Test the regularisation parameter on the convolutional Layer for keras

test_conv_reg.py

# Test the regularisation parameter on the convolutional Layer

from keras.datasets import mnist
from keras.models import Sequential
from keras.layers.convolutional import Convolution2D
from keras.layers.core import Dense,Flatten
from keras.utils import np_utils
from keras.regularizers import l2

import numpy as np

np.random.seed(1337)

max_train_samples = 128*4
max_test_samples = 1000

(X_train, y_train), (X_test, y_test) = mnist.load_data()

X_test = X_test.reshape(-1,1,28,28)
X_train = X_train.reshape(-1,1,28,28)

nb_classes = len(np.unique(y_train))
Y_train = np_utils.to_categorical(y_train, nb_classes)[:max_train_samples]
Y_test = np_utils.to_categorical(y_test, nb_classes)[:max_test_samples]

model_noreg = Sequential()
model_noreg.add(Convolution2D(1, 1, 20, 20))
model_noreg.add(Flatten())
model_noreg.add(Dense(9*9, 10))

model_noreg.compile(loss='categorical_crossentropy', optimizer='rmsprop')
model_noreg.fit(X_train, Y_train)

score_noreg = model_noreg.evaluate(X_test, Y_test)
score_train_noreg = model_noreg.evaluate(X_train, Y_train)

model = Sequential()
model.add(Convolution2D(1, 1, 20, 20, W_regularizer=l2(0.05)))
model.add(Flatten())
model.add(Dense(9*9, 10))

model.compile(loss='categorical_crossentropy', optimizer='rmsprop')
model.fit(X_train, Y_train)

score_reg = model.evaluate(X_test, Y_test)
score_train_reg = model.evaluate(X_train, Y_train)

print
print
print "Overfitting without regularisation: %f - %f = %f" % ( score_noreg , score_train_noreg , score_noreg-score_train_noreg)
print "Overfitting with regularisation: %f - %f = %f" % ( score_reg , score_train_reg , score_reg-score_train_reg)