Special issue on modern search heuristics and applications

Abstract

Since its inception in the early 1980s, we have seen a lot of exciting developments in the field of metaheuristics. The complexity of many real-world problems, which are often associated with large search spaces, real-time performance demands and dynamic environments, has made exact solution methods impractical to solve them within a reasonable amount of time. This gives rise to various types of non-exact metaheuristic approaches, including the natureinspired and non nature-inspired ones (see [1–4]). In general, metaheuristics can be viewed as higher level frameworks aimed at efficiently and effectively exploring a search space [5]. Unlike conventional methods, which assume that the objective functions can be solved mathematically, metaheuristics typically do not make much assumption about the problem to be solved or the underlying search space. This makes them applicable to a wide domain of tasks where little information is known about the characteristics of the utility measure. Among the most well-known metaheuristic approaches are those based on the process of natural selection, such as Genetic Algorithms (GA), Genetic Programming (GP), Evolution Strategies (ES), Evolutionary Programming (EP) and Differential Evolution (DE). Other popular metaheuristics include Simulated Annealing that takes inspiration from physics and Swarm Intelligence algorithms such as Ant Colony Optimization (ACO) and Particle Swarm Optimization (PSO) that imitate the social behavior of ants or birds. Scatter Search and Tabu Search are examples of non nature-inspired metaheuristics. These metaheuristics have been applied to areas as diverse as chemistry, computer graphics and visual arts, computer security, data mining, distributed systems, learning and teaching, economics and finance, engineering, healthcare, telecommunication networks, transportation and logistics, manufacturing, military and defense, production, and other combinatorial optimization problems. Ever since the publication of the No Free Lunch theorem [6], a theoretical result proving that the performance of every optimization algorithm over all possible (finite) problems is the same, researchers and practitioners alike have radically changed their view about designing and developing modern search heuristics. Instead of trying to propose universally applicable algorithms, they now start to propose approaches that are tailored to specific problems. Following the trend, this special issue brings together four papers in which the use of metaheuristics in specific application domains is discussed. Three of the papers included were among the invited submissions from the Special Session on Evolutionary Computing that was held at the 9th IEEE International Conference on Cognitive Informatics (ICCI 2010), Tsinghua University, Beijing, China [7]. Each of these papers was substantially revised and extended based on the original conference version. All accepted papers were rigorously reviewed in two rounds by The list of reviewers in alphabetical order: Francisco Chicano, Maurice Clerc, Zong Woo Geem, Guillermo Leguizamon, Ferrante Neri, Albert Orriols-Puig, Rong Qu, Patrick Siarry, Zhenyu Yang, Michael Zapf