|
|
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.
|
|
|
Rozpoznávání pozic deskové hry go z fotografií
Musil, Tomáš ; Baudiš, Petr (advisor) ; Hauzar, David (referee)
It is customary to keep a written game record of professional or high-rank amateur tournament games of Go. Even informal games are worth recording for subsequent analysis. Writing the game record by hand distracts the player from the game and it is not very reliable. Video or photographic record lacks the flexibility of abstract notation. In this thesis we discuss several ways of automatically extracting Go game records from photographs. We propose our own method based on Hough transform and RANSAC paradigm. We implement a reliable and easy to use system that allows players to take a game record effortlessly. Powered by TCPDF (www.tcpdf.org)
|
|
| |
|
|
Rozpoznávání pozic deskové hry go z fotografií
Musil, Tomáš ; Baudiš, Petr (advisor) ; Hauzar, David (referee)
It is customary to keep a written game record of professional or high-rank amateur tournament games of Go. Even informal games are worth recording for subsequent analysis. Writing the game record by hand distracts the player from the game and it is not very reliable. Video or photographic record lacks the flexibility of abstract notation. In this thesis we discuss several ways of automatically extracting Go game records from photographs. We propose our own method based on Hough transform and RANSAC paradigm. We implement a reliable and easy to use system that allows players to take a game record effortlessly. Powered by TCPDF (www.tcpdf.org)
|
|
|
Efficient Network Backup System
Filípek, Tomáš ; Baudiš, Petr (advisor) ; Marek, Lukáš (referee)
There are elegant algorithms for storing similar files, as well as for sending them efficiently through a network. But can these algorithms be combined to form a back-up system? We have constructed such a system using the client-server architecture. Its main features include support for versioning and intelligent history keeping. To assess the right values for some of the algorithm parameters, a simple performance testing was performed. Using the chosen values, the application is reasonably efficient both in space and time.
|
|
|
Arimaa challenge - static evaluation function
Hřebejk, Tomáš ; Majerech, Vladan (advisor) ; Baudiš, Petr (referee)
Arimaa is a strategic board game for two players. It was designed with the aim that it will be hard to create a computer program that could defeat the best human players. In this thesis, we focus on the design of the static evaluation function for Arimaa. The purpose of a static evaluation function is to determine which player is leading in a given position and how significant the lead is. We have divided the problem into a few parts, which were solved separately. We paid most attention to the efficient recognition of important patterns on the board, such as goal threats. The basic element of the proposed evaluation function is mobility. For each piece, the number of steps that the piece would need to get to other places on the board is estimated. We also examined machine learning. We developed a new algorithm for learning a static evaluation function from expert games. An implementation of an Arimaa playing program, which demonstrates the proposed methods, is part of the thesis. Powered by TCPDF (www.tcpdf.org)
|
|
|
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 ::
RSS feed.