In [1]:

    

import os
import keras
import numpy as np
import tensorflow as tf
from keras.datasets import cifar10
from keras.preprocessing.image import ImageDataGenerator
from keras.models import Sequential
from keras.layers import Dense, Dropout, Activation, Flatten
from keras.layers import Conv2D, MaxPooling2D, GlobalAveragePooling2D, AveragePooling2D
from keras.initializers import he_normal
from keras import optimizers
from keras.callbacks import LearningRateScheduler, TensorBoard
from keras.layers.normalization import BatchNormalization
from keras.utils.data_utils import get_file


    

    




Using TensorFlow backend.

In [2]:

    

os.environ["CUDA_VISIBLE_DEVICES"] = "3"
from keras.backend.tensorflow_backend import set_session
config = tf.ConfigProto()
config.gpu_options.per_process_gpu_memory_fraction = 0.5
set_session(tf.Session(config=config))


    

In [3]:

    

num_classes  = 10
batch_size   = 128
epochs       = 170
iterations   = 391
dropout      = 0.5
log_filepath = r'./vgg19_retrain_WHE/'


    

In [4]:

    

def color_preprocessing(x_train,x_test):
    x_train = x_train.astype('float32')
    x_test = x_test.astype('float32')
    # data preprocessing 
    x_train[:,:,:,0] = (x_train[:,:,:,0]-123.680)
    x_train[:,:,:,1] = (x_train[:,:,:,1]-116.779)
    x_train[:,:,:,2] = (x_train[:,:,:,2]-103.939)
    x_test[:,:,:,0] = (x_test[:,:,:,0]-123.680)
    x_test[:,:,:,1] = (x_test[:,:,:,1]-116.779)
    x_test[:,:,:,2] = (x_test[:,:,:,2]-103.939)

    return x_train, x_test


    

In [5]:

    

def scheduler(epoch):
  learning_rate_init = 0.01
  if epoch > 80:
    learning_rate_init = 0.001
  if epoch > 120:
    learning_rate_init = 0.0001
  return learning_rate_init


    

In [6]:

    

def VGG19():
    model = Sequential()

    # Block 1
    model.add(Conv2D(64, (3, 3), padding='same', kernel_initializer=he_normal(), name='block1_conv1', input_shape=x_train.shape[1:]))
    model.add(Activation('relu'))
    model.add(Conv2D(64, (3, 3), padding='same', kernel_initializer=he_normal(), name='block1_conv2'))
    model.add(Activation('relu'))
    model.add(MaxPooling2D((2, 2), strides=(2, 2), name='block1_pool'))

    # Block 2
    model.add(Conv2D(128, (3, 3), padding='same', kernel_initializer=he_normal(), name='block2_conv1'))
    model.add(Activation('relu'))
    model.add(Conv2D(128, (3, 3), padding='same', kernel_initializer=he_normal(), name='block2_conv2'))
    model.add(Activation('relu'))
    model.add(MaxPooling2D((2, 2), strides=(2, 2), name='block2_pool'))

    # Block 3
    model.add(Conv2D(256, (3, 3), padding='same', kernel_initializer=he_normal(), name='block3_conv1'))
    model.add(Activation('relu'))
    model.add(Conv2D(256, (3, 3), padding='same', kernel_initializer=he_normal(), name='block3_conv2'))
    model.add(Activation('relu'))
    model.add(Conv2D(256, (3, 3), padding='same', kernel_initializer=he_normal(), name='block3_conv3'))
    model.add(Activation('relu'))
    model.add(Conv2D(256, (3, 3), padding='same', kernel_initializer=he_normal(), name='block3_conv4'))
    model.add(Activation('relu'))
    model.add(MaxPooling2D((2, 2), strides=(2, 2), name='block3_pool'))

    # Block 4
    model.add(Conv2D(512, (3, 3), padding='same', kernel_initializer=he_normal(), name='block4_conv1'))
    model.add(Activation('relu'))
    model.add(Conv2D(512, (3, 3), padding='same', kernel_initializer=he_normal(), name='block4_conv2'))
    model.add(Activation('relu'))
    model.add(Conv2D(512, (3, 3), padding='same', kernel_initializer=he_normal(), name='block4_conv3'))
    model.add(Activation('relu'))
    model.add(Conv2D(512, (3, 3), padding='same', kernel_initializer=he_normal(), name='block4_conv4'))
    model.add(Activation('relu'))
    model.add(MaxPooling2D((2, 2), strides=(2, 2), name='block4_pool'))

    # Block 5
    model.add(Conv2D(512, (3, 3), padding='same', kernel_initializer=he_normal(), name='block5_conv1'))
    model.add(Activation('relu'))
    model.add(Conv2D(512, (3, 3), padding='same', kernel_initializer=he_normal(), name='block5_conv2'))
    model.add(Activation('relu'))
    model.add(Conv2D(512, (3, 3), padding='same', kernel_initializer=he_normal(), name='block5_conv3'))
    model.add(Activation('relu'))
    model.add(Conv2D(512, (3, 3), padding='same', kernel_initializer=he_normal(), name='block5_conv4'))
    model.add(Activation('relu'))

