mnist_cnn_all.py

from __future__ import absolute_import
from __future__ import print_function
from keras.datasets import mnist
from keras.models import Sequential
from keras.layers.core import Dense, Dropout, Activation, Flatten
from keras.layers.convolutional import Convolution2D, MaxPooling2D
from keras.utils import np_utils
from scipy.special import expit
import numpy as np
np.set_printoptions(suppress=True)
np.set_printoptions(precision=6)

'''
    Train a simple convnet on the MNIST dataset.

    Run on GPU: THEANO_FLAGS=mode=FAST_RUN,device=gpu,floatX=float32 python mnist_cnn.py

    Get to 99.25% test accuracy after 12 epochs (there is still a lot of margin for parameter tuning).
    16 seconds per epoch on a GRID K520 GPU.
'''

TRAIN = True
batch_size = 128
nb_classes = 10
nb_epoch = 5

# the data, shuffled and split between tran and test sets
(X_train, y_train), (X_test, y_test) = mnist.load_data()

X_train = X_train.reshape(X_train.shape[0], 1, 28, 28)
X_test = X_test.reshape(X_test.shape[0], 1, 28, 28)
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 = np_utils.to_categorical(y_train, nb_classes)
Y_test = np_utils.to_categorical(y_test, nb_classes)

model = Sequential()
model.add(Convolution2D(32, 1, 3, 3, border_mode='valid'))
model.add(Activation('relu'))
model.add(Convolution2D(32, 32, 3, 3))
model.add(Activation('relu'))
model.add(MaxPooling2D(poolsize=(2, 2)))
model.add(Dropout(0.25))
model.add(Convolution2D(128, 32, 12, 12))
model.add(Activation('relu'))
model.add(Dropout(0.5))
model.add(Convolution2D(nb_classes, 128, 1, 1))
model.add(Flatten())
model.add(Activation('sigmoid'))

model.compile(loss='categorical_crossentropy', optimizer='adadelta')

if TRAIN:
	model.fit(X_train, Y_train, batch_size=batch_size, nb_epoch=nb_epoch, show_accuracy=True, verbose=1, validation_data=(X_test, Y_test))
	score = model.evaluate(X_test, Y_test, show_accuracy=True, verbose=0)
	print('Test score:', score[0])
	print('Test accuracy:', score[1])

	model.save_weights('/tmp/model_weights.hdf5', overwrite=True)
else:
	model.load_weights('/tmp/model_weights.hdf5')

# create mosaic image with twice the same digit
data = np.zeros([1,1,28, 28*2], 'float32')
data[0,0,:,:28] = X_train[0,0,:,:]
data[0,0,:,28:] = X_train[0,0,:,:]

print('real label: %d' % y_train[0])
r0 = (model.predict(data[:,:,:,:28])).ravel()

print('expected result (centered):', r0, '(label=%d)' % np.argmax(r0))
pred = model.predict(data).ravel()
pred = pred.reshape(nb_classes, len(pred)/nb_classes).T
for i,p in enumerate(pred):
	print('pos:%02d' % i, p, '(label=%d)' % np.argmax(p))

What's this import for?

from scipy.special import expit

Doesn't seem to get used anywhere...
Sorry for the n00b question, but I'm new to both Keras and ANNs but not to Python, so this is a shallow question but would help me understand one more bit...

Hi Guy,
It is indeed useless here, used in other parts of our library, left by mistake.
Victor