【读书1】【2017】MATLAB与深度学习——单层神经网络的训练：增量规则(1)

根据给定信息修改权重的系统方法称为学习规则。

The systematic approach to modifying theweights according to the given information is called the learning rule.

由于训练是神经网络将信息系统存储的唯一途径，因此，学习规则是神经网络研究的一个重要组成部分。

Since training is the only way for theneural network to store the information systematically, the learning rule is avital component in neural network research.

在这一部分中，我们研究增量规则，一种单层神经网络的学习规则。

In this section, we deal with the deltarule, the representative learning rule of the single-layer neural network.

增量规则也被称为Adaline规则或者Widrow-Hoff规则。

It is also referred to as Adaline rule aswell as Widrow-Hoff rule.

虽然增量规则不能进行多层神经网络训练，但对于研究神经网络学习规则这一重要概念非常有用。

Although it is not capable of multi-layerneural network training, it is very useful for studying the important conceptsof the learning rule of the neural network.

考虑如图2-11所示的单层神经网络。

Consider a single-layer neural network, asshown in Figure 2-11.

——本文译自Phil Kim所著的《Matlab Deep Learning》

更多精彩文章请关注微信号：