    # model modification for cifar-10
    model.add(Flatten(name='flatten'))
    model.add(Dense(4096, use_bias = True, kernel_initializer=he_normal(), name='fc_cifa10'))
    model.add(Activation('relu'))
    model.add(Dropout(dropout))
    model.add(Dense(4096, kernel_initializer=he_normal(), name='fc2'))  
    model.add(Activation('relu'))
    model.add(Dropout(dropout))      
    model.add(Dense(10, kernel_initializer=he_normal(), name='predictions_cifa10'))        
    model.add(Activation('softmax'))
    
    return model


    

In [7]:

    

(x_train, y_train), (x_test, y_test) = cifar10.load_data()
y_train = keras.utils.to_categorical(y_train, num_classes)
y_test = keras.utils.to_categorical(y_test, num_classes)
# color preprocessing
x_train, x_test = color_preprocessing(x_train, x_test)

# build network
WEIGHTS_PATH = 'https://github.com/fchollet/deep-learning-models/releases/download/v0.1/vgg19_weights_tf_dim_ordering_tf_kernels.h5'
filepath = get_file('vgg19_weights_tf_dim_ordering_tf_kernels.h5', WEIGHTS_PATH, cache_subdir='models')

model = VGG19()
print(model.summary())

# load pretrained weight from VGG19 by name      
model.load_weights(filepath, by_name=True)

# -------- optimizer setting -------- #
sgd = optimizers.SGD(lr=.1, momentum=0.9, nesterov=True)
model.compile(loss='categorical_crossentropy', optimizer=sgd, metrics=['accuracy'])


    

    




_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
block1_conv1 (Conv2D)        (None, 32, 32, 64)        1792      
_________________________________________________________________
activation_1 (Activation)    (None, 32, 32, 64)        0         
_________________________________________________________________
block1_conv2 (Conv2D)        (None, 32, 32, 64)        36928     
_________________________________________________________________
activation_2 (Activation)    (None, 32, 32, 64)        0         
_________________________________________________________________
block1_pool (MaxPooling2D)   (None, 16, 16, 64)        0         
_________________________________________________________________
block2_conv1 (Conv2D)        (None, 16, 16, 128)       73856     
_________________________________________________________________
activation_3 (Activation)    (None, 16, 16, 128)       0         
_________________________________________________________________
block2_conv2 (Conv2D)        (None, 16, 16, 128)       147584    
_________________________________________________________________
activation_4 (Activation)    (None, 16, 16, 128)       0         
_________________________________________________________________
block2_pool (MaxPooling2D)   (None, 8, 8, 128)         0         
_________________________________________________________________
block3_conv1 (Conv2D)        (None, 8, 8, 256)         295168    
_________________________________________________________________
activation_5 (Activation)    (None, 8, 8, 256)         0         
_________________________________________________________________
block3_conv2 (Conv2D)        (None, 8, 8, 256)         590080    
_________________________________________________________________
activation_6 (Activation)    (None, 8, 8, 256)         0         
_________________________________________________________________
block3_conv3 (Conv2D)        (None, 8, 8, 256)         590080    
_________________________________________________________________
activation_7 (Activation)    (None, 8, 8, 256)         0         
_________________________________________________________________
block3_conv4 (Conv2D)        (None, 8, 8, 256)         590080    
_________________________________________________________________
activation_8 (Activation)    (None, 8, 8, 256)         0         
_________________________________________________________________
block3_pool (MaxPooling2D)   (None, 4, 4, 256)         0         
_________________________________________________________________
block4_conv1 (Conv2D)        (None, 4, 4, 512)         1180160   
_________________________________________________________________
activation_9 (Activation)    (None, 4, 4, 512)         0         
_________________________________________________________________
block4_conv2 (Conv2D)        (None, 4, 4, 512)         2359808   
_________________________________________________________________
activation_10 (Activation)   (None, 4, 4, 512)         0         
_________________________________________________________________
block4_conv3 (Conv2D)        (None, 4, 4, 512)         2359808   
_________________________________________________________________
activation_11 (Activation)   (None, 4, 4, 512)         0         
_________________________________________________________________
block4_conv4 (Conv2D)        (None, 4, 4, 512)         2359808   
_________________________________________________________________
activation_12 (Activation)   (None, 4, 4, 512)         0         
_________________________________________________________________
block4_pool (MaxPooling2D)   (None, 2, 2, 512)         0         
_________________________________________________________________
block5_conv1 (Conv2D)        (None, 2, 2, 512)         2359808   
_________________________________________________________________
activation_13 (Activation)   (None, 2, 2, 512)         0         
_________________________________________________________________
block5_conv2 (Conv2D)        (None, 2, 2, 512)         2359808   
_________________________________________________________________
activation_14 (Activation)   (None, 2, 2, 512)         0         
_________________________________________________________________
block5_conv3 (Conv2D)        (None, 2, 2, 512)         2359808   
_________________________________________________________________
activation_15 (Activation)   (None, 2, 2, 512)         0         
_________________________________________________________________
block5_conv4 (Conv2D)        (None, 2, 2, 512)         2359808   
_________________________________________________________________
activation_16 (Activation)   (None, 2, 2, 512)         0         
_________________________________________________________________
flatten (Flatten)            (None, 2048)              0         
_________________________________________________________________
fc_cifa10 (Dense)            (None, 4096)              8392704   
_________________________________________________________________
activation_17 (Activation)   (None, 4096)              0         
_________________________________________________________________
dropout_1 (Dropout)          (None, 4096)              0         
_________________________________________________________________
fc2 (Dense)                  (None, 4096)              16781312  
_________________________________________________________________
activation_18 (Activation)   (None, 4096)              0         
_________________________________________________________________
dropout_2 (Dropout)          (None, 4096)              0         
_________________________________________________________________
predictions_cifa10 (Dense)   (None, 10)                40970     
_________________________________________________________________
activation_19 (Activation)   (None, 10)                0         
=================================================================
Total params: 45,239,370
Trainable params: 45,239,370
Non-trainable params: 0
_________________________________________________________________
None

