A Hybrid Approach for Automatic Generation of Fuzzy Inference Systems without Supervised Learning

A hybrid approach with dynamic self-generated fuzzy q-learning (DSGFQL) and genetic algorithms (GA) for automatic generation of fuzzy inference systems (FISs) termed evolutionary dynamic self-generated fuzzy inference systems (EDSGFIS) is proposed in this paper. The structure and parameters of an FIS are generated through a dynamic self- generated fuzzy q-learning (DSGFQL) while an evolutive action set for the consequents of the FIS is obtained via GA. Contribution of this paper is that the EDSGFIS algorithm suggests a heuristic approach to organize the structure of an FIS and adjust the parameters based on the reinforcement only and without supervised learning (SL). GA is adopted here to obtain a satisfactory set of actions for the training of the DSGFQL methodology. Moreover, a hierarchical learning structure is proposed to reduce the computational cost and increase the speed of learning. The proposed EDSGFIS algorithm can automatically create, delete and adjust fuzzy rules according to the performance of the entire system as well as the evaluation of individual fuzzy agents. Simulation studies on a wall-following task by a mobile robot show the superiority of the proposed approach. Further discussions on the proposed approach are presented in this work.