In [18]:

    

import keras
from keras.datasets import mnist
from keras.models import Sequential
from keras.layers import Dense, Dropout, Flatten, Activation
from keras.layers import Conv2D, MaxPooling2D, BatchNormalization
from keras import backend as K

import numpy as np


    

In [19]:

    

batch_size = 2400
num_classes = 2
epochs = 200


    

In [20]:

    

train_X = np.load('./train_X.npy')
train_Y = np.load('./train_Y.npy')


    

In [21]:

    

test_X = np.load('./test_X.npy')
test_Y = np.load('./test_Y.npy')


    

In [22]:

    

train_X.dtype


    

    
Out[22]:





dtype('float64')

In [23]:

    

train_X.shape


    

    
Out[23]:





(48000, 40, 40, 1)

In [24]:

    

test_X.shape


    

    
Out[24]:





(1000, 40, 40, 1)

In [25]:

    

model = Sequential()
model.add(Conv2D(32, kernel_size=(3, 3), activation='relu',input_shape=(40, 40, 1)))
model.add(Conv2D(32, (3, 3), strides=1, activation='relu'))
model.add(MaxPooling2D(pool_size=(2, 2)))
model.add(Dropout(0.2))

model.add(Conv2D(64, (3, 3), strides=1, activation='relu'))
model.add(MaxPooling2D(pool_size=(2, 2)))
model.add(Dropout(0.4))

model.add(Flatten())
model.add(Dense(128, activation='relu'))
model.add(Dropout(0.5))
model.add(Dense(num_classes))
model.add(Activation('softmax'))


    

In [26]:

    

model.compile(loss=keras.losses.categorical_crossentropy,
              optimizer=keras.optimizers.Adadelta(),
              metrics=['accuracy'])


    

In [27]:

    

model.fit(train_X, train_Y, batch_size=batch_size, epochs=epochs,
          verbose=1, validation_data=(test_X, test_Y))


    

    




