IMU_Model_Example.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "import scipy.io\n",
    "\n",
    "# random seed.\n",
    "rand_seed = 2\n",
    "\n",
    "from numpy.random import seed\n",
    "seed(rand_seed)\n",
    "from tensorflow import set_random_seed\n",
    "set_random_seed(rand_seed)\n",
    "\n",
    "import keras\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Conv2D, MaxPooling2D, LSTM, Dense, Dropout, Flatten, Activation\n",
    "from keras.layers.core import Permute, Reshape\n",
    "from keras import backend as K"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "def one_hot_encoding(y_data):\n",
    "    Mapping=dict()\n",
    "    sub_dirs=['downstair','upstair','run','jump','walk','handwashing','exercise']\n",
    "\n",
    "    categories=10\n",
    "\n",
    "    count=0\n",
    "    for i in sub_dirs:\n",
    "        Mapping[i]=count\n",
    "        count=count+1\n",
    "\n",
    "    y_features2=[]\n",
    "    for i in range(len(y_data)):\n",
    "        Type=y_data[i]\n",
    "        lab=Mapping[Type]\n",
    "        y_features2.append(lab)\n",
    "\n",
    "    y_features=np.array(y_features2)\n",
    "    y_features=y_features.reshape(y_features.shape[0],1)\n",
    "    from keras.utils import to_categorical\n",
    "    y_features = to_categorical(y_features)\n",
    "\n",
    "    return y_features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "path='ADD_PATH_To_Data'\n",
    "\n",
    "\n",
    "def get_train_data(path=path):\n",
    "    Train_data=np.load(path+'train_5000.npz')\n",
    "    Features_imu=Train_data['arr_0']\n",
    "    Labels=Train_data['arr_1']\n",
    "    Features_sound=Train_data['arr_2']\n",
    "\n",
    "    Labels = one_hot_encoding(Labels)\n",
    "    Features_imu = Features_imu.reshape(Features_imu.shape[0],1, Features_imu.shape[1], Features_imu.shape[2]) \n",
    "\n",
    "    return Features_imu,Labels,Features_sound\n",
    "\n",
    "def get_valid_data(path=path):\n",
    "    Train_data=np.load(path+'valid_1000.npz')\n",
    "    Features_imu=Train_data['arr_0']\n",
    "    Labels=Train_data['arr_1']\n",
    "    Features_sound=Train_data['arr_2']\n",
    "\n",
    "    Labels = one_hot_encoding(Labels)\n",
    "    Features_imu = Features_imu.reshape(Features_imu.shape[0],1, Features_imu.shape[1], Features_imu.shape[2]) \n",
    "\n",
    "    return Features_imu,Labels,Features_sound\n",
    "\n",
    "def get_test_data(path=path):\n",
    "    Train_data=np.load(path+'test_1377.npz')\n",
    "    Features_imu=Train_data['arr_0']\n",
    "    Labels=Train_data['arr_1']\n",
    "    Features_sound=Train_data['arr_2']\n",
    "\n",
    "    Labels = one_hot_encoding(Labels)\n",
    "    Features_imu = Features_imu.reshape(Features_imu.shape[0],1, Features_imu.shape[1], Features_imu.shape[2]) \n",
    "\n",
    "    return Features_imu,Labels,Features_sound"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(5000, 1, 40, 12) (5000, 7) (5000, 193)\n",
      "Train Classes distribution:  [760. 994. 546. 780. 532. 732. 656.]\n",
      "(1000, 1, 40, 12) (1000, 7) (1000, 193)\n",
      "Valid Classes distribution:  [150. 188. 136. 141.  98. 157. 130.]\n",
      "(1377, 1, 40, 12) (1377, 7) (1377, 193)\n",
      "Test Classes distribution:  [219. 268. 124. 146. 143. 234. 243.]\n"
     ]
    }
   ],
   "source": [
    "Features_imu,Labels,Features_sound = get_train_data()\n",
    "print(Features_imu.shape, Labels.shape, Features_sound.shape)\n",
    "print('Train Classes distribution: ',np.sum(Labels, axis =0))\n",
    "\n",
    "Features_imu2,Labels2,Features_sound2 = get_valid_data()\n",
    "print(Features_imu2.shape, Labels2.shape, Features_sound2.shape)\n",
    "print('Valid Classes distribution: ',np.sum(Labels2, axis =0))\n",
    "\n",
    "Features_imu3,Labels3,Features_sound3 = get_test_data()\n",
    "print(Features_imu3.shape, Labels3.shape, Features_sound3.shape)\n",
    "print('Test Classes distribution: ',np.sum(Labels3, axis =0))\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# IMU Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# With dropout in the network\n",
    "\n",
    "num_feat_map = 32\n",
    "\n",
    "_, dim, win_len, _, = Features_imu.shape\n",
    "num_classes=7\n",
    "\n",
    "\n",
    "def full_imu_model():  \n",
    "    print('building the model ... ')        \n",
    "    model = Sequential()\n",
    "    model.add(Conv2D(num_feat_map, kernel_size=(1, 5),\n",
    "                 activation='relu',\n",
    "                  input_shape=(dim, win_len, 12),\n",
    "                 padding='same'))\n",
    "    model.add(MaxPooling2D(pool_size=(1, 2)))\n",
    "    model.add(Conv2D(num_feat_map, kernel_size=(1, 5), activation='relu',padding='same'))\n",
    "    model.add(MaxPooling2D(pool_size=(1, 2)))\n",
    "    model.add(Flatten())\n",
    "    model.add(Dense(256, name='DENSE_3'))\n",
    "    model.add(Activation('relu', name = 'ACT_3'))\n",
    "    model.add(Dropout(0.3))\n",
    "    model.add(Dense(128, name='DENSE_2'))\n",
    "    model.add(Activation('relu', name = 'ACT_2'))\n",
    "    model.add(Dense(64, name='DENSE_1'))\n",
    "    model.add(Activation('relu', name = 'ACT_1'))\n",
    "\n",
    "    model.add(Dense(num_classes, activation='softmax', name = 'OUTPUT'))\n",
    "    \n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "building the model ... \n"
     ]
    }
   ],
   "source": [
    "IMU_model = full_imu_model()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d_1 (Conv2D)            (None, 1, 40, 32)         1952      \n",
      "_________________________________________________________________\n",
      "max_pooling2d_1 (MaxPooling2 (None, 1, 20, 32)         0         \n",
      "_________________________________________________________________\n",
      "conv2d_2 (Conv2D)            (None, 1, 20, 32)         5152      \n",
      "_________________________________________________________________\n",
      "max_pooling2d_2 (MaxPooling2 (None, 1, 10, 32)         0         \n",
      "_________________________________________________________________\n",
      "flatten_1 (Flatten)          (None, 320)               0         \n",
      "_________________________________________________________________\n",
      "DENSE_3 (Dense)              (None, 256)               82176     \n",
      "_________________________________________________________________\n",
      "ACT_3 (Activation)           (None, 256)               0         \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 256)               0         \n",
      "_________________________________________________________________\n",
      "DENSE_2 (Dense)              (None, 128)               32896     \n",
      "_________________________________________________________________\n",
      "ACT_2 (Activation)           (None, 128)               0         \n",
      "_________________________________________________________________\n",
