Practical Issues in Temporal Difference Learning Export

@article{tesauro1992, abstract = {This paper examines whether temporal difference methods for training connectionist networks, such as Sutton''s TD(lambda) algorithm, can be successfully applied to complex real-world problems. A number of important practical issues are identified and discussed from a general theoretical perspective. These practical issues are then examined in the context of a case study in which TD(lambda) is applied to learning the game of backgammon from the outcome of self-play. This is apparently the first application of this algorithm to a complex non-trivial task. It is found that, with zero knowledge built in, the network is able to learn from scratch to play the entire game at a fairly strong intermediate level of performance, which is clearly better than conventional commercial programs, and which in fact surpasses comparable networks trained on a massive human expert data set. This indicates that TD learning may work better in practice than one would expect based on current theory, and it suggests that further analysis of TD methods, as well as applications in other complex domains, may be worth investigating.}, address = {Hingham, MA, USA}, author = {Tesauro, Gerald}, citeulike-article-id = {465792}, citeulike-linkout-0 = {http://portal.acm.org/citation.cfm?id=139611.139616}, citeulike-linkout-1 = {http://dx.doi.org/10.1007/BF00992697}, citeulike-linkout-2 = {http://www.springerlink.com/(hx2gkz55d5pg2if5litnzl55)/app/home/contribution.asp?referrer=parent&backto=issue,3,8;journal,74,124;linkingpublicationresults,1:400309,1}, doi = {10.1007/BF00992697}, issn = {0885-6125}, journal = {Machine Learning}, keywords = {reinforcement, td}, month = {May}, number = {3-4}, pages = {257--277}, posted-at = {2006-01-15 18:36:51}, priority = {2}, publisher = {Kluwer Academic Publishers}, title = {Practical Issues in Temporal Difference Learning}, url = {http://dx.doi.org/10.1007/BF00992697}, volume = {8}, year = {1992} }

TY - JOUR ID - tesauro1992 L3 - citeulike-article-id:465792 N2 - This paper examines whether temporal difference methods for training connectionist networks, such as Sutton''s TD(lambda) algorithm, can be successfully applied to complex real-world problems. A number of important practical issues are identified and discussed from a general theoretical perspective. These practical issues are then examined in the context of a case study in which TD(lambda) is applied to learning the game of backgammon from the outcome of self-play. This is apparently the first application of this algorithm to a complex non-trivial task. It is found that, with zero knowledge built in, the network is able to learn from scratch to play the entire game at a fairly strong intermediate level of performance, which is clearly better than conventional commercial programs, and which in fact surpasses comparable networks trained on a massive human expert data set. This indicates that TD learning may work better in practice than one would expect based on current theory, and it suggests that further analysis of TD methods, as well as applications in other complex domains, may be worth investigating. IS - 3-4 JF - Machine Learning SN - 0885-6125 AD - Hingham, MA, USA EP - 277 TI - Practical Issues in Temporal Difference Learning PB - Kluwer Academic Publishers VL - 8 SP - 257 KW - reinforcement KW - td AU - Tesauro, Gerald PY - 1992/05// UR - http://dx.doi.org/10.1007/BF00992697 ER -