Title :
Fuzzy advantage learning
Author :
Yan, X.W. ; Deng, Z.D. ; Sun, Z.-Q.
Author_Institution :
State Key Lab. of Intelligent Technol. & Syst., Tsinghua Univ., Beijing, China
Abstract :
Based on advantage learning and fuzzy inference systems (FISs), fuzzy advantage learning (FAL) is presented. In FAL, a FIS is applied to introduce strong generalization ability and to generate continuous actions for advantage learning. This approach is illustrated and compared to other reinforcement learning methods. The simulation study on inverted pendulum control shows its effectiveness with faster learning speed
Keywords :
fuzzy logic; fuzzy systems; generalisation (artificial intelligence); inference mechanisms; learning (artificial intelligence); nonlinear control systems; pendulums; continuous actions; fuzzy advantage learning; fuzzy inference systems; inverted pendulum control; learning speed; reinforcement learning methods; strong generalization ability; Delay effects; Dynamic programming; Fuzzy neural networks; Fuzzy systems; Intelligent systems; Laboratories; Learning systems; Neural networks; State-space methods; Sun;
Conference_Titel :
Fuzzy Systems, 2000. FUZZ IEEE 2000. The Ninth IEEE International Conference on
Conference_Location :
San Antonio, TX
Print_ISBN :
0-7803-5877-5
DOI :
10.1109/FUZZY.2000.839145