      "DENSE_1 (Dense)              (None, 64)                8256      \n",
      "_________________________________________________________________\n",
      "ACT_1 (Activation)           (None, 64)                0         \n",
      "_________________________________________________________________\n",
      "OUTPUT (Dense)               (None, 7)                 455       \n",
      "=================================================================\n",
      "Total params: 130,887\n",
      "Trainable params: 130,887\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "IMU_model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "model_path = '/home/nesl/Sandeep/Trained_models/IMU_Example'\n",
    "\n",
    "\n",
    "from keras.callbacks import ReduceLROnPlateau\n",
    "from keras.callbacks import Callback\n",
    "from keras.callbacks import ModelCheckpoint\n",
    "\n",
    "checkpointer = ModelCheckpoint(filepath=model_path, verbose=1,monitor='val_acc' ,save_best_only=True, save_weights_only=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 5000 samples, validate on 1000 samples\n",
      "Epoch 1/50\n",
      "5000/5000 [==============================] - 4s 778us/step - loss: 7.1730 - acc: 0.2948 - val_loss: 1.7285 - val_acc: 0.5480\n",
      "\n",
      "Epoch 00001: val_acc improved from -inf to 0.54800, saving model to /home/nesl/Sandeep/Trained_models/IMU_Example\n",
      "Epoch 2/50\n",
      "5000/5000 [==============================] - 2s 367us/step - loss: 1.4818 - acc: 0.5706 - val_loss: 0.9326 - val_acc: 0.6940\n",
      "\n",
      "Epoch 00002: val_acc improved from 0.54800 to 0.69400, saving model to /home/nesl/Sandeep/Trained_models/IMU_Example\n",
      "Epoch 3/50\n",
      "5000/5000 [==============================] - 2s 369us/step - loss: 0.8863 - acc: 0.7014 - val_loss: 0.6737 - val_acc: 0.7660\n",
      "\n",
      "Epoch 00003: val_acc improved from 0.69400 to 0.76600, saving model to /home/nesl/Sandeep/Trained_models/IMU_Example\n",
      "Epoch 4/50\n",
      "5000/5000 [==============================] - 2s 377us/step - loss: 0.7075 - acc: 0.7600 - val_loss: 0.5527 - val_acc: 0.8260\n",
      "\n",
      "Epoch 00004: val_acc improved from 0.76600 to 0.82600, saving model to /home/nesl/Sandeep/Trained_models/IMU_Example\n",
      "Epoch 5/50\n",
      "5000/5000 [==============================] - 2s 381us/step - loss: 0.5108 - acc: 0.8196 - val_loss: 0.4919 - val_acc: 0.8480\n",
      "\n",
      "Epoch 00005: val_acc improved from 0.82600 to 0.84800, saving model to /home/nesl/Sandeep/Trained_models/IMU_Example\n",
      "Epoch 6/50\n",
      "5000/5000 [==============================] - 2s 380us/step - loss: 0.4079 - acc: 0.8582 - val_loss: 0.4105 - val_acc: 0.8770\n",
      "\n",
      "Epoch 00006: val_acc improved from 0.84800 to 0.87700, saving model to /home/nesl/Sandeep/Trained_models/IMU_Example\n",
      "Epoch 7/50\n",
      "5000/5000 [==============================] - 2s 379us/step - loss: 0.3525 - acc: 0.8758 - val_loss: 0.3866 - val_acc: 0.8720\n",
      "\n",
      "Epoch 00007: val_acc did not improve from 0.87700\n",
      "Epoch 8/50\n",
      "5000/5000 [==============================] - 2s 381us/step - loss: 0.2782 - acc: 0.8962 - val_loss: 0.3512 - val_acc: 0.8960\n",
      "\n",
      "Epoch 00008: val_acc improved from 0.87700 to 0.89600, saving model to /home/nesl/Sandeep/Trained_models/IMU_Example\n",
      "Epoch 9/50\n",
      "5000/5000 [==============================] - 2s 375us/step - loss: 0.2436 - acc: 0.9144 - val_loss: 0.3313 - val_acc: 0.9110\n",
