Some Studies in Machine Learning Using the Game of Checkers. II—Recent Progress

Samuel, Arthur L.

doi:10.1007/978-1-4613-8716-9_15

Cited by 63 publications

(34 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the absence of theoretical guidance, we now empirically examine the previously discussed practical issues within the context of a specific application: learning to play the game of backgammon from the outcome of self-play. Complex board games such as checkers, chess, Othello and backgammon have been widely studied as testing grounds for various machine learning procedures (Samuel, 1959;Samuel, 1967;Griffith, 1974;Quinlan, 1983;Mitchell, 1984;Frey, 1986;Christensen & Korf, 1986;Lee & Mahajan, 1988;Tesauro & Sejnowski, 1989;Tesauro, 1990). Several of these studies have employed temporal difference learning methods.…”

Section: O a Case Study: Td Learning Of Backgammon Strategymentioning

confidence: 99%

Practical issues in temporal difference learning

1992

View full text Add to dashboard Cite

Abstract. This paper examines whether temporal difference methods for training connectionist networks, such as Sutton's TD(X) 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(X) 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.

show abstract

Section: O a Case Study: Td Learning Of Backgammon Strategymentioning

confidence: 99%

Practical issues in temporal difference learning

1992

View full text Add to dashboard Cite

show abstract

“…Samuel [18,19] describes the use of rote-learning and argues that a program can learn to play a domain better than the creator. A lookahead of two or three plays is used to find moves to be scored for a checker game.…”

Section: Related Workmentioning

confidence: 99%

Evaluating the Effectiveness of Exploration and Accumulated Experience in Automatic Case Elicitation

Powell

Hauff

Hastings

2005

Case-Based Reasoning Research and Development

View full text Add to dashboard Cite

Abstract. Non-learning problem solvers have been applied to many interesting and complex domains. Experience-based learning techniques have been developed to augment the capabilities of certain non-learning problem solvers in order to improve overall performance. An alternative approach to enhancing preexisting systems is automatic case elicitation, a learning technique in which a case-based reasoning system with no prior domain knowledge acquires knowledge automatically through real-time exploration and interaction with its environment. In empirical testing in the domain of checkers, results suggest not only that experience can substitute for the inclusion of pre-coded model-based knowledge, but also that the ability to explore is crucial to the performance of automatic case elicitation.

show abstract

“…But note that significant behavioral changes can be achieved within this simple framework. Samuel's (1959Samuel's ( , 1967 checker player is a striking example of the power of such an approach. If one views the adjustable weights and thresholds as parameters of a structurally-fixed neural network, then much of the research on neural net learning also falls into this category.…”

Section: Using Genetic Algorithms To Change Parametersmentioning

confidence: 99%

Untitled

Jong

1988

Machine Learning

View full text Add to dashboard Cite

Abstract. Genetic algorithms represent a class of adaptive search techniques that have been intensively studied in recent years. Much of the interest in genetic algorithms is due to the fact that they provide a set of efficient domain-independent search heuristics which are a significant improvement over traditional "weak methods" without the need for incorporating highly domain-specific knowledge. There is now considerable evidence that genetic algorithms are useful for global function optimization and NP-hard problems. Recently, there has been a good deal of interest in using genetic algorithms for machine learning problems. This paper provides a brief overview of how one might use genetic algorithms as a key element in learning systems.

show abstract

Some Studies in Machine Learning Using the Game of Checkers. II—Recent Progress

Cited by 63 publications

References 0 publications

Practical issues in temporal difference learning

Practical issues in temporal difference learning

Evaluating the Effectiveness of Exploration and Accumulated Experience in Automatic Case Elicitation

Untitled

Contact Info

Product

Resources

About