1.]\n",
      "[0. 1.] [0. 1.] [0. 1.] * * * | * | [0. 1.]\n",
      "[0. 1.] [0. 1.] [0. 1.] * * * | * | [0. 1.]\n",
      "[1. 0.] [1. 0.] [0. 1.]     * |   | [0. 1.]\n",
      "[0. 1.] [1. 0.] [0. 1.] *   * | * | [0. 1.]\n",
      "[1. 0.] [0. 1.] [0. 1.]   * * | * | [0. 1.]\n",
      "[0. 1.] [0. 1.] [0. 1.] * * * | * | [0. 1.]\n",
      "[0. 1.] [0. 1.] [1. 0.] * *   | * | [0. 1.]\n",
      "[0. 1.] [0. 1.] [0. 1.] * * * | * | [0. 1.]\n",
      "[0. 1.] [1. 0.] [0. 1.]   *   |   | [1. 0.]\n",
      "[1. 0.] [1. 0.] [0. 1.] * *   | * | [1. 0.]\n",
      "[1. 0.] [1. 0.] [1. 0.] * * * | * | [1. 0.]\n",
      "[1. 0.] [0. 1.] [1. 0.] *   * | * | [1. 0.]\n",
      "[1. 0.] [0. 1.] [0. 1.] *     |   | [1. 0.]\n",
      "[1. 0.] [1. 0.] [1. 0.] * * * | * | [1. 0.]\n",
      "[1. 0.] [1. 0.] [1. 0.] * * * | * | [1. 0.]\n",
      "[1. 0.] [1. 0.] [1. 0.] * * * | * | [1. 0.]\n",
      "[1. 0.] [1. 0.] [1. 0.] * * * | * | [1. 0.]\n",
      "[1. 0.] [1. 0.] [0. 1.] * *   | * | [1. 0.]\n",
      "[0. 1.] [1. 0.] [1. 0.]   * * | * | [1. 0.]\n",
      "[1. 0.] [1. 0.] [1. 0.] * * * | * | [1. 0.]\n",
      "[1. 0.] [1. 0.] [1. 0.] * * * | * | [1. 0.]\n",
      "[1. 0.] [1. 0.] [1. 0.] * * * | * | [1. 0.]\n",
      "[1. 0.] [1. 0.] [1. 0.] * * * | * | [1. 0.]\n",
      "[0. 1.] [1. 0.] [0. 1.] *   * | * | [0. 1.]\n",
      "[0. 1.] [1. 0.] [0. 1.] *   * | * | [0. 1.]\n",
      "[0. 1.] [0. 1.] [0. 1.] * * * | * | [0. 1.]\n",
      "[0. 1.] [0. 1.] [0. 1.] * * * | * | [0. 1.]\n",
      "[0. 1.] [0. 1.] [0. 1.] * * * | * | [0. 1.]\n",
      "[0. 1.] [0. 1.] [0. 1.] * * * | * | [0. 1.]\n",
      "[0. 1.] [0. 1.] [0. 1.] * * * | * | [0. 1.]\n",
      "[0. 1.] [0. 1.] [0. 1.] * * * | * | [0. 1.]\n",
      "[0. 1.] [0. 1.] [0. 1.] * * * | * | [0. 1.]\n",
      "[0. 1.] [0. 1.] [0. 1.] * * * | * | [0. 1.]\n",
      "[0. 1.] [0. 1.] [0. 1.] * * * | * | [0. 1.]\n",
      "[0. 1.] [1. 0.] [1. 0.] *     |   | [0. 1.]\n",
      "[0. 1.] [0. 1.] [1. 0.] * *   | * | [0. 1.]\n",
      "[0. 1.] [1. 0.] [0. 1.] *   * | * | [0. 1.]\n",
      "[0. 1.] [1. 0.] [0. 1.] *   * | * | [0. 1.]\n",
      "[0. 1.] [1. 0.] [1. 0.]   * * | * | [1. 0.]\n",
      "[1. 0.] [1. 0.] [0. 1.] * *   | * | [1. 0.]\n",