      "\n",
      "Epoch 00009: val_acc improved from 0.89600 to 0.91100, saving model to /home/nesl/Sandeep/Trained_models/IMU_Example\n",
      "Epoch 10/50\n",
      "5000/5000 [==============================] - 2s 371us/step - loss: 0.2328 - acc: 0.9144 - val_loss: 0.3093 - val_acc: 0.8970\n",
      "\n",
      "Epoch 00010: val_acc did not improve from 0.91100\n",
      "Epoch 11/50\n",
      "5000/5000 [==============================] - 2s 374us/step - loss: 0.1895 - acc: 0.9324 - val_loss: 0.3275 - val_acc: 0.9060\n",
      "\n",
      "Epoch 00011: val_acc did not improve from 0.91100\n",
      "Epoch 12/50\n",
      "5000/5000 [==============================] - 2s 375us/step - loss: 0.1689 - acc: 0.9418 - val_loss: 0.3287 - val_acc: 0.9090\n",
      "\n",
      "Epoch 00012: val_acc did not improve from 0.91100\n",
      "Epoch 13/50\n",
      "5000/5000 [==============================] - 2s 373us/step - loss: 0.1497 - acc: 0.9448 - val_loss: 0.3063 - val_acc: 0.9210\n",
      "\n",
      "Epoch 00013: val_acc improved from 0.91100 to 0.92100, saving model to /home/nesl/Sandeep/Trained_models/IMU_Example\n",
      "Epoch 14/50\n",
      "5000/5000 [==============================] - 2s 375us/step - loss: 0.1506 - acc: 0.9492 - val_loss: 0.3640 - val_acc: 0.9160\n",
      "\n",
      "Epoch 00014: val_acc did not improve from 0.92100\n",
      "Epoch 15/50\n",
      "5000/5000 [==============================] - 2s 379us/step - loss: 0.1956 - acc: 0.9370 - val_loss: 0.3261 - val_acc: 0.9250\n",
      "\n",
      "Epoch 00015: val_acc improved from 0.92100 to 0.92500, saving model to /home/nesl/Sandeep/Trained_models/IMU_Example\n",
      "Epoch 16/50\n",
      "5000/5000 [==============================] - 2s 371us/step - loss: 0.1266 - acc: 0.9546 - val_loss: 0.3187 - val_acc: 0.9290\n",
      "\n",
      "Epoch 00016: val_acc improved from 0.92500 to 0.92900, saving model to /home/nesl/Sandeep/Trained_models/IMU_Example\n",
      "Epoch 17/50\n",
      "5000/5000 [==============================] - 2s 363us/step - loss: 0.1093 - acc: 0.9590 - val_loss: 0.3072 - val_acc: 0.9300\n",
      "\n",
      "Epoch 00017: val_acc improved from 0.92900 to 0.93000, saving model to /home/nesl/Sandeep/Trained_models/IMU_Example\n",
      "Epoch 18/50\n",
      "5000/5000 [==============================] - 2s 368us/step - loss: 0.0929 - acc: 0.9660 - val_loss: 0.3289 - val_acc: 0.9290\n",
      "\n",
      "Epoch 00018: val_acc did not improve from 0.93000\n",
      "Epoch 19/50\n",
      "5000/5000 [==============================] - 2s 366us/step - loss: 0.0836 - acc: 0.9692 - val_loss: 0.3515 - val_acc: 0.9240\n",
      "\n",
      "Epoch 00019: val_acc did not improve from 0.93000\n",
      "Epoch 20/50\n",
      "5000/5000 [==============================] - 2s 372us/step - loss: 0.0821 - acc: 0.9694 - val_loss: 0.3086 - val_acc: 0.9310\n",
      "\n",
      "Epoch 00020: val_acc improved from 0.93000 to 0.93100, saving model to /home/nesl/Sandeep/Trained_models/IMU_Example\n",
      "Epoch 21/50\n",
      "5000/5000 [==============================] - 2s 368us/step - loss: 0.0778 - acc: 0.9736 - val_loss: 0.3271 - val_acc: 0.9330\n",
      "\n",
      "Epoch 00021: val_acc improved from 0.93100 to 0.93300, saving model to /home/nesl/Sandeep/Trained_models/IMU_Example\n",
      "Epoch 22/50\n",
      "5000/5000 [==============================] - 2s 367us/step - loss: 0.0717 - acc: 0.9782 - val_loss: 0.3152 - val_acc: 0.9310\n",
