Genetic Takagi-Sugeno fuzzy reinforcement learning

Abstract

This paper presents two fuzzy reinforcement learning methods for solving complicated learning tasks of continuous domains. Takagi-Sugeno fuzzy reinforcement learning (TSFRL) is constructed by combining Takagi-Sugeno type fuzzy inference systems with Q-learning. Next, genetic Takagi-Sugeno fuzzy reinforcement learning (GTSFRL) is introduced by embedding TSFRL into genetic algorithms. Both proposed learning algorithms can also be used to design Takagi-Sugeno fuzzy logic controllers. Experiments on the double inverted pendulum system demonstrate the performance and applicability of the proposed schemes. Finally, the conclusion remark is drawn.