A Neighborhood-Assisted Framework for Differential Evolution

Abstract

Differential evolution (DE), as a powerful and efficient evolutionary algorithm (EA), has shown its advantages in solving the complex optimization problems. In the literature, the utilization of neighborhood information has been attracting wide attention in the DE community due to its effectiveness in enhancing the search ability of DE. However, we have observed that no general framework is presented to provide a comprehensive way of studying the neighborhood-based DE variants. Therefore, this paper suggests a threelayer mechanism neighborhood-assisted (TLNA) DE framework to facilitate the utilization of neighborhood information. In TLNA, the mechanisms of using neighborhood information are generalized into a three-layer cooperative structure, i.e., the interaction mechanism (IM) layer, the organization mechanism (OM) layer, and utilization mechanism (UM) layer. Thus, TLNA is built to provide a synergistic effect of different layers of mechanisms for systematically utilizing neighborhood information. As a general framework, TLNA can be realized with different implementations of the three-layer mechanism. Furthermore, to demonstrate the practicality of the proposed framework, a TLNA instantiation (iTLNA) is given in detail. The performance of iTLNA is extensively evaluated on a suite of benchmark functions. The experimental results have confirmed the competitiveness of iTLNA to other DE variants and EAs, which shows that the proposed TLNA framework can pave an effective way to improve the performance of DE with neighborhood information.

Elite Representative Based Individual Adaptive Regeneration Framework for Differential Evolution

“…During the guidance stage, the leader is selected and combined with promising difference vectors to guide the mutation procedure. A three-layer mechanism neighborhood-assisted (TLNA) DE framework is proposed in [40] to utilize neighborhood information systematically.…”

Differential Evolution With Adaptive Guiding Mechanism Based on Heuristic Rules

“…It makes these strategies be good at local exploration but easily lead to premature convergence. Based on these considerations, various approaches have been proposed to enhance the search ability of the mutation operator for different complex problems, which roughly fall into the following categories: designing new mutation strategies [21], [22], integrating multiple mutation strategies [9], [10], [23], and selecting parent vectors for mutation [13]- [15], [24]. These works related to the mutation operator of DE will be reviewed in Section III.…”

“…In most DE variants, however, the difference between individuals in search behavior has not yet been effectively utilized for guiding the evolution of population. As shown in [14], [20], and [24]- [26], different individuals have distinct effects on the evolution of population. The individuals with better fitness values can guide the population towards the more promising regions, while the individuals with worse fitness values can explore new search space for keeping populations diversity [26].…”