      "\n",
      "Epoch 00022: val_acc did not improve from 0.93300\n",
      "Epoch 23/50\n",
      "5000/5000 [==============================] - 2s 377us/step - loss: 0.0593 - acc: 0.9790 - val_loss: 0.3356 - val_acc: 0.9340\n",
      "\n",
      "Epoch 00023: val_acc improved from 0.93300 to 0.93400, saving model to /home/nesl/Sandeep/Trained_models/IMU_Example\n",
      "Epoch 24/50\n",
      "5000/5000 [==============================] - 2s 366us/step - loss: 0.0518 - acc: 0.9816 - val_loss: 0.3405 - val_acc: 0.9360\n",
      "\n",
      "Epoch 00024: val_acc improved from 0.93400 to 0.93600, saving model to /home/nesl/Sandeep/Trained_models/IMU_Example\n",
      "Epoch 25/50\n",
      "5000/5000 [==============================] - 2s 375us/step - loss: 0.0708 - acc: 0.9774 - val_loss: 0.3698 - val_acc: 0.9320\n",
      "\n",
      "Epoch 00025: val_acc did not improve from 0.93600\n",
      "Epoch 26/50\n",
      "5000/5000 [==============================] - 2s 365us/step - loss: 0.0825 - acc: 0.9758 - val_loss: 0.3779 - val_acc: 0.9260\n",
      "\n",
      "Epoch 00026: val_acc did not improve from 0.93600\n",
      "Epoch 27/50\n",
      "5000/5000 [==============================] - 2s 364us/step - loss: 0.0648 - acc: 0.9772 - val_loss: 0.3892 - val_acc: 0.9270\n",
      "\n",
      "Epoch 00027: val_acc did not improve from 0.93600\n",
      "Epoch 28/50\n",
      "5000/5000 [==============================] - 2s 372us/step - loss: 0.0637 - acc: 0.9780 - val_loss: 0.2924 - val_acc: 0.9380\n",
      "\n",
      "Epoch 00028: val_acc improved from 0.93600 to 0.93800, saving model to /home/nesl/Sandeep/Trained_models/IMU_Example\n",
      "Epoch 29/50\n",
      "5000/5000 [==============================] - 2s 372us/step - loss: 0.0563 - acc: 0.9818 - val_loss: 0.3104 - val_acc: 0.9390\n",
      "\n",
      "Epoch 00029: val_acc improved from 0.93800 to 0.93900, saving model to /home/nesl/Sandeep/Trained_models/IMU_Example\n",
      "Epoch 30/50\n",
      "5000/5000 [==============================] - 2s 373us/step - loss: 0.0533 - acc: 0.9830 - val_loss: 0.3488 - val_acc: 0.9390\n",
      "\n",
      "Epoch 00030: val_acc did not improve from 0.93900\n",
      "Epoch 31/50\n",
      "5000/5000 [==============================] - 2s 369us/step - loss: 0.1430 - acc: 0.9532 - val_loss: 0.2924 - val_acc: 0.9360\n",
      "\n",
      "Epoch 00031: val_acc did not improve from 0.93900\n",
      "Epoch 32/50\n",
      "5000/5000 [==============================] - 2s 371us/step - loss: 0.0813 - acc: 0.9768 - val_loss: 0.2682 - val_acc: 0.9440\n",
      "\n",
      "Epoch 00032: val_acc improved from 0.93900 to 0.94400, saving model to /home/nesl/Sandeep/Trained_models/IMU_Example\n",
      "Epoch 33/50\n",
      "5000/5000 [==============================] - 2s 374us/step - loss: 0.0509 - acc: 0.9858 - val_loss: 0.3065 - val_acc: 0.9350\n",
      "\n",
      "Epoch 00033: val_acc did not improve from 0.94400\n",
      "Epoch 34/50\n",
      "5000/5000 [==============================] - 2s 373us/step - loss: 0.0483 - acc: 0.9842 - val_loss: 0.2917 - val_acc: 0.9430\n",
      "\n",
      "Epoch 00034: val_acc did not improve from 0.94400\n",
      "Epoch 35/50\n",
      "5000/5000 [==============================] - 2s 371us/step - loss: 0.0373 - acc: 0.9874 - val_loss: 0.2978 - val_acc: 0.9420\n",
      "\n",
      "Epoch 00035: val_acc did not improve from 0.94400\n",
      "Epoch 36/50\n",
