PaDE-NPC: Parameter Adaptive Differential Evolution With Novel Parameter Control for Single-Objective Optimization

Single objective real-parameter optimization problems exist in many areas of the real world, and Differential Evolution (DE) is a powerful population based stochastic optimization approach for tackling such problems. There are many different mutation strategies mentioned in the literature, and each of them has its own advantage. In this paper, we propose combined mutation strategies which can make a full use of the advantage of each mutation strategy regarding a population diversity indicator during the evolution. Furthermore, Novel Parameter Control (NPC) for the three control parameters including the scale factor <inline-formula> <tex-math notation="LaTeX">$F$ </tex-math></inline-formula>, crossover rate <inline-formula> <tex-math notation="LaTeX">$CR$ </tex-math></inline-formula> and population size <inline-formula> <tex-math notation="LaTeX">$PS$ </tex-math></inline-formula> are also proposed in the paper. Different from employing the fitness value as a weight in recent proposed state-of-the-art DE variants, our PaDE-NPC algorithm can tackle a large optimization problems especially for those the fitness differences are unavailable; Moreover, a platform based population size reduction scheme is also involved in the NPC, which can get a better perception of the landscape at the early stage of the evolution while obtaining a balance between exploration and exploitation in the later part of the evolution. The novel PaDE-NPC algorithm is verified under 58 benchmark functions from CEC2013 and CEC2017 test suits for real-parameter optimization competitions and experiment results show that our proposed PaDE-NPC algorithm outperforms these recently proposed powerful DE variants.

Abstract licence: CC BY-NC-ND

A person counts once a day per paper. We keep no cookie and no address: a reader is a hash that changes every day. Any month under five reads shows no number.