|
|
MCTS with Information Sharing
Baudiš, Petr ; Hric, Jan (advisor) ; Majerech, Vladan (referee)
We introduce our competitive implementation of a Monte Carlo Tree Search (MCTS) algorithm for the board game of Go: Pachi. The software is based both on previously published methods and our original improvements. We then focus on improving the tree search performance by collecting information regarding tactical situations and game status from the Monte Carlo simulations and sharing it with and within the game tree. We propose specific methods of such sharing --- dynamic komi, criticality-based biasing, and liberty maps --- and demonstrate their positive effect. based on collected play-testing measurements. We also outline some promising future research directions related to our work.
|
|
|
Artificial intelligence in abstract 2-player games
Veselý, Pavel ; Valla, Tomáš (advisor) ; Baudiš, Petr (referee)
In this thesis we focus on algorithms for searching for the best move in a given position in an abstract strategy 2-player game. We describe algorithms Alpha-beta and Proof-number Search with their enhancements and we come with new ideas for making them quicker. We also propose an algorithm for choosing randomly between moves not much worse than the best move and ideas how to play in lost positions. We apply the algorithms on the game Tzaar which is special for having a lot of possible moves which makes the game hard for a computer. Our goal is to create a robot for playing Tzaar as good as possible. We show that our artificial intelligence can play on the level of best human and computer players. We also examine experimentally how enhancements of the algorithms help making computations quicker in this game.
|
|
|
MCTS with Information Sharing
Baudiš, Petr ; Hric, Jan (advisor) ; Majerech, Vladan (referee)
We introduce our competitive implementation of a Monte Carlo Tree Search (MCTS) algorithm for the board game of Go: Pachi. The software is based both on previously published methods and our original improvements. We then focus on improving the tree search performance by collecting information regarding tactical situations and game status from the Monte Carlo simulations and sharing it with and within the game tree. We propose specific methods of such sharing --- dynamic komi, criticality-based biasing, and liberty maps --- and demonstrate their positive effect. based on collected play-testing measurements. We also outline some promising future research directions related to our work.
|
|
|
Artificial intelligence in abstract 2-player games
Veselý, Pavel ; Valla, Tomáš (advisor) ; Baudiš, Petr (referee)
In this thesis we focus on algorithms for searching for the best move in a given position in an abstract strategy 2-player game. We describe algorithms Alpha-beta and Proof-number Search with their enhancements and we come with new ideas for making them quicker. We also propose an algorithm for choosing randomly between moves not much worse than the best move and ideas how to play in lost positions. We apply the algorithms on the game Tzaar which is special for having a lot of possible moves which makes the game hard for a computer. Our goal is to create a robot for playing Tzaar as good as possible. We show that our artificial intelligence can play on the level of best human and computer players. We also examine experimentally how enhancements of the algorithms help making computations quicker in this game.
|
guest ::
