.png){kind=icon}
Return
Potential bias when creating a differential-vector movement algorithm
.png){kind=icon}
PRE
AI
DOI：10.1016/j.asoc.2021.107925.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Simulating natural biological processes has been used widely to create new algorithms to optimize complex problems. Hundreds of new algorithms have been proposed in this growing area of research, with many of these using differential-vector movement to generate new solutions. However, the relationship between the natural inspiration for an algorithm and its algorithmic behavior is sometimes tenuous. Unfortunately, high levels of solution quality and speed on narrowly defined benchmarks are often prioritized over general theoretical understanding. Several algorithms, including teaching-learning-based optimization, symbiotic organisms search, sine cosine algorithm, forensic-based investigation optimization, and grey wolf optimizer were examined to explore the search regions achieved by their suggested strategies. Interestingly, all of the five algorithms were found to frequently consider the origin of coordinates when generating new solutions. Accordingly, these algorithms sometimes obtained extraordinary results when applied to benchmark functions where the optimum solutions reside at the origin. The results highlight the importance of properly checking how and from which regions points are sampled when developing new metaheuristics. Furthermore, providing clear descriptions of the underlying designs of strategies and parameter settings is encouraged. (C) 2021 Elsevier B.V. All rights reserved.
Keywords:
Nature-inspired algorithms
Bio-inspired algorithms
Differential vectors
Journal
IF:
6.6
Papers:
1.4W
Citations:
4.8W
Organization
Cited Papers
.png){kind=icon}
ShareSaveShareSaveShareSaveShareSaveShareSaveShareSave
View more