In [8]:

    

tb_cb = TensorBoard(log_dir=log_filepath, histogram_freq=0)
change_lr = LearningRateScheduler(scheduler)
cbks = [change_lr,tb_cb]


    

In [9]:

    

print('Using real-time data augmentation.')
datagen = ImageDataGenerator(horizontal_flip=True,
        width_shift_range=0.125,height_shift_range=0.125,fill_mode='constant',cval=0.)

datagen.fit(x_train)


    

    




Using real-time data augmentation.

In [10]:

    

model.fit_generator(datagen.flow(x_train, y_train,batch_size=batch_size),
                    steps_per_epoch=iterations,
                    epochs=epochs,
                    callbacks=cbks,
                    validation_data=(x_test, y_test))
model.save('vgg19_retrain_WHE.h5')


    

    




Epoch 1/170
391/391 [==============================] - 62s - loss: 2.1615 - acc: 0.1582 - val_loss: 1.8489 - val_acc: 0.2243
Epoch 2/170
391/391 [==============================] - 58s - loss: 1.6493 - acc: 0.3601 - val_loss: 1.2095 - val_acc: 0.5590
Epoch 3/170
391/391 [==============================] - 58s - loss: 0.9550 - acc: 0.6790 - val_loss: 0.7549 - val_acc: 0.7524
Epoch 4/170
391/391 [==============================] - 58s - loss: 0.7024 - acc: 0.7652 - val_loss: 0.6260 - val_acc: 0.7955
Epoch 5/170
391/391 [==============================] - 58s - loss: 0.5807 - acc: 0.8084 - val_loss: 0.5459 - val_acc: 0.8199
Epoch 6/170
391/391 [==============================] - 58s - loss: 0.5178 - acc: 0.8301 - val_loss: 0.4862 - val_acc: 0.8433
Epoch 7/170
391/391 [==============================] - 58s - loss: 0.4574 - acc: 0.8490 - val_loss: 0.4639 - val_acc: 0.8450
Epoch 8/170
391/391 [==============================] - 58s - loss: 0.4093 - acc: 0.8644 - val_loss: 0.4629 - val_acc: 0.8520
Epoch 9/170
391/391 [==============================] - 58s - loss: 0.3807 - acc: 0.8733 - val_loss: 0.4652 - val_acc: 0.8531
Epoch 10/170
391/391 [==============================] - 59s - loss: 0.3498 - acc: 0.8849 - val_loss: 0.4790 - val_acc: 0.8561
Epoch 11/170
391/391 [==============================] - 59s - loss: 0.3245 - acc: 0.8920 - val_loss: 0.4133 - val_acc: 0.8645
Epoch 12/170
391/391 [==============================] - 59s - loss: 0.3012 - acc: 0.9002 - val_loss: 0.4040 - val_acc: 0.8667
Epoch 13/170
391/391 [==============================] - 59s - loss: 0.2820 - acc: 0.9053 - val_loss: 0.3990 - val_acc: 0.8716
Epoch 14/170
391/391 [==============================] - 59s - loss: 0.2650 - acc: 0.9126 - val_loss: 0.3840 - val_acc: 0.8799
Epoch 15/170
391/391 [==============================] - 59s - loss: 0.2441 - acc: 0.9184 - val_loss: 0.3814 - val_acc: 0.8779
Epoch 16/170
391/391 [==============================] - 59s - loss: 0.2315 - acc: 0.9233 - val_loss: 0.4091 - val_acc: 0.8743
Epoch 17/170
391/391 [==============================] - 59s - loss: 0.2192 - acc: 0.9268 - val_loss: 0.3718 - val_acc: 0.8816
Epoch 18/170
391/391 [==============================] - 59s - loss: 0.2024 - acc: 0.9338 - val_loss: 0.4119 - val_acc: 0.8794
Epoch 19/170
391/391 [==============================] - 59s - loss: 0.1921 - acc: 0.9366 - val_loss: 0.4260 - val_acc: 0.8778
Epoch 20/170
391/391 [==============================] - 59s - loss: 0.1867 - acc: 0.9379 - val_loss: 0.4223 - val_acc: 0.8730
Epoch 21/170
391/391 [==============================] - 59s - loss: 0.1738 - acc: 0.9423 - val_loss: 0.3785 - val_acc: 0.8866
Epoch 22/170
391/391 [==============================] - 59s - loss: 0.1630 - acc: 0.9466 - val_loss: 0.4024 - val_acc: 0.8831
Epoch 23/170
391/391 [==============================] - 59s - loss: 0.1524 - acc: 0.9504 - val_loss: 0.3883 - val_acc: 0.8874
Epoch 24/170
391/391 [==============================] - 59s - loss: 0.1518 - acc: 0.9496 - val_loss: 0.3689 - val_acc: 0.8928
