Ying-Yang approach to the optimization of fuzzy cerebellar model articulation controller
Minh Nhut Nguyen · 2008
As an associative memory neural network model, the Cerebellar Model Articulation Controller (CMAC) has attractive properties of fast learning speed and simple computation, but its rigid structure makes it difficult to approximate certain functions.In order to overcome this shortcoming, our research aims at the fuzzification phase and the rule weighting process to improve FCMAC by Bayesian Ying-Yang (BYY) learning, cooperative coevolution computation, and online learning.The contributions of this research can be claimed as follows.Firstly, Bayesian Ying-Yang learning is embedded in fuzzy CMAC, named FCMAC-BYY, to find the optimal fuzzy sets in the fuzzification phase.Bayesian Ying-Yang learning is motivated from the famous Chinese ancient Ying-Yang philosophy: everything in the universe can be viewed as a product of a constant conflict between opposites -Ying and Yang, a perfect status is reached if Ying and Yang achieves harmony.Apart from the optimal fuzzy sets systematically obtained by BYY, the proposed FCMAC-BYY also enjoys a consistent rule base, intuitive fuzzy logic reasoning and clear semantic meanings.Secondly, cooperative coevolution computation is integrated into FCMAC-BYY to search for the global solutions.As a matter of fact, FCMAC-BYY suffers from two problems: the fuzzification phase is separated from the learning phase, and the weights of rules are trained by gradient-based methods.Both of these two problems may lead to a local minimum during the learning process.Cooperative coevolutionary learning is hence vii