Layer.conv2d(input_layer=model[0],
kernel_shape=(5, 5),
num_kernels=16,
stride=1,
padding=2))
model.append(Layer.pool2d(input_layer=model[1], pool_shape=(2, 2), stride=2))
model.append(
Layer.conv2d(input_layer=model[2],
kernel_shape=(5, 5),
num_kernels=32,
stride=1,
padding=2))
model.append(Layer.pool2d(input_layer=model[3], pool_shape=(2, 2), stride=2))
model.append(Layer.dense(input_layer=model[4], num_nodes=64, rectified=True))
model.append(
Layer.dense(input_layer=model[5], num_nodes=10, rectified=False)
) # We need linear support vector machine output layer | Set rectified to false during training
# set learning rates per layer
model[1].learning_rate = 0.02
model[2].learning_rate = 0.02
model[3].learning_rate = 0.02
model[4].learning_rate = 0.02
model[5].learning_rate = 0.02
model[6].learning_rate = 0.02
#traing model
X = np.array([[0,0],
[1,0],
[0,1],
[1,1]])
Y = np.array([[1],
[0],
[0],
[1]])
np.random.seed(1) # seed random number generator for reproducible results
# Set up model layers
model.append(Layer._input_layer(input_shape = (2,1)))
model.append(Layer.dense(input_layer = model[0], num_nodes = 2, rectified = True))
model.append(Layer.dense(input_layer = model[1], num_nodes = 1, rectified = False)) # We need linear support vector machine output layer | Set rectified to false during training
model[1].learning_rate = 0.02
model[2].learning_rate = 0.02
# train model using the BSSP learning algorithm
Model.sgd_bssp_train(model, X, Y, 5000)
# rectify the output layer | This line can be commented out
model[len(model) - 1].rectified = True
# Test model
for x in X:
output = Model.predict(model, x)