      "[1. 0.] [0. 1.] [1. 0.] *   * | * | [1. 0.]\n",
      "[1. 0.] [1. 0.] [0. 1.] * *   | * | [1. 0.]\n",
      "[1. 0.] [1. 0.] [1. 0.] * * * | * | [1. 0.]\n",
      "[1. 0.] [1. 0.] [1. 0.] * * * | * | [1. 0.]\n",
      "[0. 1.] [1. 0.] [0. 1.]   *   |   | [1. 0.]\n",
      "[1. 0.] [1. 0.] [1. 0.] * * * | * | [1. 0.]\n",
      "[1. 0.] [1. 0.] [0. 1.] * *   | * | [1. 0.]\n",
      "[1. 0.] [1. 0.] [1. 0.] * * * | * | [1. 0.]\n",
      "[1. 0.] [1. 0.] [1. 0.] * * * | * | [1. 0.]\n",
      "[1. 0.] [1. 0.] [1. 0.] * * * | * | [1. 0.]\n",
      "[1. 0.] [1. 0.] [1. 0.] * * * | * | [1. 0.]\n",
      "[1. 0.] [1. 0.] [1. 0.] * * * | * | [1. 0.]\n",
      "[1. 0.] [1. 0.] [1. 0.] * * * | * | [1. 0.]\n"
     ]
    }
   ],
   "source": [
    "# look closer at the predictions\n",
    "for i in range(len(test_labels[0])):\n",
    "    print(m1_decoded_test_hot[i,:],\n",
    "          m2_decoded_test_hot[i,:],\n",
    "          m3_decoded_test_hot[i,:],\n",
    "          '*' if np.array_equal(m1_decoded_test_hot[i,:], test_labels[0][i,:]) else ' ',\n",
    "          '*' if np.array_equal(m2_decoded_test_hot[i,:], test_labels[0][i,:]) else ' ',\n",
    "          '*' if np.array_equal(m3_decoded_test_hot[i,:], test_labels[0][i,:]) else ' ',\n",
    "          '|',\n",
    "          '*' if np.array_equal(voting[i,:], test_labels[0][i,:]) else ' ',\n",
    "          '|',\n",
    "          test_labels[0][i,:])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0.       , 0.       , 0.       , 0.       , 0.       , 0.       ,\n",
       "        0.       , 0.       , 0.       , 0.       , 0.       , 0.       ,\n",
       "        0.       , 0.       , 0.       ],\n",
       "       [0.       , 0.       , 0.       , 0.       , 0.       , 0.       ,\n",
       "        0.       , 0.       , 0.       , 0.       , 0.       , 0.       ,\n",
       "        0.       , 0.       , 0.       ],\n",
       "       [0.       , 0.       , 0.       , 0.       , 0.       , 6.9492188,\n",
       "        7.015625 , 7.0390625, 6.9921875, 6.6484375, 0.       , 0.       ,\n",
       "        0.       , 0.       , 0.       ],\n",
       "       [0.       , 0.       , 0.       , 6.7695312, 7.1601562, 7.3554688,\n",