      "5000/5000 [==============================] - 2s 371us/step - loss: 0.0414 - acc: 0.9868 - val_loss: 0.3097 - val_acc: 0.9450\n",
      "\n",
      "Epoch 00036: val_acc improved from 0.94400 to 0.94500, saving model to /home/nesl/Sandeep/Trained_models/IMU_Example\n",
      "Epoch 37/50\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "5000/5000 [==============================] - 2s 371us/step - loss: 0.0446 - acc: 0.9862 - val_loss: 0.3315 - val_acc: 0.9360\n",
      "\n",
      "Epoch 00037: val_acc did not improve from 0.94500\n",
      "Epoch 38/50\n",
      "5000/5000 [==============================] - 2s 371us/step - loss: 0.0467 - acc: 0.9880 - val_loss: 0.3178 - val_acc: 0.9400\n",
      "\n",
      "Epoch 00038: val_acc did not improve from 0.94500\n",
      "Epoch 39/50\n",
      "5000/5000 [==============================] - 2s 371us/step - loss: 0.0410 - acc: 0.9884 - val_loss: 0.3011 - val_acc: 0.9440\n",
      "\n",
      "Epoch 00039: val_acc did not improve from 0.94500\n",
      "Epoch 40/50\n",
      "5000/5000 [==============================] - 2s 377us/step - loss: 0.0306 - acc: 0.9914 - val_loss: 0.3132 - val_acc: 0.9500\n",
      "\n",
      "Epoch 00040: val_acc improved from 0.94500 to 0.95000, saving model to /home/nesl/Sandeep/Trained_models/IMU_Example\n",
      "Epoch 41/50\n",
      "5000/5000 [==============================] - 2s 376us/step - loss: 0.0297 - acc: 0.9932 - val_loss: 0.3043 - val_acc: 0.9430\n",
      "\n",
      "Epoch 00041: val_acc did not improve from 0.95000\n",
      "Epoch 42/50\n",
      "5000/5000 [==============================] - 2s 380us/step - loss: 0.0370 - acc: 0.9892 - val_loss: 0.3681 - val_acc: 0.9380\n",
      "\n",
      "Epoch 00042: val_acc did not improve from 0.95000\n",
      "Epoch 43/50\n",
      "5000/5000 [==============================] - 2s 373us/step - loss: 0.0662 - acc: 0.9812 - val_loss: 0.3723 - val_acc: 0.9420\n",
      "\n",
      "Epoch 00043: val_acc did not improve from 0.95000\n",
      "Epoch 44/50\n",
      "5000/5000 [==============================] - 2s 372us/step - loss: 0.0687 - acc: 0.9800 - val_loss: 0.3172 - val_acc: 0.9380\n",
      "\n",
      "Epoch 00044: val_acc did not improve from 0.95000\n",
      "Epoch 45/50\n",
      "5000/5000 [==============================] - 2s 369us/step - loss: 0.0536 - acc: 0.9850 - val_loss: 0.3293 - val_acc: 0.9370\n",
      "\n",
      "Epoch 00045: val_acc did not improve from 0.95000\n",
      "Epoch 46/50\n",
      "5000/5000 [==============================] - 2s 374us/step - loss: 0.0404 - acc: 0.9866 - val_loss: 0.2879 - val_acc: 0.9440\n",
      "\n",
      "Epoch 00046: val_acc did not improve from 0.95000\n",
      "Epoch 47/50\n",
      "5000/5000 [==============================] - 2s 375us/step - loss: 0.0312 - acc: 0.9908 - val_loss: 0.2984 - val_acc: 0.9450\n",
      "\n",
      "Epoch 00047: val_acc did not improve from 0.95000\n",
      "Epoch 48/50\n",
      "5000/5000 [==============================] - 2s 375us/step - loss: 0.0307 - acc: 0.9912 - val_loss: 0.3275 - val_acc: 0.9400\n",
      "\n",
      "Epoch 00048: val_acc did not improve from 0.95000\n",
      "Epoch 49/50\n",
      "5000/5000 [==============================] - 2s 387us/step - loss: 0.0235 - acc: 0.9912 - val_loss: 0.2939 - val_acc: 0.9450\n",
      "\n",
      "Epoch 00049: val_acc did not improve from 0.95000\n",
      "Epoch 50/50\n",
      "5000/5000 [==============================] - 2s 370us/step - loss: 0.0373 - acc: 0.9894 - val_loss: 0.3405 - val_acc: 0.9420\n",