Epoch 25/170
391/391 [==============================] - 59s - loss: 0.1411 - acc: 0.9532 - val_loss: 0.4278 - val_acc: 0.8837
Epoch 26/170
391/391 [==============================] - 59s - loss: 0.1362 - acc: 0.9552 - val_loss: 0.4309 - val_acc: 0.8841
Epoch 27/170
391/391 [==============================] - 58s - loss: 0.1278 - acc: 0.9578 - val_loss: 0.4042 - val_acc: 0.8862
Epoch 28/170
391/391 [==============================] - 59s - loss: 0.1255 - acc: 0.9578 - val_loss: 0.4596 - val_acc: 0.8807
Epoch 29/170
391/391 [==============================] - 59s - loss: 0.1203 - acc: 0.9604 - val_loss: 0.4113 - val_acc: 0.8982
Epoch 30/170
391/391 [==============================] - 59s - loss: 0.1165 - acc: 0.9628 - val_loss: 0.4128 - val_acc: 0.8902
Epoch 31/170
391/391 [==============================] - 59s - loss: 0.1042 - acc: 0.9652 - val_loss: 0.4029 - val_acc: 0.8953
Epoch 32/170
391/391 [==============================] - 59s - loss: 0.1054 - acc: 0.9647 - val_loss: 0.4011 - val_acc: 0.8930
Epoch 33/170
391/391 [==============================] - 59s - loss: 0.1021 - acc: 0.9660 - val_loss: 0.4034 - val_acc: 0.8980
Epoch 34/170
391/391 [==============================] - 59s - loss: 0.0930 - acc: 0.9685 - val_loss: 0.4055 - val_acc: 0.8903
Epoch 35/170
391/391 [==============================] - 59s - loss: 0.0905 - acc: 0.9702 - val_loss: 0.3711 - val_acc: 0.8985
Epoch 36/170
391/391 [==============================] - 59s - loss: 0.0924 - acc: 0.9697 - val_loss: 0.3568 - val_acc: 0.9046
Epoch 37/170
391/391 [==============================] - 59s - loss: 0.0832 - acc: 0.9725 - val_loss: 0.4074 - val_acc: 0.8993
Epoch 38/170
391/391 [==============================] - 59s - loss: 0.0858 - acc: 0.9713 - val_loss: 0.4066 - val_acc: 0.8977
Epoch 39/170
391/391 [==============================] - 59s - loss: 0.0852 - acc: 0.9714 - val_loss: 0.4103 - val_acc: 0.8949
Epoch 40/170
391/391 [==============================] - 59s - loss: 0.0759 - acc: 0.9748 - val_loss: 0.4220 - val_acc: 0.8986
Epoch 41/170
391/391 [==============================] - 59s - loss: 0.0714 - acc: 0.9762 - val_loss: 0.3726 - val_acc: 0.9064
Epoch 42/170
391/391 [==============================] - 59s - loss: 0.0734 - acc: 0.9755 - val_loss: 0.3923 - val_acc: 0.8980
Epoch 43/170
391/391 [==============================] - 59s - loss: 0.0692 - acc: 0.9775 - val_loss: 0.4600 - val_acc: 0.8953
Epoch 44/170
391/391 [==============================] - 59s - loss: 0.0701 - acc: 0.9769 - val_loss: 0.4043 - val_acc: 0.9002
Epoch 45/170
391/391 [==============================] - 59s - loss: 0.0656 - acc: 0.9787 - val_loss: 0.4119 - val_acc: 0.9008
Epoch 46/170
391/391 [==============================] - 59s - loss: 0.0664 - acc: 0.9773 - val_loss: 0.4240 - val_acc: 0.8962
Epoch 47/170
391/391 [==============================] - 59s - loss: 0.0607 - acc: 0.9802 - val_loss: 0.4133 - val_acc: 0.9012
Epoch 48/170
391/391 [==============================] - 59s - loss: 0.0594 - acc: 0.9796 - val_loss: 0.4440 - val_acc: 0.8974
Epoch 49/170
391/391 [==============================] - 59s - loss: 0.0594 - acc: 0.9806 - val_loss: 0.4591 - val_acc: 0.8946
Epoch 50/170
391/391 [==============================] - 59s - loss: 0.0559 - acc: 0.9815 - val_loss: 0.4300 - val_acc: 0.8970
Epoch 51/170
391/391 [==============================] - 59s - loss: 0.0534 - acc: 0.9826 - val_loss: 0.4257 - val_acc: 0.8995
Epoch 52/170
391/391 [==============================] - 59s - loss: 0.0542 - acc: 0.9826 - val_loss: 0.4197 - val_acc: 0.8965
Epoch 53/170
391/391 [==============================] - 59s - loss: 0.0526 - acc: 0.9824 - val_loss: 0.4181 - val_acc: 0.9029
Epoch 54/170
