Tuning evaluation functions by maximizing concordance

Gomboc, Dave; Buro, Michael; Marsland, T.A.

doi:10.1016/j.tcs.2005.09.047

Cited by 16 publications

(11 citation statements)

References 48 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Except for not using the desired moves, Buro's method has properties that are similar to those listed in Table 1; his objective function has continuity as well as an assured local minimum, and his method is scalable. Gomboc, Buro, and Marsland (2005) proposed to learn from game records annotated by human experts; however, the feature weights that were adjusted in their experiments were only a small part of the full evaluation functions. Reinforcement learning (Sutton & Barto, 1998), especially temporal difference learning, of which a famous success is Backgammon (Tesauro, 2002), is considered to be promising way to avoid the difficulty in finding the desired values for regression.…”

Section: Other Methods Of Learning Evaluation Functionsmentioning

confidence: 99%

Large-Scale Optimization for Evaluation Functions with Minimax Search

Hoki¹,

Kaneko²

2014

jair

View full text Add to dashboard Cite

This paper presents a new method, Minimax Tree Optimization (MMTO), to learn a heuristic evaluation function of a practical alpha-beta search program. The evaluation function may be a linear or non-linear combination of weighted features, and the weights are the parameters to be optimized. To control the search results so that the move decisions agree with the game records of human experts, a well-modeled objective function to be minimized is designed. Moreover, a numerical iterative method is used to find local minima of the objective function, and more than forty million parameters are adjusted by using a small number of hyper parameters. This method was applied to shogi, a major variant of chess in which the evaluation function must handle a larger state space than in chess. Experimental results show that the large-scale optimization of the evaluation function improves the playing strength of shogi programs, and the new method performs significantly better than other methods. Implementation of the new method in our shogi program Bonanza made substantial contributions to the program's first-place finish in the 2013 World Computer Shogi Championship. Additionally, we present preliminary evidence of broader applicability of our method to other two-player games such as chess.

show abstract

Section: Other Methods Of Learning Evaluation Functionsmentioning

confidence: 99%

Large-Scale Optimization for Evaluation Functions with Minimax Search

Hoki¹,

Kaneko²

2014

jair

View full text Add to dashboard Cite

show abstract

“…Man vs Machine games have become scarcer. There was an annual event in Bilbao called "People vs Computers", but the results in 2005 were extremely favorable to computer programs (Levy, 2005). David Levy, who was the referee of the match, even suggested that games should be played with odds and the event was apparently canceled the next year.…”

Section: Evaluation Of the Elo Strength Of The Program Usedmentioning

confidence: 99%

“…Finding such a mapping is however not a real problem for chess programmers because their problem is more to find a good ranking of the moves in a given position than an evaluation of the probability of winning the game, which has no direct practical interest. See for example (Gomboc, Buro, and Marsland, 2005) for the problem of tuning evaluation functions.…”

Section: The Experimental Settingsmentioning

confidence: 99%

Who is the Master?

Alliot

2017

ICG

View full text Add to dashboard Cite

Abstract. There has been debates for years on how to rate chess players living and playing at different periods (see Keene and Divinsky (1989)). Some attempts were made to rank them not on the results of games played, but on the moves played in these games, evaluating these moves with computer programs. However, the previous attempts were subject to different criticisms, regarding the strengths of the programs used, the number of games evaluated, and other methodological problems.In the current study, 26,000 games (over 2 millions of positions) played at regular time control by all world champions since Wilhelm Steinitz have been analyzed using an extremely strong program running on a cluster of 640 processors. Using this much larger database, the indicators presented in previous studies (along with some new, similar, ones) have been correlated with the outcome of the games. The results of these correlations show that the interpretation of the strength of players based on the similarity of their moves with the ones played by the computer is not as straightforward as it might seem. Then, to overcome these difficulties, a new Markovian interpretation of the game of chess is proposed, which enables to create, using the same database, Markovian matrices for each year a player was active. By using classical linear algebra methods on these matrices, the outcome of games between any players can be predicted, and this prediction is shown to be at least as good as the classical ELO prediction for players who actually played against each others.

show abstract

“…However, the domains where such analyses can be applied are limited. Similarly, we can see how the evaluation values for each position produced by an evaluation function agree on the preferences of human players, if positions with the assessments made by human players are available [21]. The applicability of this method is limited to domains in which such assessments can be carried out.…”

Section: B Accuracy Of Game-tree Searchmentioning

confidence: 99%

Evaluation of Monte Carlo tree search and the application to Go

Takeuchi

Kaneko

Yamaguchi

2008

2008 IEEE Symposium on Computational Intelligence and Games

View full text Add to dashboard Cite

Abstract-Recent improvements to Monte Carlo tree search have produced strong computer Go programs. This paper presents a method of measuring the accuracy of Monte Carlo tree search in game programming. We use the win percentage of positions in a large database of game records as a benchmark and compare the win probability obtained by simulations with the benchmark. By applying our method to Monte Carlo tree search in Go, we found differences between search methods and their parameters, and the effect of the properties of positions such as the move numbers and the existence of stones in threats. This paper also introduces numerical metrics to evaluate the performance of search methods. Our experiments in Go, as well as Chess, Othello, and Shogi revealed that the metrics were quite close to our empirical understanding of the performance of various search methods and their parameters.

show abstract

Tuning evaluation functions by maximizing concordance

Cited by 16 publications

References 48 publications

Large-Scale Optimization for Evaluation Functions with Minimax Search

Large-Scale Optimization for Evaluation Functions with Minimax Search

Who is the Master?

Evaluation of Monte Carlo tree search and the application to Go

Contact Info

Product

Resources

About