Information: Join today and make your research activities more affordable! Technical workshop participation fees and annual registration fees are available at member rates.
Notice: [Important] Announcement of Changes to Registration Fee Payment and Manuscript Upload Procedures for IEICE Technical Meetings
IEICE Technical Committee Submission System
Conference Paper's Information		 Online Proceedings
[Sign in]
Tech. Rep. Archives
 Go Top Page Go Previous   [Japanese] / [English] 

Paper Abstract and Keywords
Presentation 2007-06-15 09:00
Off-policy least-squares temporal difference learning and its convergence guarantee in finite horizon prorblems
Takeshi Mori, Shin-ichi Maeda, Shin Ishii (NAIST) NC2007-14
Abstract (in Japanese) (See Japanese page) 
(in English) Recently-developed off-policy temporal difference (TD) learning with linear function approximation has attracted attention because of the possibility of sample reuse and dealing effectively with exploration and exploitation. However, the variance of the value function becomes exponentially large as the length of trajectory grows and hence the learning diverges. It is then necessary to truncate the length of trajectory, but the bias of such a finite horizon trajectory can be so harmful that the value function also diverges. Therefore, both in such infinite and finite horizon problems, the off-policy TD learning has no convergence guarantee.
In this study, we propose an off-policy least-squares temporal difference (LSTD) learning and show the convergence in finite horizon problems. Computer simulation shows that our method converges in a finite horizon problem whereas the off-policy TD learning diverges.
Keyword (in Japanese) (See Japanese page) 
(in English) reinforcement learning / off-policy learning / importance sampling / least-squares temporal difference learning / finite horizon problem / / /  
Reference Info. IEICE Tech. Rep., vol. 107, no. 92, NC2007-14, pp. 35-40, June 2007.
Paper # NC2007-14 
Date of Issue 2007-06-07 (NC) 
ISSN Print edition: ISSN 0913-5685    Online edition: ISSN 2432-6380
Copyright
and
reproduction		 All rights are reserved and no part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher. Notwithstanding, instructors are permitted to photocopy isolated articles for noncommercial classroom use without fee. (License No.: 10GA0019/12GB0052/13GB0056/17GB0034/18GB0034)
Download PDF NC2007-14

Conference Information
Committee NC  
Conference Date 2007-06-14 - 2007-06-15 
Place (in Japanese) (See Japanese page) 
Place (in English) OIST Seaside House 
Topics (in Japanese) (See Japanese page) 
Topics (in English)  
Paper Information
Registration To NC 
Conference Code 2007-06-NC 
Language Japanese 
Title (in Japanese) (See Japanese page) 
Sub Title (in Japanese) (See Japanese page) 
Title (in English) Off-policy least-squares temporal difference learning and its convergence guarantee in finite horizon prorblems 
Sub Title (in English)  
Keyword(1) reinforcement learning  
Keyword(2) off-policy learning  
Keyword(3) importance sampling  
Keyword(4) least-squares temporal difference learning  
Keyword(5) finite horizon problem  
Keyword(6)  
Keyword(7)  
Keyword(8)  
1st Author's Name Takeshi Mori  
1st Author's Affiliation Nara Institute of Science and Technology (NAIST)
2nd Author's Name Shin-ichi Maeda  
2nd Author's Affiliation Nara Institute of Science and Technology (NAIST)
3rd Author's Name Shin Ishii  
3rd Author's Affiliation Nara Institute of Science and Technology (NAIST)
4th Author's Name  
4th Author's Affiliation ()
5th Author's Name  
5th Author's Affiliation ()
6th Author's Name  
6th Author's Affiliation ()
7th Author's Name  
7th Author's Affiliation ()
8th Author's Name  
8th Author's Affiliation ()
9th Author's Name  
9th Author's Affiliation ()
10th Author's Name  
10th Author's Affiliation ()
11th Author's Name  
11th Author's Affiliation ()
12th Author's Name  
12th Author's Affiliation ()
13th Author's Name  
13th Author's Affiliation ()
14th Author's Name  
14th Author's Affiliation ()
15th Author's Name  
15th Author's Affiliation ()
16th Author's Name  
16th Author's Affiliation ()
17th Author's Name  
17th Author's Affiliation ()
18th Author's Name  
18th Author's Affiliation ()
19th Author's Name  
19th Author's Affiliation ()
20th Author's Name  
20th Author's Affiliation ()
21st Author's Name  
21st Author's Affiliation ()
22nd Author's Name  
22nd Author's Affiliation ()
23rd Author's Name  
23rd Author's Affiliation ()
24th Author's Name  
24th Author's Affiliation ()
25th Author's Name  
25th Author's Affiliation ()
26th Author's Name / /
26th Author's Affiliation ()
()
27th Author's Name / /
27th Author's Affiliation ()
()
28th Author's Name / /
28th Author's Affiliation ()
()
29th Author's Name / /
29th Author's Affiliation ()
()
30th Author's Name / /
30th Author's Affiliation ()
()
31st Author's Name / /
31st Author's Affiliation ()
()
32nd Author's Name / /
32nd Author's Affiliation ()
()
33rd Author's Name / /
33rd Author's Affiliation ()
()
34th Author's Name / /
34th Author's Affiliation ()
()
35th Author's Name / /
35th Author's Affiliation ()
()
36th Author's Name / /
36th Author's Affiliation ()
()
Speaker Author-1 
Date Time 2007-06-15 09:00:00 
Presentation Time 25 minutes 
Registration for NC 
Paper # NC2007-14 
Volume (vol) vol.107 
Number (no) no.92 
Page pp.35-40 
#Pages 6 
Date of Issue 2007-06-07 (NC) 

[Return to Top Page]

[Return to IEICE Web Page]
The Institute of Electronics, Information and Communication Engineers (IEICE), Japan