391/391 [==============================] - 59s - loss: 0.0515 - acc: 0.9832 - val_loss: 0.4212 - val_acc: 0.9002
Epoch 55/170
391/391 [==============================] - 58s - loss: 0.0454 - acc: 0.9852 - val_loss: 0.4485 - val_acc: 0.9020
Epoch 56/170
391/391 [==============================] - 59s - loss: 0.0465 - acc: 0.9847 - val_loss: 0.4382 - val_acc: 0.8989
Epoch 57/170
391/391 [==============================] - 59s - loss: 0.0445 - acc: 0.9858 - val_loss: 0.4004 - val_acc: 0.9053
Epoch 58/170
391/391 [==============================] - 59s - loss: 0.0446 - acc: 0.9853 - val_loss: 0.4324 - val_acc: 0.9018
Epoch 59/170
391/391 [==============================] - 59s - loss: 0.0430 - acc: 0.9866 - val_loss: 0.4499 - val_acc: 0.8993
Epoch 60/170
391/391 [==============================] - 59s - loss: 0.0464 - acc: 0.9852 - val_loss: 0.4415 - val_acc: 0.9030
Epoch 61/170
391/391 [==============================] - 59s - loss: 0.0413 - acc: 0.9872 - val_loss: 0.4242 - val_acc: 0.9046
Epoch 62/170
391/391 [==============================] - 58s - loss: 0.0422 - acc: 0.9871 - val_loss: 0.4770 - val_acc: 0.9020
Epoch 63/170
391/391 [==============================] - 58s - loss: 0.0412 - acc: 0.9867 - val_loss: 0.4547 - val_acc: 0.9046
Epoch 64/170
391/391 [==============================] - 59s - loss: 0.0429 - acc: 0.9855 - val_loss: 0.3991 - val_acc: 0.9005
Epoch 65/170
391/391 [==============================] - 59s - loss: 0.0390 - acc: 0.9872 - val_loss: 0.4650 - val_acc: 0.9068
Epoch 66/170
391/391 [==============================] - 58s - loss: 0.0388 - acc: 0.9879 - val_loss: 0.4395 - val_acc: 0.9044
Epoch 67/170
391/391 [==============================] - 59s - loss: 0.0361 - acc: 0.9888 - val_loss: 0.4630 - val_acc: 0.9035
Epoch 68/170
391/391 [==============================] - 59s - loss: 0.0378 - acc: 0.9878 - val_loss: 0.4434 - val_acc: 0.9072
Epoch 69/170
391/391 [==============================] - 59s - loss: 0.0384 - acc: 0.9877 - val_loss: 0.4974 - val_acc: 0.8975
Epoch 70/170
391/391 [==============================] - 59s - loss: 0.0365 - acc: 0.9882 - val_loss: 0.4688 - val_acc: 0.9012
Epoch 71/170
391/391 [==============================] - 59s - loss: 0.0377 - acc: 0.9874 - val_loss: 0.4433 - val_acc: 0.9121
Epoch 72/170
391/391 [==============================] - 59s - loss: 0.0352 - acc: 0.9882 - val_loss: 0.4744 - val_acc: 0.9019
Epoch 73/170
391/391 [==============================] - 59s - loss: 0.0329 - acc: 0.9895 - val_loss: 0.4751 - val_acc: 0.9040
Epoch 74/170
391/391 [==============================] - 59s - loss: 0.0331 - acc: 0.9894 - val_loss: 0.4696 - val_acc: 0.9046
Epoch 75/170
391/391 [==============================] - 58s - loss: 0.0321 - acc: 0.9899 - val_loss: 0.4673 - val_acc: 0.9052
Epoch 76/170
391/391 [==============================] - 59s - loss: 0.0326 - acc: 0.9900 - val_loss: 0.4540 - val_acc: 0.9042
Epoch 77/170
391/391 [==============================] - 58s - loss: 0.0313 - acc: 0.9895 - val_loss: 0.4447 - val_acc: 0.9050
Epoch 78/170
391/391 [==============================] - 59s - loss: 0.0291 - acc: 0.9906 - val_loss: 0.4563 - val_acc: 0.9016
Epoch 79/170
391/391 [==============================] - 59s - loss: 0.0319 - acc: 0.9899 - val_loss: 0.4568 - val_acc: 0.9012
Epoch 80/170
391/391 [==============================] - 59s - loss: 0.0292 - acc: 0.9906 - val_loss: 0.5000 - val_acc: 0.9053
Epoch 81/170
391/391 [==============================] - 59s - loss: 0.0309 - acc: 0.9908 - val_loss: 0.4738 - val_acc: 0.9046
Epoch 82/170
391/391 [==============================] - 59s - loss: 0.0154 - acc: 0.9951 - val_loss: 0.4346 - val_acc: 0.9148
Epoch 83/170
391/391 [==============================] - 59s - loss: 0.0076 - acc: 0.9975 - val_loss: 0.4411 - val_acc: 0.9166
Epoch 84/170
