Function Optimization using Evolutionary Game Theory Algorithm

There are many models that can be used to make intelligent behaviors in solving complex problems. Game theory is one of them and can be used efficiently in decision making. In this paper, we study an evolutionary game theory algorithm (EGTA) and tested the algorithm in optimization function. It works with a set of players (i.e., solutions) and used an expected payoff mechanism. The payoff estimation mechanism makes a player being able to make a decision rationally. On the other hand, it generates new offspring using the imitation operator and belief-learning operator. The imitation operator tries to learn from other player's strategies; one player updates its quality by strategically learning from another better player. Belief learning is a strategy where a player tries to improve its chromosome/solution by estimation and analyze previous information. The algorithm has been tested in solving several benchmark function optimization problems and compared with the other four methods. The game theory-based algorithm outperformed other algorithms on accuracy and stability.