       "        7.2773438, 7.25     , 7.0195312, 6.5742188, 6.0078125, 5.4804688,\n",
       "        0.       , 0.       , 0.       ],\n",
       "       [0.       , 0.       , 6.4804688, 7.1484375, 7.5742188, 7.5546875,\n",
       "        7.3789062, 7.2304688, 6.921875 , 6.515625 , 6.2382812, 5.921875 ,\n",
       "        5.484375 , 0.       , 0.       ],\n",
       "       [0.       , 5.7382812, 6.3945312, 7.5117188, 7.8359375, 7.359375 ,\n",
       "        6.7929688, 6.9960938, 6.625    , 6.3085938, 6.3710938, 6.5      ,\n",
       "        6.3242188, 5.9101562, 0.       ],\n",
       "       [0.       , 5.3359375, 6.1953125, 7.34375  , 8.40625  , 7.9960938,\n",
       "        6.8125   , 6.9414062, 6.7070312, 6.21875  , 6.1132812, 6.296875 ,\n",
       "        6.59375  , 6.4492188, 0.       ],\n",
       "       [0.       , 5.2070312, 5.40625  , 6.3789062, 8.8984375, 8.5390625,\n",
       "        7.1640625, 7.2617188, 7.5507812, 6.859375 , 6.15625  , 6.1953125,\n",
       "        6.7890625, 6.8515625, 0.       ],\n",
       "       [0.       , 5.6015625, 5.6914062, 5.8984375, 7.3671875, 7.8710938,\n",
       "        7.3515625, 7.8515625, 8.3515625, 7.8476562, 7.28125  , 7.3632812,\n",
       "        7.7070312, 6.9179688, 0.       ],\n",
       "       [0.       , 5.1835938, 5.       , 5.828125 , 6.828125 , 7.4296875,\n",
       "        7.5351562, 8.2734375, 8.328125 , 8.0625   , 7.2851562, 6.4453125,\n",
       "        7.0820312, 6.5429688, 0.       ],\n",
       "       [0.       , 4.4257812, 4.3007812, 6.4335938, 7.6289062, 7.3320312,\n",
       "        7.421875 , 8.2421875, 7.0351562, 7.3515625, 5.546875 , 4.4609375,\n",
       "        6.984375 , 6.5585938, 0.       ],\n",
       "       [0.       , 5.0546875, 4.8164062, 5.8554688, 6.9492188, 7.0585938,\n",
       "        7.3710938, 7.6796875, 6.734375 , 6.9765625, 6.2851562, 6.28125  ,\n",
       "        7.0859375, 6.8867188, 0.       ],\n",
       "       [0.       , 0.       , 0.       , 0.       , 6.7617188, 7.1015625,\n",
       "        7.2539062, 6.890625 , 6.5351562, 6.578125 , 6.65625  , 0.       ,\n",
       "        0.       , 0.       , 0.       ],\n",
       "       [0.       , 0.       , 0.       , 0.       , 0.       , 0.       ,\n",
       "        0.       , 0.       , 0.       , 0.       , 0.       , 0.       ,\n",
       "        0.       , 0.       , 0.       ],\n",
       "       [0.       , 0.       , 0.       , 0.       , 0.       , 0.       ,\n",
       "        0.       , 0.       , 0.       , 0.       , 0.       , 0.       ,\n",
       "        0.       , 0.       , 0.       ]], dtype=float32)"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_train[1,:,:,1].astype('float32')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7fe5e4d58dd8>"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvhp/UCwAAD2dJREFUeJzt3WuMXdV5xvH/47ky9oDtggFfAJsQKEkqgSwuSUWjulCHIpwPUWXUtC5EQlGVFnpRYoTUSP3UNBW9qFEjC9ISFUFaAgVFkGCRoDZScYONwRADnlAuDsY2mNr4gu2Zefvh7ImG4Yx9WPsyZ7qenzQ6t728Xu8zz6xz9jlrL0UEZpafOTNdgJnNDIffLFMOv1mmHH6zTDn8Zply+M0y5fCbZcrhN8uUw2+Wqd4mO+vXQAwyt8kuzbLyHoc4FkfVybaNhn+QuVyuVU12aZaVTfF4x9v6Zb9Zphx+s0yVCr+k1ZJelDQiaX1VRZlZ/ZLDL6kH+AbwGeBi4AZJF1dVmJnVq8zIfxkwEhEvR8Qx4D5gTTVlmVndyoR/CfD6pNs7i/vMbBYo81Ffu88SP3BaIEk3AzcDDDJUojszq1KZkX8nsGzS7aXAG1M3iogNEbEyIlb2MVCiOzOrUpnw/wS4QNJySf3AWuDhasoys7olv+yPiFFJXwJ+APQA34qI5yurzMxqVerrvRHxCPBIRbWYWYP8DT+zTDn8Zply+M0y5fCbZcrhN8uUw2+WKYffLFMOv1mmHH6zTDn8Zply+M0y5fCbZcrhN8uUw2+WKYffLFMOv1mmGl2rz07ulzenPSV9Gkvuc44+cN7VjoxHR+tBtrX76HBauysPJPdp7+eR3yxTDr9Zphx+s0yVWatvmaQfSdou6XlJt1RZmJnVq8wBv1HgTyNii6RhYLOkjRHx04pqM7MaJY/8EbErIrYU198FtuO1+sxmjUre80s6D7gE2FTFv2dm9Sv9Ob+kecB3gVsj4gMfwnqhTrPuVGrkl9RHK/j3RMQD7bbxQp1m3anM0X4BdwHbI+KO6koysyaUGfk/Bfwu8OuSthY/11ZUl5nVrMwqvT8G0r/cbWYzyt/wM8uUw2+WKU/pPYFzNs1Nard48H+T+zytJ23K6vHoSe5zLNLGgHdG0z+6HR1Pq/fg91ck9/n2gbTn89zf3pbcZzfzyG+WKYffLFMOv1mmHH6zTDn8Zply+M0y5fCbZcrhN8uUw2+WKYffLFMOv1mmHH6zTDn8Zpn6fz+rb/g/T09uu3hwV1K7c/rfTu4z1bvjg8ltD4+lnVuxzEzC0cSZhPP6jiX3OTYvrc9X//UTyX1284xAj/xmmXL4zTLl8JtlqnT4JfVIelrS96ooyMyaUcXIfwutdfrMbBYpu2LPUuC3gDurKcfMmlJ25P9b4MvAeAW1mFmDyizXdR2wJyI2n2S7myU9Jemp4xxN7c7MKlZ2ua7rJb0C3Edr2a5/mbqRF+o0607J4Y+I2yJiaUScB6wFfhgRn6+sMjOrlT/nN8tUJd/tj4gngCeq+LfMrBke+c0y5fCbZWpWTOnd89BFyW0HxtMXzVzS/05SuzN60xbbBBg5elZSuxcPn5nc56m97yW1e+vovOQ+Dx5P++RnoHc0uc+l/Wm/CwtPOZzc575HPpLU7rRrR5L77JRHfrNMOfxmmXL4zTLl8JtlyuE3y5TDb5Yph98sUw6/WaYcfrNMOfxmmXL4zTLl8JtlyuE3y9SsmNV38FD6IpRH5vWl9zuW1u/HhvYk93nqnLQZdmVm9aUuuNk3Zyy9z/G0Pg8f70/ucyhxkc/5/UeS+xzsOZ7U7lByj53zyG+WKYffLFMOv1mmyi7XNV/S/ZJekLRd0pVVFWZm9Sp7wO/vgO9HxOck9QNDFdRkZg1IDr+kU4GrgN8HiIhjQNrhVDNrXJmX/SuAvcA/SXpa0p2S5lZUl5nVrEz4e4FLgX+MiEtofTS5fupGXqjTrDuVCf9OYGdEbCpu30/rj8H7eKFOs+5UZqHON4HXJV1Y3LUK+GklVZlZ7coe7f9D4J7iSP/LwI3lSzKzJpQKf0RsBVZWVIuZNcjf8DPLlMNvlqlZMaV3fEzJbV/duyC57RO9H01q9ycLX07uc0hpi3wuHtif3Gef0qbmfmJoZ3KfT/ecm9Ru+/4SU5cTpxEP96VNswZY2D+e1G5Hco+d88hvlimH3yxTDr9Zphx+s0w5/GaZcvjNMuXwm2XK4TfLlMNvlimH3yxTDr9Zphx+s0w5/GaZmhWz+k6Zm35G8Dlz0mZVAbx5cDip3R37ViT3eeHgG0nthnvSZ559ZODNpHYX9e9N7vOM3rTZi8sG9yX3mTp7cXHfO8l9Pnnw/OS2dfPIb5Yph98sUw6/WabKLtT5x5Kel/ScpHslDVZVmJnVKzn8kpYAfwSsjIiPAz3A2qoKM7N6lX3Z3wucIqmX1gq9aYeqzaxxZVbs+Tnw18BrwC5gf0Q8VlVhZlavMi/7FwBrgOXAYmCupM+32c4LdZp1oTIv+38D+J+I2BsRx4EHgE9O3cgLdZp1pzLhfw24QtKQJNFaqHN7NWWZWd3KvOffRGtZ7i3AtuLf2lBRXWZWs7ILdX4V+GpFtZhZg/wNP7NMOfxmmZoVU3qPHOpPbjt+LG1xRoC+RWnTTkeOLErvM3Ha6bwSU3rP73s7qd1YpC+gOr/ncHLbVD/elza99qLh3cl9vvneqYkt06cud8ojv1mmHH6zTDn8Zply+M0y5fCbZcrhN8uUw2+WKYffLFMOv1mmHH6zTDn8Zply+M0y5fCbZWpWzOpjb/q5/3pH02eenXtB2gKNVwyPJPe5e/S0tIbp65EyRto++vY7Vyb3eWZf2ozJu1+8PLnPo6/PS2q3/Kq0WY/dziO/WaYcfrNMOfxmmTpp+CV9S9IeSc9Num+hpI2SdhSXC+ot08yq1snI/8/A6in3rQcej4gLgMeL22Y2i5w0/BHxH3zwhGJrgLuL63cDn624LjOrWep7/jMjYhdAcZl+xkozmxG1f84v6WbgZoBBhuruzsw6lDry75Z0NkBxuWe6Db1Qp1l3Sg3/w8C64vo64KFqyjGzpnTyUd+9wH8BF0raKekLwF8CV0vaAVxd3DazWeSk7/kj4oZpHlpVcS1m1iB/w88sUw6/WaZmxZTe8QXH09seTV+o8/UDad9a3jL3vOQ+h+YcS2q3qD9tiizAtqNLktq9dTRtiizA6HjauHPOwrRp1gAkth2YM5rc5aKBd5Palfhfdswjv1mmHH6zTDn8Zply+M0y5fCbZcrhN8uUw2+WKYffLFMOv1mmHH6zTDn8Zply+M0y5fCbZWpWzOr76I2bZ7qED+XQprnJbecNHE1qd3gs/fyIm46ckdTu0Fh/cp9HxvqS2i06JW2WHMDuK9NmPm4uNUamz0itm0d+s0w5/GaZcvjNMpW6UOfXJb0g6VlJD0qaX2+ZZla11IU6NwIfj4hfAV4Cbqu4LjOrWdJCnRHxWERMnNjsSWBpDbWZWY2qeM9/E/BoBf+OmTWo1Of8km4HRoF7TrCNF+o060LJ4Ze0DrgOWBURMd12EbEB2ABwqhZOu52ZNSsp/JJWA18Bfi0iDldbkpk1IXWhzn8AhoGNkrZK+mbNdZpZxVIX6ryrhlrMrEH+hp9Zphx+s0zNiim9s81rlx9Kb0vqwqKnJPeZPu1038k3sa7lkd8sUw6/WaYcfrNMOfxmmXL4zTLl8JtlyuE3y5TDb5Yph98sUw6/WaYcfrNMOfxmmXL4zTLl8JtlyuE3y5TDb5Yph98sU0kLdU567M8khaTT6ynPzOqSulAnkpYBVwOvVVyTmTUgaaHOwt8AXwa8Co/ZLJT0nl/S9cDPI+KZiusxs4Z86LP3ShoCbgeu6XB7L9Rp1oVSRv7zgeXAM5JeAZYCWySd1W7jiNgQESsjYmUfA