LEARNING THE WEIGHTS OF WEIGHTED FUZZY IF-THEN RULES VIA TRAINING T-S NORMS NEURAL NETWORK

摘要：

In this paper, an approach of learning the values of the weights in weighted fuzzy if-then rules is presented. Based on the concept of T-S norms, firstly, this paper presents the T-S norm-based fuzzy reasoning algorithm; secondly, we map a set of initial fuzzy fi-then rules, in which all the weights are equal to 1.0, and the T-S norm-based fuzzy reasoning methodology into a forward fuzzy neural network, named T-S norm neural network, and the nodes of hidden layer are T norm neural cells, while the nodes of output layer are S norm neural cells;finally, we complete the training of the T-S norm neural network via a derived T-S norm BP algorithm. The experimental results have shown that our approach can learn the weights of weighted fuzzy if-then rules efficiently.

关键词： Local weight Global weight Weighted fuzzy fi-then rules T-S norms

作者: CHUN-RU DONG XI-ZHAO WANG XIAO-DONG DAI

作者单位: Faculty of Mathematics and Computer Science, Hebei University, B aoding 071002, China

会议类型: 国际会议

会议名称: 2006 International Conference on Machine Learning and Cybernetics(IEEE第五届机器学习与控制论坛)

会议地点: 大连

会议语种:英文

页码: 2920-2924

在线出版日期: 2006-08-13（万方平台首次上网日期，不代表论文的发表时间）