end += size
return [x_train, y_train, x_test, y_test]
kfold = kfoldSplit(X, Y, k)
cv_scores = [] # store the score
batch_size = [64]
epoch = range(30, 31)
model = ['CNN_model_LeNet', 'CNN_model_simple']
if __name__ == '__main__':
for e in epoch:
for size in batch_size:
for i in range(k):
print(i, kfold[0][i].shape)
model = CNN.LeNet(kernel_size=(3, 3), activation='relu')
model.compile(
loss='categorical_crossentropy',
optimizer=CNN.adam,
metrics=['accuracy'] # 评价函数
)
model.fit(kfold[0][i],
kfold[1][i],
epochs=e,
batch_size=size,
verbose=0)
score = model.evaluate(kfold[2][i], kfold[3][i], verbose=0)
cv_scores.append(score[1] * 100)
print("%s: %.2f%%" % (model.metrics_names[1], score[1] * 100))
print(
