Coupling genetic algorithms and rule-based systems for complex decisions

摘要

Due to the increasing need of supporting complex decisions in an effective manner, emergent approaches have been proposed to combine different AI technologies to form hybrid systems. Although hybrid systems are still largely experimental, many preliminary results indicate that such coupling can enable more complex problem-solving techniques than have been available to decision makers in the past. In this research, we design an interactive, Genetic Algorithms (GA)-based hybrid system for the support of the batch selection decision. We utilize a GA's stochastic search capability in identifying promising areas in the solution space. Then, the rule-based system is applied to reinforce the feasibility of the GA-identified potential solutions. Our experimental results indicate that our hybrid system has better performance than both the pure GA system and the rule-based system for complex decision-making.