Bruno Bouzy: Associating Shallow and Selective Global Tree Search with Monte Carlo for 9*9 Go. Computers and Games Bruno Bouzy of Paris Descartes, CPSC, Paris (Paris 5) with expertise in: Artificial Intelligence. Read 73 publications, and contact Bruno Bouzy on ResearchGate. Bruno Bouzy is a player and programmer from France. Born in , his highest rank was 3 dan. He was vice champion of France, losing in the.

Author: Kazisar Fektilar
Country: Guadeloupe
Language: English (Spanish)
Genre: Sex
Published (Last): 25 April 2016
Pages: 108
PDF File Size: 13.10 Mb
ePub File Size: 17.81 Mb
ISBN: 151-5-44250-256-6
Downloads: 54090
Price: Free* [*Free Regsitration Required]
Uploader: Kigakree

Programming backgammon using self-teaching neural nets.

In incomplete information games, such as poker [Billings et al. Besides, Olga has a fuzzy and optimistic definition: Amongst them, Monte-Carlo led to promising … More.

Let us start the random games from the root by two given moves, one move for the friendly side, and, then, one move for the opponent, and make statistics on the terminal position evaluation for each node situated at depth 2 in the min-max tree. In this context, each module is independent of the other one, and does not use the strength of the other one. Simulated annealing normally has an evaluation that depends only on the current state in the case of Gobble, a state is the lists of moves for both players ; instead in Gobble the evaluation of a state is the average of all the random games that are based on all the states reached so far.

Discussion This section discusses the strengths and weaknesses of the statistical ap- proach and opens up some promising perspectives. Damien Pellier 1 AuthorId: Conversely, it looks for weaknesses in the opponent position that do not exist. This idea constitutes the cornerstone of our bokzy. Its main difficulty in terms of game programming is the huge branching factor. We have addressed two problems due to the use of transpositions.


This phenomenon happens when captures have already occurred at the time when the move is played. Nevertheless, this approach looks promising. The diagonal intersections do not intervene in these definitions since we do not want to insert too much domain-dependent knowledge into the program. Compared to the very slow basic idea, the gain in speed is important. This paper experimentally evaluates multiagent learning algorithms playing repeated matrix games to maximize their cumulative return.

Computer Science > Artificial Intelligence

In figure 3 the point C is good for white and bad for black. As suggested by [Bruegmann, ], a first experiment would be to make the random program use patterns giving the probability of a move advised by the pattern. Experiments Starting from the basic idea, this section describes the various enhancements with their effect on the level of go programs: However, it does have some drawbacks because the evaluation of a move from a random game in which it was played at a late stage is less reliable than when it is played at an early stage.

Patterns are generated by browsing recorded games of professional … More.

There was a problem providing the content you requested

Log In Sign Up. Click here to sign up. For instance, it would suppress moves that work well only when the opponent does not respond. How can this failure be explained? The time to carry out these tests is proportional to the time spent to play one random game. This result underlines that the use of transpositions significantly speeds up the program but decreases its performance.

Bruno Bouzy – Semantic Scholar

For the latter, an intermediate possibility can be adopted: Oleg was written by Bernard Helmstetter. Another idea would consist in making the two modules interact. Game-tree searching by min-max approximation. The only domain-dependent knowledge required is the definition of an eye.

Since Oleg is boouzy to play Thus, in the beginning, the games were almost completely random, and at the end they were almost completely determined by the evaluations of the moves. This paper experimentally shows that well-selected hedging algorithms are better on average than all previous MAL algorithms on the task of playing RMG against various players. This is an adaptation of [Abramson, ]. From a practical point of view, how does this boouzy to the level of play?


The re- sult out of 14 9×9 games has been an average 9. Olga and Oleg are still inferior to them but we believe that, with the help of the ever-increasing power of computers, this approach is promising for computer go in the future. In these two cases, the move with the highest expected outcome is chosen.

Remember me on this computer.

dblp: Bruno Bouzy

The idea is attractive, because one random game helps evaluate almost all possible moves at the root. The author showed that the expected outcome is a powerful heuristic. This paper underlines the association of two computer go approaches, a domaindependent knowledge approach and Monte Carlo.

Create your web page Haltools: This approach is based on statistics or Monte Carlo methods. Progressive pruning loses little strength compared to the basic idea.

It has already been considered theoretically within the framework of [Rivest, ]. Prospective methods of programming the game of Go will probably be of interest in other domains as well. They are shown in figure 1. Jaap Van Den Herik. Bruon game of Go is one of the games that still withstand classical Artificial Intelligence approaches.