Differential Evolution With Adaptive Guiding Mechanism Based on Heuristic Rules

Abstract

This paper proposes to resolve the limitation of differential evolution (DE) that the difference between the individuals in search behavior has not yet been utilized effectively for guiding the evolution of the population. An adaptive guiding mechanism (AGM) based on the heuristic rules is thus suggested to make possible, individual-dependent guidance. The AGM mainly comprises three stages: construction, separation, and guidance. In the construction stage, the elite leadership team (ELT ) is established with an adaptive control scheme by using good information of the population. In the separation stage, the ELT is divided into distinct elite groups that are allocated to different individuals based on their search behaviors. In the guidance stage, the leader that is chosen from the respective elite group, as well as the promising directions extracted from the population, are used together to guide the search of each individual. By incorporating AGM into DE, a novel algorithm framework, named DE with AGM (DE-AGM), is proposed to enhance the performance of DE. As a general framework, DE-AGM can be easily and seamlessly applied to most DE variants. The experimental results on 58 benchmark functions have demonstrated the competitive performance of DE-AGM.

Differential Evolution with Level-Based Learning Mechanism

“…The three individuals for the mutation strategy are randomly selected from three parts, respectively. Unlike in the above DE variants that utilize the poor individuals to maintain diversity, the local elite is selected [35] . Thus, for each individual, two elites are selected from its neighbors and the whole population, respectively, to balance diversity and convergence.…”

“…Seven peer DE algorithms including EAGDE [70] , EFADE [71] , AMECoDEs [35] , TSDE [56] , RNDE [21] , MPEDE [45] , and TVDE [72] , are used for comparison. EAGDE adopts the fitness-based population structure; EFADE designs the triangular mutation operator to balance diversity and convergence; AMECoDEs and RNDE adopt the neighbor-based population structure, and elite information is used; TSDE is a two-stage algorithm that adopts multiple strategies to enhance the search abilities of the population; MPEDE is a multipopulation algorithm, and it self-adaptively adjusts the utilization of multiple strategies; and TVDE designs a time-varying strategy to gradually increase the utilization of excellent individuals.…”

“…AMECoDEs [35] p ε = 0.1, = 0.001 RNDE [21] F = 0.5 TSDE [56] F CR F CR F CR [ = 1, = 0.1], [ = 1, = 0.9], [ = 0.8, = 0.2] MPEDE [45] λ ng = 0.2, = 20 TVDE [72] Freq G = 0.05, = 10 000 0.00×10 0 ± 0.00×10 0 F7 1.76×10 1 ± 3.77×10 0 (+)…”

“…Unlike the traditional random selection method, the new method prevents good individuals from being influenced by poor individuals. To clarify the difference between them, the diversity ( ) and success rate ( ) in the generation are calculated [35] ,…”

Elite Representative Based Individual Adaptive Regeneration Framework for Differential Evolution

“…CMA-ES [37] introduced Covariance Matrix Adaptation Evolution Strategy and more recently, Mohamed et al [38] proposed LSHADE-SPACMA as a hybridization framework between L-SHADE [25] and CMA-ES. In [39], Cai et al proposed an adaptive guiding mechanism (AGM) based on the heuristic rules as a novel framework for DE to make individual-dependent guidance possible. DE-AGM consists of three stages: the elite leadership team (ELT) is created in the construction stage and divided and allocated to respective individuals during the separation stage.…”