This is because the only parameters that are adjusted in the learning process are the linear mapping from hidden layer to output layer.

• 这是因为唯一在学习过程中被调整过的参数是那此从隐藏层到输出层的线性映射。