Bounds on the number of hidden units in binary-valued three-layer neural networks

✍ Scribed by Masahiko Arai

• Publisher: Elsevier Science
• Year: 1993
• Tongue: English
• Weight: 461 KB
• Volume: 6
• Category: Article
• ISSN: 0893-6080
• DOI: 10.1016/s0893-6080(05)80130-3

✦ Synopsis

For three-layer art~l~cial neural networks (TANs) that take binao, vahtes, the number of hidden units is considered regarding two problems: One is to find the necessary and sufficient number to make mapping between the binary output values of TANs and learning patterns (inputs) arbitrary; and the other is to get the sufficient numberJbr two-categoo, classification (TCC) problems. We show that for the.former I -1 hidden units are necessary and sufficient for I learning patterns and that for the latter about 1/3 hidden units are sufficient. These results mean that we can reduce the necessary mtmber of hidden units by taking into account the features of learning pattern distributions.