A novel bat algorithm with dynamic membrane structure for optimization problems
- PDF / 2,660,790 Bytes
- 26 Pages / 595.224 x 790.955 pts Page_size
- 98 Downloads / 279 Views
A novel bat algorithm with dynamic membrane structure for optimization problems Bisan Alsalibi1 · Laith Abualigah2
· Ahamad Tajudin Khader1
© Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract To improve the optimization efficiency for different optimization problems and take advantage of the dynamic membrane computing framework, this paper proposes an improved bat algorithm, namely, Dynamic Membrane-driven Bat Algorithm (DMBA). The dynamic construction of the DMBA algorithm aims at enhancing population diversity by balancing the exploration-exploitation tradeoff. Unlike the static membrane algorithms, the membranes in DMBA will be dynamically evolved by using merging and separation rules which help in maintaining the diversity of the population. The experimental results on a set of well-known benchmark functions including CEC 2005, CEC 2011, and CEC 2017 clearly prove the effectiveness of the proposed DMBA algorithm in terms of maintaining the diversity and exploitation capabilities compared to that of the others. It is shown that the proposed DMBA algorithm is superior to recent variants of the bat algorithm and other well-known algorithms in terms of solution accuracy and convergence speed. Keywords Dynamic membrane structure · Parallel membrane framework · Bat algorithm · Optimization problems
1 Introduction Recently, a myriad of meta-heuristic optimization algorithms has been introduced in the literature to solve a variety of optimization problems [1–4]. Among these, Bat Algorithm (BA) is motivated by the typical echolocation behavior of bats [5]. Despite its successful applications in several areas, the classical BA has two drawbacks [6, 7]. Firstly, it cannot keep track of the current best positions which causes a rapid loss of population diversity and premature convergence. Secondly, it suffers from a slow convergence speed [8–10].
Laith Abualigah
[email protected] Bisan Alsalibi [email protected] Ahamad Tajudin Khader [email protected] 1
School of Computer Sciences, Universiti Sains Malaysia, Penang, USM 11800, Malaysia
2
Faculty of Computer Sciences and Informatics, Amman Arab University, Amman 11953, Jordan
To overcome these drawbacks, BA algorithm has been enhanced by various researchers to improve its efficiency. To that aim, multiple strategies have been employed such as diversifying the initial bat population and enhancing the search characteristics. For instance, chaos-based bat algorithm (CBA) has been proposed by [11] where different chaotic maps have been incorporated into the standard BA to improve its global search behavior. Similarly, chaotic maps have been used for the same purposes by [8, 12]. In their work, a modified version of BA is proposed by employing the concepts of multi-population strategies [13]. A discrete BA is proposed to solve feature selection and classification problems [14], this version demonstrated the superiority of BA over other swarm-based approaches. Along similar lines, another discrete version of BA has been proposed by [15] to
Data Loading...
