Memetic Computing (2021) 13:147–148 https://doi.org/10.1007/s12293-021-00337-6

GUEST EDITORIAL

Guest Editorial: Special issue on memetic with learning strategy

Ling Wang1 · Liang Feng2

Published online: 19 May 2021 © The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature 2021

Memetic (MA) represents one of the successful optimization and decision-making, have to date received far extension of (EA). The term MA is less attention. now widely used as a synergy of evolutionary or any pop- This special issue aims to explore and encourage the cur- ulation-based approach with separate individual learning rent and ongoing research progress in memetic computation, or local improvement procedures for optimization. During especially the design and utilization of learning strategy in the last decade, it has been demonstrated that MAs are able MAs towards automated and advanced memetic optimi- to converge to high quality solutions more efciently than zation process. Following a rigorous peer review process, their conventional counterparts on a wide range of real world seven papers have been accepted to be included in the spe- problems. Despite the success and surge in interests on MAs, cial issue. many successful MAs reported in the literature have been The frst paper, “A light-robust-optimization model and crafted to suit problems in very specifc domains. an efective memetic algorithm for an open vehicle routing Nowadays, it is well recognized that the process of learn- problem under uncertain travel times” by Sun et al. presents ing is central to humans in problem-solving. Learning has an efective MA for solving the open vehicle routing prob- been established to be fundamental to human in functioning lem with predetermined time windows under uncertain travel and adapting to the fast evolving society. To enhance the times (OVRP-UT). In the proposed method, learning strat- optimization capability when solving complex problems, egy has been proposed to adaptively control the frequency of it is very important to use learning strategy and control performing crossover and mutation during the evolutionary scheme in MAs. Particular examples may include the learn- search process. New initialization, crossover and mutation ing of adaptive approach to control the confguration of local operators have also been proposed in the MA regarding the searches in MA along the evolutionary search, the learn- property of OVRP-UT. Comprehensive empirical study on ing of historical successes in algorithm confguration and 320 benchmark instances have been conducted to investigate high quality solutions to enhance the MA search on unseen the performance of the proposed method. problems, reinforcement learning assisted MA, deep learn- In the second paper entitled “An efective memetic algo- ing assisted MA, etc. Within the context of computational rithm for UAV routing and orientation under uncertain navi- intelligence, several core learning technologies in fuzzy sys- gation environments”, Shang et al. propose an efective MA tem and neural network have been notable for the ability in for unmanned aerial vehicle (UAV) routing and orientation modeling human’s leaning and generalization capabilities under navigational, steering and uncertain constraints. In the for dealing with complex real-world applications. In spite proposed MA, the global search performs the outer loop for of the accomplishments made in computational intelligence, optimizing the route, while the local search the attempts to emulate the learning mechanisms of human focuses on the inner loop for optimizing the orientations. in search, MAs in particular, for intelligent and automated A database recording knowledge of high-quality subroutes along the search process is used to accelerate the inner opti- * mization in the MA. Experiments on open-access datasets Liang Feng show the efectiveness of the proposed MA for providing [email protected] high-quality route with orientations for UAV. 1 Department of Automation, Tsinghua University, Beijing, To improve the operational efciency of supply chain, an China adaptive human-learning-based for solv- 2 College of , Chongqing University, ing the integrated production and distribution scheduling Chongqing, China

Vol.:(0123456789)1 3 148 Memetic Computing (2021) 13:147–148 problem has been proposed by Qin et al. in their paper “Inte- present an adaptive multi-objective evolutionary algorithm grated Production and Distribution Scheduling in Distrib- with the concepts of the grid system. Based on grid, the uted Hybrid Flow Shops”. The proposed method hybridizes objective space is divided into subspaces, which are adap- an adaptive learning operator with crossover and mutation tively selected for allocating search resources based on the to enhance the global and local search ability. quality and dominance relationship between subspaces. The forth paper, “Integrated scheduling problem for earth Last but not the least, Peng et al. in their paper “A novel observation satellites based on three modeling frameworks: non-dominated sorting genetic algorithm for solving the an adaptive bi-objective memetic algorithm” by Chang et al. triple objective project scheduling problem”, addressed the embarks a study on the scheduling of data acquisition and multi-mode resource constrained project scheduling prob- data transmission for earth observation satellites. In particu- lem with three objectives, which are minimizing the project lar, an adaptive bi-objective MA which integrates the non- duration, minimizing the resource investment and maximiz- dominated sorting genetic algorithm II (NSGA-II) and an ing the robustness of the schedule. To solve this problem, adaptive large neighborhood local search algorithm has been this paper proposed a modifed serial schedule generation proposed in this paper. The efcacy of the proposed method scheme and a simplifed normalization method in the non- has been extensively investigated using real-world data sets. dominated sorting genetic algorithm III (NSGA-III). To solve the challenging bilevel optimization problems, The guest editors would like to thank all the authors who Wu et al. proposed an efcient self-adaptive bilevel diferen- submitted their work to the special issue, and all reviewers tial evolution (SABiLDE) with kNN approximation for the for their hard work in completing timely and constructive lower level optimization in “An efcient bilevel diferential reviews. Special thanks also go to the Editors-in-Chief and evolution algorithm with adaptation of lower level popu- members of the editorial team for their support during the lation size and search radius”. The proposed self-adaptive editing process of this Special Issue. control rate and the introduction of the archiving technique play key roles in improving the efciency of the optimization algorithm. The performance of the proposed algorithm has Publisher’s Note been evaluated on 10 standard bilevel test problems and the Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional afliations. benchmark suite with 12 scalable test problems. In the sixth paper entitled “An Adaptive Multiobjective Evolutionary Algorithm Based on Grid Subspaces”, to solve multi-objective optimization problems efciently, Li et al.

1 3