摘要
对不变性的认识,特别是对变换群下的不变性的认知,一直是神经网络系统研究中最重要也是最困难的问题之一。本文首先介绍了用神经网络来实现变换群下的不变性主要方法,然后对这些方法进行了简要的评述,最后提出了一些我们自己的看法。
The problem of invariance is both important and difficult in studies of neural networks. In this paper, four typical methods for invariance implementation under transform group using neural networks are introduced and the corresponding discussions are carried out. Finally, the prospects of the problem are outlined.
出处
《系统工程与电子技术》
EI
CSCD
1992年第12期50-55,共6页
Systems Engineering and Electronics