Fuzzy inference neural networks which automatically partition a pattern space and extract fuzzy if - then rules

This paper proposes Fuzzy Inference Neural Networks (FINNs) which automatically partition a pattern space and extract fuzzy if-then rules from numerical data. There are three distinctive features in our model: 1) the membership functions of the fuzzified part are constructed in the connection between the input-part and the rule-layer; 2) Kohonen's self-organizing algorithm is applied to partition the input-output space. Consequently, they can extract polished fuzzy if-then rules; 3) they can adapt the number of rules automatically. We deal with two illustrative examples: 1) fuzzy control of unmanned vehicle; 2) prediction of the trend of stock prices. Computer simulation results indicate the effectiveness of the proposed FINNs.