Train on 48000 samples, validate on 1000 samples
Epoch 1/200
48000/48000 [==============================] - 385s - loss: 0.6927 - acc: 0.5508 - val_loss: 0.6470 - val_acc: 0.6920
Epoch 2/200
48000/48000 [==============================] - 503s - loss: 0.6404 - acc: 0.6337 - val_loss: 0.5783 - val_acc: 0.7180
Epoch 3/200
48000/48000 [==============================] - 453s - loss: 0.5845 - acc: 0.6960 - val_loss: 0.5441 - val_acc: 0.7360
Epoch 4/200
48000/48000 [==============================] - 422s - loss: 0.5517 - acc: 0.7243 - val_loss: 0.5011 - val_acc: 0.7690
Epoch 5/200
48000/48000 [==============================] - 389s - loss: 0.5256 - acc: 0.7403 - val_loss: 0.5087 - val_acc: 0.7600
Epoch 6/200
48000/48000 [==============================] - 526s - loss: 0.5064 - acc: 0.7508 - val_loss: 0.4869 - val_acc: 0.7710
Epoch 7/200
48000/48000 [==============================] - 628s - loss: 0.4900 - acc: 0.7604 - val_loss: 0.4785 - val_acc: 0.7740
Epoch 8/200
48000/48000 [==============================] - 566s - loss: 0.4696 - acc: 0.7725 - val_loss: 0.4573 - val_acc: 0.7800
Epoch 9/200
48000/48000 [==============================] - 427s - loss: 0.4594 - acc: 0.7816 - val_loss: 0.4252 - val_acc: 0.8150
Epoch 10/200
48000/48000 [==============================] - 427s - loss: 0.4433 - acc: 0.7920 - val_loss: 0.4098 - val_acc: 0.8340
Epoch 11/200
48000/48000 [==============================] - 416s - loss: 0.4196 - acc: 0.8063 - val_loss: 0.4869 - val_acc: 0.7440
Epoch 12/200
48000/48000 [==============================] - 407s - loss: 0.4109 - acc: 0.8094 - val_loss: 0.3855 - val_acc: 0.8280
Epoch 13/200
48000/48000 [==============================] - 399s - loss: 0.4107 - acc: 0.8102 - val_loss: 0.4034 - val_acc: 0.8300
Epoch 14/200
48000/48000 [==============================] - 397s - loss: 0.3917 - acc: 0.8233 - val_loss: 0.3944 - val_acc: 0.8260
Epoch 15/200
48000/48000 [==============================] - 403s - loss: 0.3691 - acc: 0.8336 - val_loss: 0.3844 - val_acc: 0.8300
Epoch 16/200
48000/48000 [==============================] - 401s - loss: 0.3643 - acc: 0.8373 - val_loss: 0.3561 - val_acc: 0.8510
Epoch 17/200
48000/48000 [==============================] - 400s - loss: 0.3509 - acc: 0.8443 - val_loss: 0.3027 - val_acc: 0.8760
Epoch 18/200
48000/48000 [==============================] - 399s - loss: 0.3410 - acc: 0.8488 - val_loss: 0.3089 - val_acc: 0.8740
Epoch 19/200
48000/48000 [==============================] - 396s - loss: 0.3293 - acc: 0.8552 - val_loss: 0.3184 - val_acc: 0.8640
Epoch 20/200
48000/48000 [==============================] - 395s - loss: 0.3110 - acc: 0.8659 - val_loss: 0.3693 - val_acc: 0.8280
Epoch 21/200
48000/48000 [==============================] - 400s - loss: 0.3063 - acc: 0.8681 - val_loss: 0.2350 - val_acc: 0.9030
Epoch 22/200
48000/48000 [==============================] - 398s - loss: 0.2972 - acc: 0.8734 - val_loss: 0.3153 - val_acc: 0.8830
Epoch 23/200
48000/48000 [==============================] - 406s - loss: 0.2773 - acc: 0.8832 - val_loss: 0.3457 - val_acc: 0.8410
Epoch 24/200
48000/48000 [==============================] - 407s - loss: 0.2636 - acc: 0.8908 - val_loss: 0.1952 - val_acc: 0.9400
Epoch 25/200
48000/48000 [==============================] - 411s - loss: 0.2687 - acc: 0.8864 - val_loss: 0.2378 - val_acc: 0.9170
Epoch 26/200
48000/48000 [==============================] - 407s - loss: 0.2389 - acc: 0.9051 - val_loss: 0.2115 - val_acc: 0.9140
Epoch 27/200
48000/48000 [==============================] - 405s - loss: 0.2386 - acc: 0.8994 - val_loss: 0.1878 - val_acc: 0.9310
Epoch 28/200
48000/48000 [==============================] - 414s - loss: 0.2308 - acc: 0.9042 - val_loss: 0.2186 - val_acc: 0.9120
Epoch 29/200
48000/48000 [==============================] - 408s - loss: 0.2211 - acc: 0.9093 - val_loss: 0.1741 - val_acc: 0.9420