391/391 [==============================] - 59s - loss: 0.0065 - acc: 0.9980 - val_loss: 0.4588 - val_acc: 0.9178
Epoch 85/170
391/391 [==============================] - 59s - loss: 0.0044 - acc: 0.9987 - val_loss: 0.4674 - val_acc: 0.9172
Epoch 86/170
391/391 [==============================] - 59s - loss: 0.0039 - acc: 0.9986 - val_loss: 0.4736 - val_acc: 0.9182
Epoch 87/170
391/391 [==============================] - 58s - loss: 0.0038 - acc: 0.9987 - val_loss: 0.4812 - val_acc: 0.9182
Epoch 88/170
391/391 [==============================] - 59s - loss: 0.0037 - acc: 0.9990 - val_loss: 0.4870 - val_acc: 0.9176
Epoch 89/170
391/391 [==============================] - 59s - loss: 0.0037 - acc: 0.9990 - val_loss: 0.4909 - val_acc: 0.9171
Epoch 90/170
391/391 [==============================] - 59s - loss: 0.0031 - acc: 0.9991 - val_loss: 0.4925 - val_acc: 0.9170
Epoch 91/170
391/391 [==============================] - 59s - loss: 0.0027 - acc: 0.9991 - val_loss: 0.4987 - val_acc: 0.9176
Epoch 92/170
391/391 [==============================] - 59s - loss: 0.0031 - acc: 0.9990 - val_loss: 0.5063 - val_acc: 0.9161
Epoch 93/170
391/391 [==============================] - 59s - loss: 0.0027 - acc: 0.9991 - val_loss: 0.5027 - val_acc: 0.9177
Epoch 94/170
391/391 [==============================] - 59s - loss: 0.0029 - acc: 0.9991 - val_loss: 0.5013 - val_acc: 0.9187
Epoch 95/170
391/391 [==============================] - 59s - loss: 0.0022 - acc: 0.9991 - val_loss: 0.5109 - val_acc: 0.9195
Epoch 96/170
391/391 [==============================] - 59s - loss: 0.0020 - acc: 0.9994 - val_loss: 0.5169 - val_acc: 0.9196
Epoch 97/170
391/391 [==============================] - 59s - loss: 0.0020 - acc: 0.9995 - val_loss: 0.5171 - val_acc: 0.9209
Epoch 98/170
391/391 [==============================] - 59s - loss: 0.0021 - acc: 0.9993 - val_loss: 0.5183 - val_acc: 0.9189
Epoch 99/170
391/391 [==============================] - 59s - loss: 0.0015 - acc: 0.9996 - val_loss: 0.5193 - val_acc: 0.9190
Epoch 100/170
391/391 [==============================] - 59s - loss: 0.0023 - acc: 0.9992 - val_loss: 0.5184 - val_acc: 0.9190
Epoch 101/170
391/391 [==============================] - 59s - loss: 0.0022 - acc: 0.9994 - val_loss: 0.5211 - val_acc: 0.9207
Epoch 102/170
391/391 [==============================] - 59s - loss: 0.0021 - acc: 0.9994 - val_loss: 0.5272 - val_acc: 0.9198
Epoch 103/170
391/391 [==============================] - 59s - loss: 0.0021 - acc: 0.9993 - val_loss: 0.5214 - val_acc: 0.9192
Epoch 104/170
391/391 [==============================] - 58s - loss: 0.0014 - acc: 0.9996 - val_loss: 0.5257 - val_acc: 0.9196
Epoch 105/170
391/391 [==============================] - 58s - loss: 0.0017 - acc: 0.9994 - val_loss: 0.5303 - val_acc: 0.9185
Epoch 106/170
391/391 [==============================] - 58s - loss: 0.0018 - acc: 0.9994 - val_loss: 0.5276 - val_acc: 0.9190
Epoch 107/170
391/391 [==============================] - 58s - loss: 0.0013 - acc: 0.9996 - val_loss: 0.5315 - val_acc: 0.9203
Epoch 108/170
391/391 [==============================] - 59s - loss: 0.0013 - acc: 0.9997 - val_loss: 0.5394 - val_acc: 0.9192
Epoch 109/170
391/391 [==============================] - 58s - loss: 0.0016 - acc: 0.9994 - val_loss: 0.5406 - val_acc: 0.9200
Epoch 110/170
391/391 [==============================] - 58s - loss: 0.0014 - acc: 0.9995 - val_loss: 0.5446 - val_acc: 0.9201
Epoch 111/170
391/391 [==============================] - 58s - loss: 0.0014 - acc: 0.9996 - val_loss: 0.5461 - val_acc: 0.9202
Epoch 112/170
391/391 [==============================] - 58s - loss: 0.0016 - acc: 0.9996 - val_loss: 0.5477 - val_acc: 0.9197
Epoch 113/170
391/391 [==============================] - 58s - loss: 0.0013 - acc: 0.9996 - val_loss: 0.5483 - val_acc: 0.9198
