Opponent Modeling in Adversarial Environments through Learning Ingenuity

Authors:
Arash Afkanpour;Saeed Bagheri Shouraki
Affiliations:
Computer Engineering Department, Sharif University of Technology, Tehran, Iran, {afkanpour, sbagheri}@ce.sharif.edu;Computer Engineering Department, Sharif University of Technology, Tehran, Iran, {afkanpour, sbagheri}@ce.sharif.edu
Venue:
Proceedings of the 2005 conference on Self-Organization and Autonomic Informatics (I)
Year:
2005

Citing 7
Cited 0

Technical Note: \cal Q-Learning

Machine Learning
Numerical recipes in C (2nd ed.): the art of scientific computing

Numerical recipes in C (2nd ed.): the art of scientific computing
Reinforcement Learning

Reinforcement Learning
Friend-or-Foe Q-learning in General-Sum Games

ICML '01 Proceedings of the Eighteenth International Conference on Machine Learning
Artificial Intelligence: A Modern Approach

Artificial Intelligence: A Modern Approach
Multiagent learning in the presence of agents with limitations

Multiagent learning in the presence of agents with limitations
Reinforcement learning: a survey

Journal of Artificial Intelligence Research

Quantified Score

Hi-index	0.01

Visualization

Abstract

In Multiagent systems there are several agents with cooperative or competitive goals. Here, we are especially interested in zero-sum games which contain exactly two players with fully opposite goals. We describe a method based on Maximum-Expected-Utility [7] principle that learns the ingenuity of the opponent based on the moves of the opponent through a game and exploits this knowledge to play better against that opponent. Then we demonstrate an application of proposed method in the popular board game of Connect-4. The results show that the proposed method is superior compared to previous methods for adversarial environments especially when there is not adequate training for appropriate adaptation against an opponent.