      "\n",
      "Epoch 00050: val_acc did not improve from 0.95000\n",
      "---Time 96.7810389995575 seconds ---\n"
     ]
    }
   ],
   "source": [
    "batch_size = 128\n",
    "epochs = 50\n",
    "from keras import optimizers\n",
    "from keras.optimizers import Adam\n",
    "\n",
    "adam = optimizers.Adam(lr=0.001, beta_1=0.9, beta_2=0.999, epsilon=None, decay=0.0, amsgrad=False)\n",
    "IMU_model.compile(loss=keras.losses.categorical_crossentropy,\n",
    "              optimizer= adam,\n",
    "              metrics=['accuracy'])\n",
    "\n",
    "import time\n",
    "start_time = time.time()\n",
    "\n",
    "learning_hist = IMU_model.fit(Features_imu, Labels,\n",
    "                    verbose=1,\n",
    "                    epochs=epochs,\n",
    "                        shuffle=True,\n",
    "                    batch_size=batch_size,\n",
    "                        callbacks=[checkpointer],\n",
    "                    validation_data=(Features_imu2, Labels2))\n",
    "\n",
    "end_time = time.time()\n",
    "\n",
    "print(\"---Time %s seconds ---\" % (end_time - start_time))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.models import load_model\n",
    "\n",
    "IMU_model = load_model(model_path)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "5000/5000 [==============================] - 2s 374us/step\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[0.009534283232755843, 0.9982]"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Train Accuracy\n",
    "IMU_model.evaluate(Features_imu,Labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1000/1000 [==============================] - 0s 359us/step\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[0.31315830117464066, 0.95]"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#valid Accuracy\n",
    "IMU_model.evaluate(Features_imu2,Labels2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1377/1377 [==============================] - 1s 371us/step\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[0.41778624388540914, 0.90559186637618]"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Test Accuracy\n",
    "IMU_model.evaluate(Features_imu3,Labels3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Plotting"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "dict_keys(['val_loss', 'val_acc', 'acc', 'loss'])\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from matplotlib import pyplot as plt\n",
    "\n",
    "print(learning_hist.history.keys())\n",
    "plt.rcParams.update({'font.size': 18})\n",
    "\n",
    "plt.figure(figsize=(10,5))\n",
    "\n",
    "plt.plot(learning_hist.history['acc'])\n",
    "plt.plot(learning_hist.history['val_acc'])\n",
    "plt.title('model accuracy')\n",
    "plt.ylabel('accuracy')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'Valid'], loc='upper right')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# summarize history for loss\n",
    "\n",
    "from matplotlib import pyplot as plt\n",
    "\n",
    "plt.figure(figsize=(10,5))\n",
    "plt.rcParams.update({'font.size': 18})\n",
    "\n",
    "\n",
    "plt.plot(learning_hist.history['loss'])\n",
    "plt.plot(learning_hist.history['val_loss'])\n",
    "plt.title('model loss')\n",
    "plt.ylabel('loss')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'Valid'], loc='upper right')\n",
    "plt.show()"
   ]
  },
  {
   "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": "3.5.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}