Overview of Parallel Computing for Meta-Heuristic Algorithms

The meta-heuristic algorithm is used in the research of various complex problems. Due to the limitations of the original meta-heuristic algorithm, many improved meta-heuristic algorithms have been proposed, such as compact, adaptive, multi-objective and parallel schemes. Among them, the parallel strategy may get a significant improvement for related applications. This paper mainly studies the application of parallel computing in meta-heuristic algorithms. There are two main types of parallelism: one is absolute parallelism, using multiple processors which can solve optimization problems with high computational costs and improve execution efficiency. The other is virtual parallelism (multi-grouping), which decomposes the population into multiple sub-populations, and each sub-population communicates between species to generate better solutions. In addition, the combination of parallel computing and meta-heuristic algorithms can solve a wide variety of application problems: path planning, engineering design, large-scale optimization, image segmentation, neural networks and prediction problems, etc. This paper presents a comprehensive study and systematic survey of parallel meta-heuristics. © 2022, Journal of Network Intelligence. All rights reserved.