Recurring Two-Stage Evolutionary Programming: A Novel Approach for Numeric Optimization

In the application of evolutionary algorithms (EAs) to complex problem solving, it is essential to maintain proper balance between global exploration and local exploitation to achieve a good near-optimum solution to the problem. This paper presents a recurring two-stage evolutionary programming (RTEP) to balance the explorative and exploitative features of the conventional EAs. Unlike most previous works, RTEP is based on repeated and alternated execution of two different stages, namely, the exploration and exploitation stages, each with its own mutation operator, selection strategy, and explorative/exploitative objective. Both analytical and empirical studies have been carried out to understand the necessity of repeated and alternated exploration and exploitation operations in EAs. A suite of 48 benchmark numerical optimization problems has been used in the empirical studies. The experimental results show the remarkable effectiveness of the repeated exploration and exploitation operations employed by RTEP.