def unit_test2():
mlp = MLP([1,2,1]);
mlp.layers[1].set_params([[-.3],[4]],[.15,-2])
mlp.layers[2].set_params([[0,2]],[-1])
num_labels=1
data= [1.0, 1.0]
predicted = mlp.predict(data[:-num_labels])
print 'predicted: ', predicted
mlp.output_error_calculation(data[-num_labels:])
mlp.backpropagation()
print 'errors:'
for layer in mlp.layers:
print layer.e
mlp.gradient_descend(.5)
print 'weights:'
for layer in mlp.layers:
print zip( layer.b, layer.weight)
def unit_test():
mlp = MLP([2,2,2,2]);
mlp.layers[1].set_params([[.2,-.1],[.3,-.3]],[.1,-.2])
mlp.layers[2].set_params([[-.2,-.3],[-.1,.3]],[.1,.2])
mlp.layers[3].set_params([[-.1,.3],[-.2,-.3]],[.2,.1])
num_labels=2
data= [.3, .7, .1, 1.0]
predicted = mlp.predict(data[:-num_labels])
print 'predicted: ', predicted
mlp.output_error_calculation(data[-num_labels:])
mlp.backpropagation()
print 'errors:'
for layer in mlp.layers:
print layer.e
mlp.gradient_descend(.1)
print 'weights:'
for layer in mlp.layers:
print zip( layer.b, layer.weight)