An Agent for EinStein Würfelt Nicht! Using N-Tuple Networks
Authors:
- 1 December 2017
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 184-189
Abstract
This paper describes the implemention of an agent that plays EinStein würfelt nich!. The agent is based on the common Monte-Carlo tree search (MCTS) which is especially good at dealing with the randomness in a game. For the agent, this paper proposes to use n-tuple networks trained by Monte-Carlo learning. In the agent, the trained n-tuple networks is used together with MCTS by the following three approaches: progressive bias, prior knowledge and ε-greedy. The experimental results show that ε-greedy improved the playing strength the most, which obtained a win rate of 61.05% against the baseline agent. By combining all three approaches, the win rate increased a little to 62.25%. And the enhanced agent tournament in Computer Olympiad 2017.Keywords
This publication has 8 references indexed in Scilit:
- An analysis for strength improvement of an MCTS-based program playing Chinese dark chessTheoretical Computer Science, 2016
- Online Adaptable Learning Rates for the Game Connect-4IEEE Transactions on Computational Intelligence and AI in Games, 2014
- Temporal difference learning of N-tuple networks for the game 20482014
- EinStein Würfelt Nicht – An Analysis of Endgame PlayICGA Journal: The Journal of the Computer Games Community, 2012
- A Survey of Monte Carlo Tree Search MethodsIEEE Transactions on Computational Intelligence and AI in Games, 2012
- PROGRESSIVE STRATEGIES FOR MONTE-CARLO TREE SEARCHNew Mathematics and Natural Computation, 2008
- Combining online and offline knowledge in UCT2007
- TD-Gammon: A Self-Teaching Backgammon Program1995