Probability redistribution using time hopping for reinforcement learning

File Description SizeFormat 
Kormushev_ISIS-2009.pdfAccepted version387.93 kBAdobe PDFView/Open
Title: Probability redistribution using time hopping for reinforcement learning
Authors: Kormushev, P
Dong, F
Hirota, K
Item Type: Conference Paper
Abstract: —A method for using the Time Hopping technique as a tool for probability redistribution is proposed. Applied to reinforcement learning in a simulation, it is able to re-shape the state probability distribution of the underlying Markov decision process as desired. This is achieved by modifying the target selection strategy of Time Hopping appropriately. Experiments with a robot maze reinforcement learning problem show that the method improves the exploration efficiency by re-shaping the state probability distribution to an almost uniform distribution.
Issue Date: 17-Aug-2009
Date of Acceptance: 17-Aug-2009
URI: http://hdl.handle.net/10044/1/26089
Journal / Book Title: Proc. 10th International Symposium on Advanced Intelligent Systems, ISIS 2009
Copyright Statement: © 2009 The Authors
Conference Name: 10th International Symposium on Advanced Intelligent Systems, ISIS 2009
Publication Status: Published
Start Date: 2009-08-17
Finish Date: 2009-08-19
Conference Place: Busan, Korea
Appears in Collections:Dyson School of Design Engineering



Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.

Creative Commonsx