RETRACTED: A Rule Based Evolutionary Algorithm for Intelligent Decision Support

Abstract

Genetic or evolutionary search algorithms seek and exploit the structure of a problem by operating on an encoding which represents the problem variables. The algorithms employed are generally designed to handle a wide variety of problems and while exploitation and progress can be rapid at the initial stages, the algorithms ultimate convergence rate can be slow. In order to speed up the solution process as well as produce a more refined solution, a rule based encoding is proposed. The rule based structure injects domain specific knowledge into the optimization process. This allows for an intuitive encoding and mimics the process utilized in many decision support applications such as scheduling. The encoding for the approach is often reduced in size as well. Specific rules for a particular problem class are coded into the formulation. This requires additional programming effort, but is valuable for specific applications which are repeated over time. During the solution process, a record is maintained concerning which rules or sequences of rules were successfully applied. This allows the rules to be continuously updated over time. An approach for implementing rules within the design encoding is demonstrated and several simple game problems are solved using this technique. The results are compared to a solution generated by a traditional encoding.