An Analysis of Majority Voting in Homogeneous Groups for Checkers: Understanding Group Performance Through Unbalance
Experimental evidence and theoretical advances over the years have created an academic consensus regarding majority voting systems, namely that, under certain conditions, the group performs better than its components. However, the underlying reason for such conditions, e.g., stochastic independence of agents, is not often explored and may help to improve performance in known setups by changing agent behavior, or find new ways of combining agents to take better advantage of their characteristics. In this work, an investigation is conducted for homogeneous groups of independent agents playing the game of Checkers. The analysis aims to find the relationship between the change in performance caused by majority voting, the group size, and the underlying decision process of each agent, which is mapped to its source of non-determinism. A characteristic unbalance in Checkers, due to an apparent initiative disadvantage, serves as a pivot for the study, from which decisions can be separated into beneficial or detrimental biases. Experimental results indicate that performance changes caused by majority voting may be beneficial or not, and are linked to the game properties and player skill. Additionally, a way of improving agent performance by manipulating its non-determinism source is briefly explored.
- 9.Obata, T., Sugiyama, T., Hoki, K., Ito, T.: Consultation algorithm for computer Shogi: move decisions by majority. In: van den Herik, H.J., Iida, H., Plaat, A. (eds.) CG 2010. LNCS, vol. 6515, pp. 156–165. Springer, Heidelberg (2011). https://doi.org/10.1007/978-3-642-17928-0_15 CrossRefGoogle Scholar
- 10.Hoki, K., Kaneko, T., Yokoyama, D., Obata, T., Yamashita, H., Tsuruoka, Y., Ito, T.: A system-design outline of the distributed-Shogi-system Akara 2010. In: Proceedings of the 14th ACIS International Conference on Software Engineering, Artificial Intelligence, Networking and Parallel/Distributed Computing (SNPD). IEEE, pp. 466–471 (2013)Google Scholar
- 11.Hoki, K., Omori, S., Ito, T.: Analysis of performance of consultation methods in computer chess. J. Inf. Sci. Eng. 30, 701–712 (2014)Google Scholar
- 14.Sato, Y., Cincotti, A., Iida, H.: An analysis of voting algorithm in games. In: Computer Games Workshop at European Conference on Artificial Intelligence, ECAI, pp. 102–113 (2012)Google Scholar