In [1]:

    

'''Trains a simple convnet on the MNIST dataset for ONLY digits 3 and 8.
Gets to 98.25% test accuracy after 12 epochs
(there is still a lot of margin for parameter tuning).
4 seconds per epoch on a 2 GHz Intel Core i5.
'''

from __future__ import print_function
import keras
from keras.datasets import mnist
from keras.models import Sequential
from keras.layers import Dense, Dropout, Flatten
from keras.layers import Conv2D, MaxPooling2D
from keras import backend as K
import numpy as np

batch_size = 128
num_classes = 2
epochs = 12

# input image dimensions
img_rows, img_cols = 28, 28

# the data, shuffled and split between train and test sets
(x_train, y_train), (x_test, y_test) = mnist.load_data()

#Only look at 3s and 8s
train_picks = np.logical_or(y_train==2,y_train==7)
test_picks = np.logical_or(y_test==2,y_test==7)

x_train = x_train[train_picks]
x_test = x_test[test_picks]
y_train = np.array(y_train[train_picks]==7,dtype=int)
y_test = np.array(y_test[test_picks]==7,dtype=int)


if K.image_data_format() == 'channels_first':
    x_train = x_train.reshape(x_train.shape[0], 1, img_rows, img_cols)
    x_test = x_test.reshape(x_test.shape[0], 1, img_rows, img_cols)
    input_shape = (1, img_rows, img_cols)
else:
    x_train = x_train.reshape(x_train.shape[0], img_rows, img_cols, 1)
    x_test = x_test.reshape(x_test.shape[0], img_rows, img_cols, 1)
    input_shape = (img_rows, img_cols, 1)

x_train = x_train.astype('float32')
x_test = x_test.astype('float32')
x_train /= 255
x_test /= 255
print('x_train shape:', x_train.shape)
print(x_train.shape[0], 'train samples')
print(x_test.shape[0], 'test samples')

# convert class vectors to binary class matrices
y_train = keras.utils.to_categorical(y_train, num_classes)
y_test = keras.utils.to_categorical(y_test, num_classes)

model = Sequential()
model.add(Conv2D(4, kernel_size=(3, 3),activation='relu',input_shape=input_shape))
model.add(Conv2D(8, (3, 3), activation='relu'))
model.add(MaxPooling2D(pool_size=(2, 2)))
model.add(Dropout(0.25))
model.add(Flatten())
model.add(Dense(16, activation='relu'))
model.add(Dropout(0.5))
model.add(Dense(2, activation='softmax'))

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

model.fit(x_train, y_train,
          batch_size=batch_size,
          epochs=epochs,
          verbose=1,
          validation_data=(x_test, y_test))
score = model.evaluate(x_test, y_test, verbose=0)
print('Test loss:', score[0])
print('Test accuracy:', score[1])


    

    




Using TensorFlow backend.






    




Downloading data from https://s3.amazonaws.com/img-datasets/mnist.npz
x_train shape: (12223, 28, 28, 1)
12223 train samples
2060 test samples
Train on 12223 samples, validate on 2060 samples
Epoch 1/12
12223/12223 [==============================] - 5s - loss: 0.2756 - acc: 0.8985 - val_loss: 0.0760 - val_acc: 0.9748
Epoch 2/12
12223/12223 [==============================] - 5s - loss: 0.0835 - acc: 0.9741 - val_loss: 0.0698 - val_acc: 0.9767
Epoch 3/12
12223/12223 [==============================] - 5s - loss: 0.0670 - acc: 0.9797 - val_loss: 0.0715 - val_acc: 0.9772
Epoch 4/12
12223/12223 [==============================] - 6s - loss: 0.0560 - acc: 0.9818 - val_loss: 0.0628 - val_acc: 0.9816
Epoch 5/12
12223/12223 [==============================] - 5s - loss: 0.0586 - acc: 0.9831 - val_loss: 0.0566 - val_acc: 0.9835
Epoch 6/12
12223/12223 [==============================] - 6s - loss: 0.0519 - acc: 0.9840 - val_loss: 0.0589 - val_acc: 0.9811
Epoch 7/12
12223/12223 [==============================] - 6s - loss: 0.0503 - acc: 0.9858 - val_loss: 0.0536 - val_acc: 0.9845
Epoch 8/12
12223/12223 [==============================] - 6s - loss: 0.0503 - acc: 0.9850 - val_loss: 0.0553 - val_acc: 0.9835
Epoch 9/12
12223/12223 [==============================] - 6s - loss: 0.0457 - acc: 0.9867 - val_loss: 0.0525 - val_acc: 0.9840
Epoch 10/12
12223/12223 [==============================] - 5s - loss: 0.0474 - acc: 0.9850 - val_loss: 0.0509 - val_acc: 0.9840
Epoch 11/12
12223/12223 [==============================] - 5s - loss: 0.0461 - acc: 0.9867 - val_loss: 0.0493 - val_acc: 0.9845
Epoch 12/12
12223/12223 [==============================] - 5s - loss: 0.0449 - acc: 0.9882 - val_loss: 0.0490 - val_acc: 0.9850
Test loss: 0.049024477387
Test accuracy: 0.984951456311

In [ ]: