Guest Editorial: Special issue on memetic algorithms with learning strategy

Ling Wang, Liang Feng
2021 Memetic Computing  
Memetic algorithm (MA) represents one of the successful extension of evolutionary algorithm (EA). The term MA is now widely used as a synergy of evolutionary or any population-based approach with separate individual learning or local improvement procedures for optimization. During the last decade, it has been demonstrated that MAs are able to converge to high quality solutions more efficiently than their conventional counterparts on a wide range of real world problems. Despite the success and
more » ... rge in interests on MAs, many successful MAs reported in the literature have been crafted to suit problems in very specific domains. Nowadays, it is well recognized that the process of learning is central to humans in problem-solving. Learning has been established to be fundamental to human in functioning and adapting to the fast evolving society. To enhance the optimization capability when solving complex problems, it is very important to use learning strategy and control scheme in MAs. Particular examples may include the learning of adaptive approach to control the configuration of local searches in MA along the evolutionary search, the learning of historical successes in algorithm configuration and high quality solutions to enhance the MA search on unseen problems, reinforcement learning assisted MA, deep learning assisted MA, etc. Within the context of computational intelligence, several core learning technologies in fuzzy system and neural network have been notable for the ability in modeling human's leaning and generalization capabilities for dealing with complex real-world applications. In spite of the accomplishments made in computational intelligence, the attempts to emulate the learning mechanisms of human in search, MAs in particular, for intelligent and automated optimization and decision-making, have to date received far less attention. This special issue aims to explore and encourage the current and ongoing research progress in memetic computation, especially the design and utilization of learning strategy in MAs towards automated and advanced memetic optimization process. Following a rigorous peer review process, seven papers have been accepted to be included in the special issue. The first paper, "A light-robust-optimization model and an effective memetic algorithm for an open vehicle routing problem under uncertain travel times" by Sun et al. presents an effective MA for solving the open vehicle routing problem with predetermined time windows under uncertain travel times (OVRP-UT). In the proposed method, learning strategy has been proposed to adaptively control the frequency of performing crossover and mutation during the evolutionary search process. New initialization, crossover and mutation operators have also been proposed in the MA regarding the property of OVRP-UT. Comprehensive empirical study on 320 benchmark instances have been conducted to investigate the performance of the proposed method. In the second paper entitled "An effective memetic algorithm for UAV routing and orientation under uncertain navigation environments", Shang et al. propose an effective MA for unmanned aerial vehicle (UAV) routing and orientation under navigational, steering and uncertain constraints. In the proposed MA, the global search performs the outer loop for optimizing the route, while the local search metaheuristic focuses on the inner loop for optimizing the orientations. A database recording knowledge of high-quality subroutes along the search process is used to accelerate the inner optimization in the MA. Experiments on open-access datasets show the effectiveness of the proposed MA for providing high-quality route with orientations for UAV. To improve the operational efficiency of supply chain, an adaptive human-learning-based genetic algorithm for solving the integrated production and distribution scheduling
doi:10.1007/s12293-021-00337-6 fatcat:q3gg4vzrnfbh3lrx7s45onuewu