![]()
Chapter 1
Introduction to Hybrid Metaheuristics
Sandip Dey
Department of Computer Science & Engineering,
OmDayal Group of Institutions,
Birshibpur,
Howrah-711316,
India[email protected] Sourav De
Department of Computer Science & Engineering,
Cooch Behar Government Engineering College,
Cooch Behar,
India[email protected] Siddhartha Bhattacharyya
Department of Computer Application,
RCC Institute of Information Technology,
Canal South Road,
Beliaghata,
Kolkata-700015,
India[email protected] The action of combining the components from various algorithms is presently the utmost successful and effective trend in optimization. The foremost motivation behind the hybridization of diverse algorithmic ideas is to acquire better performing systems, which exploit and coalesce benefits of the different pure approaches, that is, hybrid systems that are supposed to be benefited from synergy. Actually, taking a suitable amalgamation of multiple algorithmic notions is often the key to achieve highest performance in solving number of hard (complex) optimization problems. Nonetheless, evolving an exceedingly effective hybrid method is not a simple task at all. The hybridization of popular metaheuristics like particle swarm optimization, evolutionary algorithms, simulated annealing, variable neighborhood search, and ant colony optimization with techniques from other fields like artificial intelligence, operations research forms the basis of evolving more efficient and robust solutions.
Keywords: Hybrid metaheuristics, quantum computing, metaheuristics, optimization, and advanced hybrid metaheuristics.
1. Introduction
Optimization is required in different fields of engineering such as mechanical engineering, civil engineering, mining engineering, nanoscience and nanoengineering, computer, communication, networking and information engineering, bioinformatics and biomedical engineering, etc. to obtain better solutions. To solve optimization problems practically in those above mentioned fields, some efficient and effective computational algorithms are very mush essential. The foremost objective of the optimization is to derive the optimal solution for a given problem. An optimization problem is determined as: deriving values of the variables that maximize or minimize the fitness (objective) function(s) in line with the constraints.1 These kinds of problems are based on three factors: firstly, they solve some minimization/- maximization objective function(s), secondly, a set of unknown variables those are involve in the objective functions and thirdly, a set of constraints that permit the unknowns for taking some specific values but exclude others.1
Most of the optimization problems may have more than one local solutions. In this circumstances, choosing of the optimization method is very much important. The optimization method must not be greedy and the searching process will not be localized in the neighborhood of the best solution as it may stick at a local solution and that will misguide the search process. It should be observed that the optimization algorithm should make a balance between global and local search. Both the mathematical and combinatorial types optimization problems can be solved by different methods. The optimization problems that have large search space or more complex in nature will become difficult to solve using conventional mathematical optimization algorithms. Here is the utility of the metaheuristic algorithms. Different metaheuristic optimization algorithms, present in the research arena are very much capable to solve difficult optimization problems.
A heuristic method can be noted as the way of solving, leaning or discovery of a problem using practical methods and ultimately, will derive immediate near optimal results rather than exact results. Basically, a metaheuristic is an iterative generation procedure to solve a subordinate heuristic by syndicating intelligently different concepts to explore and exploit the search space, learning strategies are applied to structure information in order to find efficiently near-optimal solutions.2 The main objective of metaheuristic is to derive a set of optimal solutions which is large enough to be completely sampled. The popularity of using metaheuristic techniques on a variety of problems is due to the fact that any conventional algorithm cannot manage many real world problems, in spite of the raising computational power, simply due to unrealistically large running times.3 These algorithms make few assumptions to solve the optimization problems. It is not guaranteed that the metaheuristics may generate globally optimal solutions to solve some class of problems since most of the implementations are some form of stochastic optimization and the resultant solutions may dependents on the set of generated random variables. Metaheuristic algorithms are advantageous over optimization algorithms, simple heuristics, or iterative methods as they often determine good solutions with a lesser computational effort by exploring a large set of feasible solutions.
Most well-known metaheuristic algorithms are Genetic algorithm (GA),4 simulated annealing (SA),5 Tabu search (TS)6–8 and different types of swarm intelligence algorithms. Genetic algorithm (GA) works on the principle of the evolutionary process in nature. Tabu search applies the memory structure in living beings, whereas simulated annealing imitates the annealing process in crystalline solids. Some well known and recognized swarm intelligence algorithms are particle swarm optimization (PSO),9 ant colony optimization (ACO),10 artificial bee colony optimization (ABC),11 differential opti...
