Title :
Threshold learning in the improved penalty avoiding rational policy making algorithm
Author :
Miyazaki, Kazuteru ; Kobayashi, Ryouhei ; Kobayashi, Hiroaki
Author_Institution :
Dept. of Assessment & Res. for Degree Awarding, Univ. Evaluation, Tokyo, Japan
Abstract :
The penalty avoiding rational policy making algorithm (PARP) previously improved to save memory and cope with uncertainty, i.e., Improved PARP (IPARP). The efficiency of IPARP is influenced by threshold of a penalty rule or a penalty basis function γ significantly. In this paper, we propose a technique for learning γ. We show the effectiveness of our proposal using a soccer game task called “Keepaway”.
Keywords :
game theory; learning (artificial intelligence); PARP; keepaway; penalty avoiding rational policy making algorithm; soccer game; threshold learning; Function approximation; Games; Machine learning; Memory management; Proposals; Tiles; Uncertainty; Exploitation-oriented Learning XoL; Improved PARP; Keepaway Task; Reinforcement Learning; Threshold Learning;
Conference_Titel :
SICE Annual Conference 2010, Proceedings of
Conference_Location :
Taipei
Print_ISBN :
978-1-4244-7642-8
