Apache Spark Implementation of Whale Optimization Algorithm
- PDF / 2,129,727 Bytes
- 14 Pages / 595.276 x 790.866 pts Page_size
- 59 Downloads / 313 Views
(0123456789().,-volV)(0123456789(). ,- volV)
Apache Spark Implementation of Whale Optimization Algorithm Maryam AlJame1 • Imtiaz Ahmad1 • Mohammad Alfailakawi1 Received: 1 September 2019 / Revised: 21 July 2020 / Accepted: 22 July 2020 Ó Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract Population-based meta-heuristic algorithms are among the dominant algorithms used to solve challenging real world problems in diverse fields. Whale Optimization Algorithm (WOA) is a recent swarm intelligence meta-heuristic algorithm based on the bubble-net feeding behavior of humpback whales. Despite its capability to solve complex optimization problems, WOA requires enormous amount of computations when solving large size problems. This work proposes SparkWOA, a distributed implementation of WOA on Apache Spark platform to enhance its performance and reduce computational complexity. The proposed algorithm exploits in-memory computations and broadcast features of Apache Spark to provide better performance and scalability. Details of the proposed algorithm are presented and its performance as compared to a recent Apache Hadoop implementation is discussed. Experimental results demonstrated the superiority of the proposed implementation in terms of both speed and scalability. Keywords Apache Spark Cluster Hadoop Whale Optimization Algorithm
1 Introduction Evolutionary algorithms (EAs) are population-based metaheuristic algorithms that solve optimization problems by mimicking biological processes such as mutation, reproduction, and selection. Evolutionary algorithms robustness make them widely applicable in several applications including image processing, communication optimization, and vehicle routing. In fact, EAs underlie many applications that play important roles in our daily life such as Google Maps, airline reservations, mobile communication, and more [19]. One type of evolutionary algorithms is the class of nature-inspired swarm-based meta-heuristics. During recent years, several such algorithms have emerged namely Particle Swarm Optimization (PSO) [15], Cuckoo
& Mohammad Alfailakawi [email protected] Maryam AlJame [email protected] Imtiaz Ahmad [email protected] 1
Computer Engineering Department, Kuwait University, Kuwait, Kuwait
Search (CS) [39], Ant Colony Optimization (ACO) [6], and Artificial Bee Colony (ABC) [14, 40]. In 2016, Mirjalili and Lewis introduced Whale Optimization Algorithm (WOA), a swarm-based meta-heuristic algorithm that mimics humpback whales feeding behavior named bubble-netting [26]. In bubble-net feeding, whales work as a collaborative team when hunting. Once the leader whale finds prey, it starts encircling them with a 9-shaped network of bubbles while simultaneously emitting a sound to call the group. During the encircling, the prey get scared and try to escape by swimming up to the surface where the humpback whales will hunt them. Whale Optimization Algorithm mathematically models bubble-net feeding to solve optimization problems. WOA has a per
Data Loading...
