DocumentCode
3374596
Title
Historical Temporal Difference Learning: Some Initial Results
Author
Yao, Hengshuai ; Dongcui, Diao ; Sun, Zengqi
Author_Institution
State Key Lab. of Intelligent Control & Syst., Tsinghua Univ., Beijing
Volume
2
fYear
2006
fDate
20-24 June 2006
Firstpage
678
Lastpage
685
Abstract
In this paper, we develop a multi-step prediction algorithm that is guaranteed to converge when using general function approximation. Besides, the new algorithm should satisfy the following requirements: first, it does not have to be faster than TD(0) in the look-up table representation; however, the new algorithm should be faster than residual gradient method. Second, the new algorithm should learn optimally
Keywords
function approximation; gradient methods; learning (artificial intelligence); table lookup; function approximation; look-up table representation; multistep prediction algorithm; residual gradient method; temporal difference learning; Approximation algorithms; Dynamic programming; Function approximation; Intelligent control; Machine learning; Machine learning algorithms; State-space methods; Sun; Table lookup; Vectors; Multi-step Prediction; Reinforcement Learning; Temporal Difference Learning;
fLanguage
English
Publisher
ieee
Conference_Titel
Computer and Computational Sciences, 2006. IMSCCS '06. First International Multi-Symposiums on
Conference_Location
Hanzhou, Zhejiang
Print_ISBN
0-7695-2581-4
Type
conf
DOI
10.1109/IMSCCS.2006.231
Filename
4673785
Link To Document