+mVmlmlPvTIHxHbgEUTt4s/ACsj4q0K6zKzmqUu1Glms1zqQp2THz+vsmrMrDH+hp9Zphx+s0wpornv6EjaC7w6zcOnA9100LDb6oHuq8n1nNhM1HNuRJzRyYaNhv9EJD0VEStnuo4J3VYPdF9NrufEuq2eqfyy3yxTDr9Zprop/BtmuoApuq0e6L6aXM+JdVs979M17/nNrFndNPKbWYMaD7+k1ZJelDQiaX2bxwckfad4fJOk82qsZZmkH0naLul5Sbe02ebTkvZL2lr8/Hld9Uzq8xVJ24r+nmrzuCT9fbGPnpV0aY21XDjp/75V0gFJt07ZptZ91O5sUpIWStooaUdxuWCatuuKbXZIWldjPV+X9ELxfDwoaf40bU/43DYqIhr7AXqAnwErgH7gGeDiKdv8AfDN4vpa4Ds11nM2cGlxfRh4qU09nwa+1/B+egU4/QSPXws8Cgi4AtjU4PP3Jq3PkhvbR8BVwKXAc5Pu+ytgfXF9PfC1Nu0WAi8XlwuK6wtqqucaoLe4/rV29XTy3Db50/TIfxkwEhEvR8Qx4D5gzZRt1gB3F9fvB1ZJUh3FRMSuiNhSXH8X2A4sqaOviq0Bvh0tTwLzJZ3dQL+rgJ9FxHRf1KpFtD+b1OTfk7uBz7Zp+pvAxojYFxHvABtpc0q6KuqJiMciYrS4+SStqe5drenwLwFen3R7Jx8M2y+2KXbmfuCX6i6seHtxCbCpzcNXSnpG0qOSPlZ3LbROjfaYpM3FyVCm6mQ/1mEtcO80jzW9j86MiF3Q+iPOpGnmk8zUfrqJ1iuzdk723DbmQ8/nL6ndCD7144ZOtqmUpHnAd4FbI+LAlIe30HqZe1DStcC/AxfUWQ/wqYh4Q9IiYKOkF4rR5hclt2lT9z7qB64Hbmvz8Ezso07MxH66HRgF7plmk5M9t41peuTfCSybdHsp8MZ020jqBU6j/QlEKyGpj1bw74mIB6Y+HhEHIuJgcf0RoK/uU5VHxBvF5R7gQVpvlybrZD9W7TPAlojYPfWBmdhHwO6JtzrF5Z422zS6n4oDitcBvxPFG/ypOnhuG9N0+H8CXCBpeTGSrAUenrLNw8DEUdnPAT+cbkeWVRxLuAvYHhF3TLPNWRPHHCRdRmufvV1HPUUfcyUNT1yndSBp6poJDwO/Vxz1vwLYP/ESuEY3MM1L/qb3UWHy78k64KE22/wAuEbSguLTgGuK+yonaTXwFeD6iDg8zTadPLfNafoII60j1S/ROup/e3HfX9DaaQCDwL8BI8B/AytqrOVXab0MfBbYWvxcC3wR+GKxzZeA52l9MvEk8Mma98+Koq9nin4n9tHkmgR8o9iH22idRq3OmoZohfm0Sfc1to9o/dHZBRynNZp/gdZxoMeBHcXlwmLblcCdk9reVPwujQA31ljPCK3jCxO/RxOfWC0GHjnRcztTP/6Gn1mm/A0/s0w5/GaZcvjNMuXwm2XK4TfLlMNvlimH3yxTDr9Zpv4Pr7RzC391D08AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "% matplotlib inline\n",
    "plt.imshow(x_train[1,:,:,1].astype('float32'))"
   ]
  },
  {
   "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": "3.6.6"
  },
  "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()) "
    }
   },
   "oldHeight": 123.18181799999999,
   "position": {
    "height": "506px",
    "left": "939.091px",
    "right": "20px",
    "top": "120px",
    "width": "340px"
   },
   "types_to_exclude": [
    "module",
    "function",
    "builtin_function_or_method",
    "instance",
    "_Feature"
   ],
   "varInspector_section_display": "block",
   "window_display": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}