B-Cell Algorithm as a Parallel Approach to Optimization of Moving Peaks Benchmark Tasks

Efficiency of the B-Cell algorithm applied to one of the well-known test-case generators, the moving peaks benchmark (MPB) is a subject of study presented in this paper. We especially focused on the family of fitness landscapes generated by scenario 2 of the MPB. All of them represent the class of randomly changing environments. Some properties of the algorithm as well as the properties of the environments created by the generator MPB are discussed. A side effect of modification of one of the control parameters and its influence on the offline error measure is presented.