Epoch 114/170
391/391 [==============================] - 58s - loss: 8.4724e-04 - acc: 0.9997 - val_loss: 0.5517 - val_acc: 0.9192
Epoch 115/170
391/391 [==============================] - 58s - loss: 0.0011 - acc: 0.9996 - val_loss: 0.5588 - val_acc: 0.9192
Epoch 116/170
391/391 [==============================] - 58s - loss: 0.0011 - acc: 0.9996 - val_loss: 0.5566 - val_acc: 0.9206
Epoch 117/170
391/391 [==============================] - 58s - loss: 0.0013 - acc: 0.9996 - val_loss: 0.5558 - val_acc: 0.9220
Epoch 118/170
391/391 [==============================] - 59s - loss: 9.4050e-04 - acc: 0.9997 - val_loss: 0.5635 - val_acc: 0.9203
Epoch 119/170
391/391 [==============================] - 58s - loss: 0.0013 - acc: 0.9997 - val_loss: 0.5566 - val_acc: 0.9208
Epoch 120/170
391/391 [==============================] - 58s - loss: 0.0015 - acc: 0.9996 - val_loss: 0.5559 - val_acc: 0.9204
Epoch 121/170
391/391 [==============================] - 58s - loss: 0.0011 - acc: 0.9996 - val_loss: 0.5615 - val_acc: 0.9203
Epoch 122/170
391/391 [==============================] - 59s - loss: 0.0012 - acc: 0.9997 - val_loss: 0.5606 - val_acc: 0.9201
Epoch 123/170
391/391 [==============================] - 58s - loss: 0.0015 - acc: 0.9995 - val_loss: 0.5603 - val_acc: 0.9201
Epoch 124/170
391/391 [==============================] - 58s - loss: 0.0010 - acc: 0.9997 - val_loss: 0.5607 - val_acc: 0.9201
Epoch 125/170
391/391 [==============================] - 59s - loss: 0.0011 - acc: 0.9996 - val_loss: 0.5612 - val_acc: 0.9201
Epoch 126/170
391/391 [==============================] - 58s - loss: 8.7425e-04 - acc: 0.9998 - val_loss: 0.5612 - val_acc: 0.9199
Epoch 127/170
391/391 [==============================] - 58s - loss: 7.8571e-04 - acc: 0.9997 - val_loss: 0.5603 - val_acc: 0.9202
Epoch 128/170
391/391 [==============================] - 58s - loss: 7.6603e-04 - acc: 0.9998 - val_loss: 0.5606 - val_acc: 0.9202
Epoch 129/170
391/391 [==============================] - 59s - loss: 0.0012 - acc: 0.9996 - val_loss: 0.5611 - val_acc: 0.9202
Epoch 130/170
391/391 [==============================] - 58s - loss: 9.5339e-04 - acc: 0.9997 - val_loss: 0.5615 - val_acc: 0.9204
Epoch 131/170
391/391 [==============================] - 58s - loss: 7.1335e-04 - acc: 0.9997 - val_loss: 0.5614 - val_acc: 0.9206
Epoch 132/170
391/391 [==============================] - 58s - loss: 7.4299e-04 - acc: 0.9997 - val_loss: 0.5619 - val_acc: 0.9203
Epoch 133/170
391/391 [==============================] - 58s - loss: 7.2960e-04 - acc: 0.9998 - val_loss: 0.5620 - val_acc: 0.9200
Epoch 134/170
391/391 [==============================] - 58s - loss: 8.6425e-04 - acc: 0.9998 - val_loss: 0.5623 - val_acc: 0.9203
Epoch 135/170
391/391 [==============================] - 58s - loss: 7.5843e-04 - acc: 0.9997 - val_loss: 0.5626 - val_acc: 0.9200
Epoch 136/170
391/391 [==============================] - 58s - loss: 8.1337e-04 - acc: 0.9997 - val_loss: 0.5626 - val_acc: 0.9205
Epoch 137/170
391/391 [==============================] - 59s - loss: 7.6464e-04 - acc: 0.9998 - val_loss: 0.5624 - val_acc: 0.9207
Epoch 138/170
391/391 [==============================] - 59s - loss: 9.9768e-04 - acc: 0.9997 - val_loss: 0.5629 - val_acc: 0.9204
Epoch 139/170
391/391 [==============================] - 59s - loss: 9.2453e-04 - acc: 0.9997 - val_loss: 0.5629 - val_acc: 0.9207
Epoch 140/170
391/391 [==============================] - 59s - loss: 6.6737e-04 - acc: 0.9998 - val_loss: 0.5629 - val_acc: 0.9206
Epoch 141/170
391/391 [==============================] - 59s - loss: 6.9747e-04 - acc: 0.9998 - val_loss: 0.5626 - val_acc: 0.9206
Epoch 142/170
