A Study of Multistage Inference Rules Based on Neural Networks

• Main Authors: Li Kai-Ming, 李凱名
• Other Authors: 孫光天
• Format: Others
• Language: zh-TW
• Published: 2001
• Online Access: http://ndltd.ncl.edu.tw/handle/06010187419260265843

碩士 === 臺南師範學院 === 資訊教育研究所 === 89 === Fuzzy inference rules have been developed to a wide variety of applications successfully and popularly. However, the basic principle of the fuzzy inference rules is also obeyed the inference rules. It can not separate the patterns into any graph, when the patterns are distributed disorderly. The neural networks can separate any distributed patterns well, but it can not be presented by inference rules. In this study, We will design a multistage neural network for representing inference rules. In this way, the multistage inference rules can separate the patterns into non-linear segmentations with any graph, and the accuracy rate can be increased.