Epoch 30/200
48000/48000 [==============================] - 415s - loss: 0.2109 - acc: 0.9141 - val_loss: 0.2196 - val_acc: 0.9070
Epoch 31/200
48000/48000 [==============================] - 408s - loss: 0.2128 - acc: 0.9131 - val_loss: 0.1474 - val_acc: 0.9530
Epoch 32/200
48000/48000 [==============================] - 417s - loss: 0.1923 - acc: 0.9241 - val_loss: 0.2026 - val_acc: 0.9190
Epoch 33/200
48000/48000 [==============================] - 408s - loss: 0.2005 - acc: 0.9192 - val_loss: 0.1462 - val_acc: 0.9470
Epoch 34/200
48000/48000 [==============================] - 415s - loss: 0.1778 - acc: 0.9296 - val_loss: 0.1359 - val_acc: 0.9430
Epoch 35/200
48000/48000 [==============================] - 418s - loss: 0.1694 - acc: 0.9319 - val_loss: 0.1442 - val_acc: 0.9410
Epoch 36/200
48000/48000 [==============================] - 420s - loss: 0.1611 - acc: 0.9356 - val_loss: 0.1410 - val_acc: 0.9450
Epoch 37/200
48000/48000 [==============================] - 416s - loss: 0.1744 - acc: 0.9308 - val_loss: 0.1505 - val_acc: 0.9390
Epoch 38/200
48000/48000 [==============================] - 423s - loss: 0.1508 - acc: 0.9408 - val_loss: 0.1313 - val_acc: 0.9490
Epoch 39/200
48000/48000 [==============================] - 421s - loss: 0.1532 - acc: 0.9402 - val_loss: 0.1030 - val_acc: 0.9580
Epoch 40/200
48000/48000 [==============================] - 420s - loss: 0.1511 - acc: 0.9405 - val_loss: 0.0940 - val_acc: 0.9600
Epoch 41/200
48000/48000 [==============================] - 411s - loss: 0.1411 - acc: 0.9446 - val_loss: 0.0939 - val_acc: 0.9670
Epoch 42/200
48000/48000 [==============================] - 413s - loss: 0.1323 - acc: 0.9498 - val_loss: 0.0847 - val_acc: 0.9690
Epoch 43/200
48000/48000 [==============================] - 414s - loss: 0.1258 - acc: 0.9510 - val_loss: 0.0928 - val_acc: 0.9660
Epoch 44/200
48000/48000 [==============================] - 422s - loss: 0.1029 - acc: 0.9606 - val_loss: 0.0684 - val_acc: 0.9720
Epoch 45/200
48000/48000 [==============================] - 436s - loss: 0.1176 - acc: 0.9545 - val_loss: 0.0708 - val_acc: 0.9750
Epoch 46/200
48000/48000 [==============================] - 417s - loss: 0.1167 - acc: 0.9548 - val_loss: 0.0700 - val_acc: 0.9760
Epoch 47/200
48000/48000 [==============================] - 430s - loss: 0.1010 - acc: 0.9606 - val_loss: 0.0808 - val_acc: 0.9680
Epoch 48/200
48000/48000 [==============================] - 416s - loss: 0.1029 - acc: 0.9605 - val_loss: 0.0634 - val_acc: 0.9750
Epoch 49/200
48000/48000 [==============================] - 417s - loss: 0.1041 - acc: 0.9596 - val_loss: 0.0685 - val_acc: 0.9790
Epoch 50/200
48000/48000 [==============================] - 409s - loss: 0.0877 - acc: 0.9672 - val_loss: 0.0643 - val_acc: 0.9750
Epoch 51/200
48000/48000 [==============================] - 411s - loss: 0.1009 - acc: 0.9622 - val_loss: 0.0699 - val_acc: 0.9720
Epoch 52/200
48000/48000 [==============================] - 409s - loss: 0.0951 - acc: 0.9641 - val_loss: 0.0860 - val_acc: 0.9660
Epoch 53/200
48000/48000 [==============================] - 400s - loss: 0.0922 - acc: 0.9654 - val_loss: 0.0537 - val_acc: 0.9830
Epoch 54/200
48000/48000 [==============================] - 401s - loss: 0.0786 - acc: 0.9709 - val_loss: 0.0570 - val_acc: 0.9820
Epoch 55/200
48000/48000 [==============================] - 383s - loss: 0.0860 - acc: 0.9673 - val_loss: 0.0501 - val_acc: 0.9820
Epoch 56/200
48000/48000 [==============================] - 397s - loss: 0.0864 - acc: 0.9664 - val_loss: 0.0540 - val_acc: 0.9800
Epoch 57/200
48000/48000 [==============================] - 408s - loss: 0.0761 - acc: 0.9713 - val_loss: 0.0494 - val_acc: 0.9830
Epoch 58/200
48000/48000 [==============================] - 393s - loss: 0.0845 - acc: 0.9681 - val_loss: 0.0621 - val_acc: 0.9750
Epoch 59/200
48000/48000 [==============================] - 395s - loss: 0.0792 - acc: 0.9690 - val_loss: 0.0597 - val_acc: 0.9770
Epoch 60/200
48000/48000 [==============================] - 398s - loss: 0.0730 - acc: 0.9729 - val_loss: 0.0447 - val_acc: 0.9840
Epoch 61/200
48000/48000 [==============================] - 386s - loss: 0.0798 - acc: 0.9696 - val_loss: 0.0473 - val_acc: 0.9870
Epoch 62/200
48000/48000 [==============================] - 393s - loss: 0.0656 - acc: 0.9756 - val_loss: 0.0467 - val_acc: 0.9850
Epoch 63/200
48000/48000 [==============================] - 396s - loss: 0.0719 - acc: 0.9724 - val_loss: 0.0490 - val_acc: 0.9830
Epoch 64/200
48000/48000 [==============================] - 376s - loss: 0.0669 - acc: 0.9756 - val_loss: 0.0424 - val_acc: 0.9870
Epoch 65/200
48000/48000 [==============================] - 388s - loss: 0.0604 - acc: 0.9779 - val_loss: 0.0415 - val_acc: 0.9860
Epoch 66/200
48000/48000 [==============================] - 399s - loss: 0.0723 - acc: 0.9728 - val_loss: 0.0447 - val_acc: 0.9860
Epoch 67/200
48000/48000 [==============================] - 413s - loss: 0.0658 - acc: 0.9753 - val_loss: 0.0393 - val_acc: 0.9880
Epoch 68/200
48000/48000 [==============================] - 416s - loss: 0.0662 - acc: 0.9756 - val_loss: 0.0438 - val_acc: 0.9850
Epoch 69/200
48000/48000 [==============================] - 405s - loss: 0.0588 - acc: 0.9783 - val_loss: 0.0371 - val_acc: 0.9850
Epoch 70/200
48000/48000 [==============================] - 403s - loss: 0.0566 - acc: 0.9796 - val_loss: 0.0426 - val_acc: 0.9860
Epoch 71/200
48000/48000 [==============================] - 400s - loss: 0.0635 - acc: 0.9770 - val_loss: 0.0361 - val_acc: 0.9850
Epoch 72/200
48000/48000 [==============================] - 400s - loss: 0.0604 - acc: 0.9777 - val_loss: 0.0364 - val_acc: 0.9870
Epoch 73/200
48000/48000 [==============================] - 399s - loss: 0.0543 - acc: 0.9800 - val_loss: 0.0356 - val_acc: 0.9870
Epoch 74/200
48000/48000 [==============================] - 397s - loss: 0.0561 - acc: 0.9789 - val_loss: 0.0363 - val_acc: 0.9870
Epoch 75/200
48000/48000 [==============================] - 395s - loss: 0.0516 - acc: 0.9810 - val_loss: 0.0364 - val_acc: 0.9870
Epoch 76/200
48000/48000 [==============================] - 396s - loss: 0.0551 - acc: 0.9795 - val_loss: 0.0374 - val_acc: 0.9840
Epoch 77/200
48000/48000 [==============================] - 389s - loss: 0.0568 - acc: 0.9792 - val_loss: 0.0418 - val_acc: 0.9860
Epoch 78/200
48000/48000 [==============================] - 397s - loss: 0.0556 - acc: 0.9791 - val_loss: 0.0336 - val_acc: 0.9900
Epoch 79/200
48000/48000 [==============================] - 395s - loss: 0.0556 - acc: 0.9795 - val_loss: 0.0329 - val_acc: 0.9880
Epoch 80/200
48000/48000 [==============================] - 396s - loss: 0.0483 - acc: 0.9827 - val_loss: 0.0366 - val_acc: 0.9840
Epoch 81/200
48000/48000 [==============================] - 398s - loss: 0.0572 - acc: 0.9791 - val_loss: 0.0326 - val_acc: 0.9890
Epoch 82/200
48000/48000 [==============================] - 394s - loss: 0.0461 - acc: 0.9830 - val_loss: 0.0304 - val_acc: 0.9910
Epoch 83/200
48000/48000 [==============================] - 405s - loss: 0.0467 - acc: 0.9825 - val_loss: 0.0335 - val_acc: 0.9890
Epoch 84/200
48000/48000 [==============================] - 400s - loss: 0.0498 - acc: 0.9813 - val_loss: 0.0355 - val_acc: 0.9840
Epoch 85/200
48000/48000 [==============================] - 399s - loss: 0.0446 - acc: 0.9837 - val_loss: 0.0318 - val_acc: 0.9900
Epoch 86/200
48000/48000 [==============================] - 404s - loss: 0.0542 - acc: 0.9794 - val_loss: 0.0323 - val_acc: 0.9900
Epoch 87/200
48000/48000 [==============================] - 411s - loss: 0.0428 - acc: 0.9841 - val_loss: 0.0305 - val_acc: 0.9890
Epoch 88/200
48000/48000 [==============================] - 412s - loss: 0.0462 - acc: 0.9832 - val_loss: 0.0298 - val_acc: 0.9910
Epoch 89/200
48000/48000 [==============================] - 412s - loss: 0.0414 - acc: 0.9840 - val_loss: 0.0368 - val_acc: 0.9860
Epoch 90/200
48000/48000 [==============================] - 399s - loss: 0.0463 - acc: 0.9826 - val_loss: 0.0309 - val_acc: 0.9900
Epoch 91/200
48000/48000 [==============================] - 402s - loss: 0.0477 - acc: 0.9813 - val_loss: 0.0360 - val_acc: 0.9840
Epoch 92/200
48000/48000 [==============================] - 416s - loss: 0.0447 - acc: 0.9837 - val_loss: 0.0425 - val_acc: 0.9850
Epoch 93/200
48000/48000 [==============================] - 414s - loss: 0.0415 - acc: 0.9851 - val_loss: 0.0281 - val_acc: 0.9930
Epoch 94/200
48000/48000 [==============================] - 416s - loss: 0.0412 - acc: 0.9847 - val_loss: 0.0301 - val_acc: 0.9900
Epoch 95/200
48000/48000 [==============================] - 404s - loss: 0.0441 - acc: 0.9844 - val_loss: 0.0281 - val_acc: 0.9920
Epoch 96/200
48000/48000 [==============================] - 409s - loss: 0.0460 - acc: 0.9829 - val_loss: 0.0300 - val_acc: 0.9890
Epoch 97/200
48000/48000 [==============================] - 432s - loss: 0.0386 - acc: 0.9860 - val_loss: 0.0281 - val_acc: 0.9930
Epoch 98/200
48000/48000 [==============================] - 425s - loss: 0.0375 - acc: 0.9863 - val_loss: 0.0276 - val_acc: 0.9920
Epoch 99/200
48000/48000 [==============================] - 434s - loss: 0.0393 - acc: 0.9857 - val_loss: 0.0281 - val_acc: 0.9920
Epoch 100/200
48000/48000 [==============================] - 440s - loss: 0.0382 - acc: 0.9862 - val_loss: 0.0266 - val_acc: 0.9930
Epoch 101/200
48000/48000 [==============================] - 431s - loss: 0.0382 - acc: 0.9858 - val_loss: 0.0276 - val_acc: 0.9930
Epoch 102/200
48000/48000 [==============================] - 428s - loss: 0.0351 - acc: 0.9865 - val_loss: 0.0272 - val_acc: 0.9920
Epoch 103/200
48000/48000 [==============================] - 428s - loss: 0.0384 - acc: 0.9862 - val_loss: 0.0271 - val_acc: 0.9930
Epoch 104/200
48000/48000 [==============================] - 434s - loss: 0.0370 - acc: 0.9864 - val_loss: 0.0280 - val_acc: 0.9920
Epoch 105/200
48000/48000 [==============================] - 434s - loss: 0.0378 - acc: 0.9861 - val_loss: 0.0347 - val_acc: 0.9870
Epoch 106/200
48000/48000 [==============================] - 426s - loss: 0.0413 - acc: 0.9846 - val_loss: 0.0284 - val_acc: 0.9910
Epoch 107/200
48000/48000 [==============================] - 434s - loss: 0.0358 - acc: 0.9871 - val_loss: 0.0267 - val_acc: 0.9930
Epoch 108/200
48000/48000 [==============================] - 435s - loss: 0.0311 - acc: 0.9887 - val_loss: 0.0261 - val_acc: 0.9920
Epoch 109/200
48000/48000 [==============================] - 411s - loss: 0.0375 - acc: 0.9864 - val_loss: 0.0321 - val_acc: 0.9880
Epoch 110/200
48000/48000 [==============================] - 392s - loss: 0.0395 - acc: 0.9851 - val_loss: 0.0286 - val_acc: 0.9900
Epoch 111/200
48000/48000 [==============================] - 396s - loss: 0.0334 - acc: 0.9875 - val_loss: 0.0259 - val_acc: 0.9940
Epoch 112/200
48000/48000 [==============================] - 393s - loss: 0.0332 - acc: 0.9874 - val_loss: 0.0325 - val_acc: 0.9870
Epoch 113/200
48000/48000 [==============================] - 398s - loss: 0.0435 - acc: 0.9834 - val_loss: 0.0305 - val_acc: 0.9880
Epoch 114/200
48000/48000 [==============================] - 389s - loss: 0.0309 - acc: 0.9891 - val_loss: 0.0252 - val_acc: 0.9920
Epoch 115/200
48000/48000 [==============================] - 399s - loss: 0.0296 - acc: 0.9892 - val_loss: 0.0255 - val_acc: 0.9940
Epoch 116/200
48000/48000 [==============================] - 400s - loss: 0.0330 - acc: 0.9883 - val_loss: 0.0242 - val_acc: 0.9940
Epoch 117/200
48000/48000 [==============================] - 392s - loss: 0.0322 - acc: 0.9884 - val_loss: 0.0255 - val_acc: 0.9930
Epoch 118/200
48000/48000 [==============================] - 397s - loss: 0.0295 - acc: 0.9892 - val_loss: 0.0254 - val_acc: 0.9900
Epoch 119/200
48000/48000 [==============================] - 399s - loss: 0.0316 - acc: 0.9888 - val_loss: 0.0252 - val_acc: 0.9930
Epoch 120/200
48000/48000 [==============================] - 406s - loss: 0.0306 - acc: 0.9888 - val_loss: 0.0314 - val_acc: 0.9870
Epoch 121/200
48000/48000 [==============================] - 398s - loss: 0.0310 - acc: 0.9885 - val_loss: 0.0239 - val_acc: 0.9930
Epoch 122/200
48000/48000 [==============================] - 409s - loss: 0.0282 - acc: 0.9900 - val_loss: 0.0235 - val_acc: 0.9930
Epoch 123/200
48000/48000 [==============================] - 408s - loss: 0.0323 - acc: 0.9882 - val_loss: 0.0347 - val_acc: 0.9840
Epoch 124/200
48000/48000 [==============================] - 415s - loss: 0.0346 - acc: 0.9872 - val_loss: 0.0237 - val_acc: 0.9930
Epoch 125/200
48000/48000 [==============================] - 413s - loss: 0.0260 - acc: 0.9906 - val_loss: 0.0299 - val_acc: 0.9870
Epoch 126/200
48000/48000 [==============================] - 418s - loss: 0.0276 - acc: 0.9895 - val_loss: 0.0223 - val_acc: 0.9940
Epoch 127/200
48000/48000 [==============================] - 411s - loss: 0.0295 - acc: 0.9891 - val_loss: 0.0228 - val_acc: 0.9940
Epoch 128/200
48000/48000 [==============================] - 407s - loss: 0.0287 - acc: 0.9891 - val_loss: 0.0248 - val_acc: 0.9910
Epoch 129/200
48000/48000 [==============================] - 396s - loss: 0.0269 - acc: 0.9901 - val_loss: 0.0242 - val_acc: 0.9910
Epoch 130/200
48000/48000 [==============================] - 392s - loss: 0.0269 - acc: 0.9901 - val_loss: 0.0231 - val_acc: 0.9940
Epoch 131/200
48000/48000 [==============================] - 413s - loss: 0.0293 - acc: 0.9893 - val_loss: 0.0236 - val_acc: 0.9930
Epoch 132/200
48000/48000 [==============================] - 404s - loss: 0.0255 - acc: 0.9909 - val_loss: 0.0229 - val_acc: 0.9920
Epoch 133/200
48000/48000 [==============================] - 388s - loss: 0.0289 - acc: 0.9891 - val_loss: 0.0281 - val_acc: 0.9900
Epoch 134/200
48000/48000 [==============================] - 386s - loss: 0.0264 - acc: 0.9906 - val_loss: 0.0232 - val_acc: 0.9930
Epoch 135/200
48000/48000 [==============================] - 373s - loss: 0.0236 - acc: 0.9914 - val_loss: 0.0240 - val_acc: 0.9910
Epoch 136/200
48000/48000 [==============================] - 393s - loss: 0.0274 - acc: 0.9897 - val_loss: 0.0223 - val_acc: 0.9920
Epoch 137/200
48000/48000 [==============================] - 386s - loss: 0.0264 - acc: 0.9903 - val_loss: 0.0248 - val_acc: 0.9910
Epoch 138/200
48000/48000 [==============================] - 378s - loss: 0.0233 - acc: 0.9921 - val_loss: 0.0233 - val_acc: 0.9940
Epoch 139/200
48000/48000 [==============================] - 381s - loss: 0.0310 - acc: 0.9888 - val_loss: 0.0233 - val_acc: 0.9940
Epoch 140/200
48000/48000 [==============================] - 372s - loss: 0.0244 - acc: 0.9912 - val_loss: 0.0237 - val_acc: 0.9930
Epoch 141/200
48000/48000 [==============================] - 379s - loss: 0.0216 - acc: 0.9923 - val_loss: 0.0228 - val_acc: 0.9930
Epoch 142/200
48000/48000 [==============================] - 366s - loss: 0.0231 - acc: 0.9912 - val_loss: 0.0233 - val_acc: 0.9940
Epoch 143/200
48000/48000 [==============================] - 367s - loss: 0.0246 - acc: 0.9915 - val_loss: 0.0240 - val_acc: 0.9930
Epoch 144/200
48000/48000 [==============================] - 381s - loss: 0.0267 - acc: 0.9910 - val_loss: 0.0225 - val_acc: 0.9940
Epoch 145/200
48000/48000 [==============================] - 392s - loss: 0.0228 - acc: 0.9918 - val_loss: 0.0221 - val_acc: 0.9940
Epoch 146/200
48000/48000 [==============================] - 375s - loss: 0.0205 - acc: 0.9927 - val_loss: 0.0223 - val_acc: 0.9930
Epoch 147/200
48000/48000 [==============================] - 383s - loss: 0.0219 - acc: 0.9921 - val_loss: 0.0221 - val_acc: 0.9940
Epoch 148/200
48000/48000 [==============================] - 389s - loss: 0.0230 - acc: 0.9916 - val_loss: 0.0211 - val_acc: 0.9950
Epoch 149/200
48000/48000 [==============================] - 395s - loss: 0.0223 - acc: 0.9920 - val_loss: 0.0230 - val_acc: 0.9930
Epoch 150/200
48000/48000 [==============================] - 390s - loss: 0.0216 - acc: 0.9924 - val_loss: 0.0241 - val_acc: 0.9930
Epoch 151/200
48000/48000 [==============================] - 395s - loss: 0.0217 - acc: 0.9925 - val_loss: 0.0224 - val_acc: 0.9940
Epoch 152/200
48000/48000 [==============================] - 385s - loss: 0.0231 - acc: 0.9911 - val_loss: 0.0235 - val_acc: 0.9940
Epoch 153/200
48000/48000 [==============================] - 380s - loss: 0.0234 - acc: 0.9914 - val_loss: 0.0219 - val_acc: 0.9940
Epoch 154/200
48000/48000 [==============================] - 397s - loss: 0.0208 - acc: 0.9926 - val_loss: 0.0237 - val_acc: 0.9900
Epoch 155/200
48000/48000 [==============================] - 395s - loss: 0.0214 - acc: 0.9925 - val_loss: 0.0240 - val_acc: 0.9930
Epoch 156/200
48000/48000 [==============================] - 392s - loss: 0.0215 - acc: 0.9923 - val_loss: 0.0226 - val_acc: 0.9940
Epoch 157/200
48000/48000 [==============================] - 391s - loss: 0.0246 - acc: 0.9911 - val_loss: 0.0239 - val_acc: 0.9940
Epoch 158/200
48000/48000 [==============================] - 385s - loss: 0.0228 - acc: 0.9915 - val_loss: 0.0250 - val_acc: 0.9930
Epoch 159/200
48000/48000 [==============================] - 389s - loss: 0.0211 - acc: 0.9925 - val_loss: 0.0224 - val_acc: 0.9940
Epoch 160/200
48000/48000 [==============================] - 392s - loss: 0.0200 - acc: 0.9928 - val_loss: 0.0222 - val_acc: 0.9940
Epoch 161/200
48000/48000 [==============================] - 398s - loss: 0.0191 - acc: 0.9936 - val_loss: 0.0229 - val_acc: 0.9930
Epoch 162/200
48000/48000 [==============================] - 385s - loss: 0.0187 - acc: 0.9932 - val_loss: 0.0234 - val_acc: 0.9930
Epoch 163/200
48000/48000 [==============================] - 383s - loss: 0.0264 - acc: 0.9904 - val_loss: 0.0221 - val_acc: 0.9940
Epoch 164/200
48000/48000 [==============================] - 396s - loss: 0.0182 - acc: 0.9940 - val_loss: 0.0217 - val_acc: 0.9940
Epoch 165/200
48000/48000 [==============================] - 407s - loss: 0.0179 - acc: 0.9934 - val_loss: 0.0218 - val_acc: 0.9940
Epoch 166/200
48000/48000 [==============================] - 401s - loss: 0.0206 - acc: 0.9930 - val_loss: 0.0232 - val_acc: 0.9920
Epoch 167/200
48000/48000 [==============================] - 399s - loss: 0.0170 - acc: 0.9937 - val_loss: 0.0254 - val_acc: 0.9910
Epoch 168/200
48000/48000 [==============================] - 397s - loss: 0.0175 - acc: 0.9939 - val_loss: 0.0229 - val_acc: 0.9930
Epoch 169/200
48000/48000 [==============================] - 403s - loss: 0.0174 - acc: 0.9937 - val_loss: 0.0232 - val_acc: 0.9920
Epoch 170/200
48000/48000 [==============================] - 397s - loss: 0.0189 - acc: 0.9932 - val_loss: 0.0228 - val_acc: 0.9930
Epoch 171/200
48000/48000 [==============================] - 398s - loss: 0.0210 - acc: 0.9925 - val_loss: 0.0246 - val_acc: 0.9900
Epoch 172/200
48000/48000 [==============================] - 390s - loss: 0.0168 - acc: 0.9940 - val_loss: 0.0221 - val_acc: 0.9940
Epoch 173/200
48000/48000 [==============================] - 392s - loss: 0.0180 - acc: 0.9937 - val_loss: 0.0220 - val_acc: 0.9940
Epoch 174/200
48000/48000 [==============================] - 399s - loss: 0.0169 - acc: 0.9940 - val_loss: 0.0232 - val_acc: 0.9910
Epoch 175/200
48000/48000 [==============================] - 389s - loss: 0.0207 - acc: 0.9924 - val_loss: 0.0222 - val_acc: 0.9940
Epoch 176/200
48000/48000 [==============================] - 396s - loss: 0.0164 - acc: 0.9942 - val_loss: 0.0230 - val_acc: 0.9930
Epoch 177/200
48000/48000 [==============================] - 405s - loss: 0.0178 - acc: 0.9937 - val_loss: 0.0216 - val_acc: 0.9920
Epoch 178/200
48000/48000 [==============================] - 403s - loss: 0.0189 - acc: 0.9932 - val_loss: 0.0304 - val_acc: 0.9880
Epoch 179/200
48000/48000 [==============================] - 401s - loss: 0.0217 - acc: 0.9921 - val_loss: 0.0219 - val_acc: 0.9940
Epoch 180/200
48000/48000 [==============================] - 388s - loss: 0.0159 - acc: 0.9945 - val_loss: 0.0217 - val_acc: 0.9940
Epoch 181/200
48000/48000 [==============================] - 390s - loss: 0.0169 - acc: 0.9941 - val_loss: 0.0218 - val_acc: 0.9920
Epoch 182/200
48000/48000 [==============================] - 406s - loss: 0.0166 - acc: 0.9942 - val_loss: 0.0216 - val_acc: 0.9930
Epoch 183/200
48000/48000 [==============================] - 404s - loss: 0.0151 - acc: 0.9945 - val_loss: 0.0221 - val_acc: 0.9940
Epoch 184/200
48000/48000 [==============================] - 396s - loss: 0.0149 - acc: 0.9947 - val_loss: 0.0222 - val_acc: 0.9930
Epoch 185/200
48000/48000 [==============================] - 393s - loss: 0.0171 - acc: 0.9942 - val_loss: 0.0210 - val_acc: 0.9940
Epoch 186/200
48000/48000 [==============================] - 404s - loss: 0.0174 - acc: 0.9940 - val_loss: 0.0241 - val_acc: 0.9910
Epoch 187/200
48000/48000 [==============================] - 397s - loss: 0.0169 - acc: 0.9940 - val_loss: 0.0241 - val_acc: 0.9910
Epoch 188/200
48000/48000 [==============================] - 393s - loss: 0.0160 - acc: 0.9943 - val_loss: 0.0232 - val_acc: 0.9930
Epoch 189/200
48000/48000 [==============================] - 395s - loss: 0.0189 - acc: 0.9935 - val_loss: 0.0207 - val_acc: 0.9940
Epoch 190/200
48000/48000 [==============================] - 392s - loss: 0.0150 - acc: 0.9946 - val_loss: 0.0212 - val_acc: 0.9940
Epoch 191/200
48000/48000 [==============================] - 401s - loss: 0.0170 - acc: 0.9942 - val_loss: 0.0212 - val_acc: 0.9930
Epoch 192/200
48000/48000 [==============================] - 401s - loss: 0.0159 - acc: 0.9946 - val_loss: 0.0214 - val_acc: 0.9940
Epoch 193/200
48000/48000 [==============================] - 391s - loss: 0.0164 - acc: 0.9942 - val_loss: 0.0242 - val_acc: 0.9920
Epoch 194/200
48000/48000 [==============================] - 401s - loss: 0.0151 - acc: 0.9949 - val_loss: 0.0212 - val_acc: 0.9930
Epoch 195/200
48000/48000 [==============================] - 391s - loss: 0.0160 - acc: 0.9942 - val_loss: 0.0241 - val_acc: 0.9900
Epoch 196/200
48000/48000 [==============================] - 392s - loss: 0.0146 - acc: 0.9945 - val_loss: 0.0203 - val_acc: 0.9940
Epoch 197/200
48000/48000 [==============================] - 392s - loss: 0.0145 - acc: 0.9952 - val_loss: 0.0221 - val_acc: 0.9930
Epoch 198/200
48000/48000 [==============================] - 387s - loss: 0.0166 - acc: 0.9938 - val_loss: 0.0302 - val_acc: 0.9890
Epoch 199/200
48000/48000 [==============================] - 388s - loss: 0.0168 - acc: 0.9945 - val_loss: 0.0216 - val_acc: 0.9930
Epoch 200/200
48000/48000 [==============================] - 412s - loss: 0.0159 - acc: 0.9945 - val_loss: 0.0218 - val_acc: 0.9940






    
Out[27]:





<keras.callbacks.History at 0x7f182815ba50>

In [28]:

    

score = model.evaluate(test_X, test_Y, verbose=0)
print('Test loss:', score[0])
print('Test accuracy:', score[1])


    

    




('Test loss:', 0.021821705218362696)
('Test accuracy:', 0.99399999999999999)

In [29]:

    

model.save('./zheyeV5.keras')