391/391 [==============================] - 59s - loss: 8.6272e-04 - acc: 0.9997 - val_loss: 0.5627 - val_acc: 0.9208
Epoch 143/170
391/391 [==============================] - 59s - loss: 4.7397e-04 - acc: 0.9998 - val_loss: 0.5632 - val_acc: 0.9208
Epoch 144/170
391/391 [==============================] - 59s - loss: 4.9275e-04 - acc: 0.9998 - val_loss: 0.5641 - val_acc: 0.9209
Epoch 145/170
391/391 [==============================] - 59s - loss: 5.5953e-04 - acc: 0.9998 - val_loss: 0.5644 - val_acc: 0.9207
Epoch 146/170
391/391 [==============================] - 58s - loss: 8.7547e-04 - acc: 0.9997 - val_loss: 0.5658 - val_acc: 0.9210
Epoch 147/170
391/391 [==============================] - 59s - loss: 0.0011 - acc: 0.9997 - val_loss: 0.5656 - val_acc: 0.9209
Epoch 148/170
391/391 [==============================] - 58s - loss: 0.0010 - acc: 0.9997 - val_loss: 0.5657 - val_acc: 0.9203
Epoch 149/170
391/391 [==============================] - 58s - loss: 6.3509e-04 - acc: 0.9998 - val_loss: 0.5661 - val_acc: 0.9208
Epoch 150/170
391/391 [==============================] - 59s - loss: 7.5543e-04 - acc: 0.9998 - val_loss: 0.5661 - val_acc: 0.9208
Epoch 151/170
391/391 [==============================] - 58s - loss: 0.0010 - acc: 0.9996 - val_loss: 0.5656 - val_acc: 0.9208
Epoch 152/170
391/391 [==============================] - 59s - loss: 9.9573e-04 - acc: 0.9998 - val_loss: 0.5645 - val_acc: 0.9213
Epoch 153/170
391/391 [==============================] - 58s - loss: 7.8192e-04 - acc: 0.9997 - val_loss: 0.5643 - val_acc: 0.9210
Epoch 154/170
391/391 [==============================] - 58s - loss: 8.6765e-04 - acc: 0.9998 - val_loss: 0.5652 - val_acc: 0.9212
Epoch 155/170
391/391 [==============================] - 58s - loss: 9.8961e-04 - acc: 0.9998 - val_loss: 0.5660 - val_acc: 0.9212
Epoch 156/170
391/391 [==============================] - 58s - loss: 7.4320e-04 - acc: 0.9998 - val_loss: 0.5665 - val_acc: 0.9211
Epoch 157/170
391/391 [==============================] - 59s - loss: 0.0010 - acc: 0.9997 - val_loss: 0.5666 - val_acc: 0.9212
Epoch 158/170
391/391 [==============================] - 58s - loss: 0.0010 - acc: 0.9997 - val_loss: 0.5674 - val_acc: 0.9208
Epoch 159/170
391/391 [==============================] - 58s - loss: 7.3119e-04 - acc: 0.9999 - val_loss: 0.5669 - val_acc: 0.9210
Epoch 160/170
391/391 [==============================] - 58s - loss: 8.2529e-04 - acc: 0.9997 - val_loss: 0.5669 - val_acc: 0.9209
Epoch 161/170
391/391 [==============================] - 58s - loss: 6.8612e-04 - acc: 0.9998 - val_loss: 0.5665 - val_acc: 0.9218
Epoch 162/170
391/391 [==============================] - 58s - loss: 8.9734e-04 - acc: 0.9997 - val_loss: 0.5676 - val_acc: 0.9210
Epoch 163/170
391/391 [==============================] - 58s - loss: 6.4278e-04 - acc: 0.9999 - val_loss: 0.5672 - val_acc: 0.9213
Epoch 164/170
391/391 [==============================] - 58s - loss: 7.9099e-04 - acc: 0.9998 - val_loss: 0.5672 - val_acc: 0.9216
Epoch 165/170
391/391 [==============================] - 58s - loss: 5.6363e-04 - acc: 0.9998 - val_loss: 0.5673 - val_acc: 0.9213
Epoch 166/170
391/391 [==============================] - 58s - loss: 8.7347e-04 - acc: 0.9997 - val_loss: 0.5671 - val_acc: 0.9206
Epoch 167/170
391/391 [==============================] - 44s - loss: 6.9125e-04 - acc: 0.9998 - val_loss: 0.5672 - val_acc: 0.9209
Epoch 168/170
391/391 [==============================] - 30s - loss: 6.5141e-04 - acc: 0.9998 - val_loss: 0.5669 - val_acc: 0.9210
Epoch 169/170
391/391 [==============================] - 30s - loss: 8.7084e-04 - acc: 0.9997 - val_loss: 0.5677 - val_acc: 0.9210
Epoch 170/170
391/391 [==============================] - 30s - loss: 9.6132e-04 - acc: 0.9997 - val_loss: 0.5676 - val_acc: 0.9208

In [ ]: