We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
An improved binary snake optimizer with Gaussian mutation transfer function and hamming distance for feature selection.
- Authors
Bao, Xinyu; Kang, Hui; Li, Hongjuan
- Abstract
The snake optimizer (SO) is a highly efficient bio-inspired algorithm for solving continuous optimization problems. This algorithm mathematically simulates the unique foraging and mating behaviors observed in snake populations in nature. However, this algorithm cannot be directly applied to solve binary optimization problems, such as feature selection. Feature selection is a significant data preprocessing step in data mining, aimed at reducing data dimensionality, lowering computational costs in terms of time and space, and improving the predictive accuracy of classifiers. To address this limitation, an improved binary version of SO (IBSO) is proposed for feature selection, which incorporates the concept of hamming distance and introduces a novel mutation transfer function (MTF). IBSO extends the application of the conventional SO to binary optimization problems by introducing hamming distance and presents a new binary position update strategy. Furthermore, IBSO utilizes a MTF based on a Gaussian distribution. The MTF not only transforms each dimension of each individual in the population into binary space but also enhances the local random search capability and increases the population diversity of the algorithm. Finally, the experiment is conducted on 27 standard benchmark datasets from UC Irvine Machine Learning Repository and IBSO is compared with several state-of-the-art binary swarm intelligence algorithms to analyze the effectiveness and efficiency of IBSO. The results show that IBSO can obtain the best fitness values with less CPU time. Besides, to evaluate the validity of the proposed MTF, IBSO is compared with other binary versions of SO with different well-known transfer functions.
- Subjects
UNIVERSITY of California, Irvine; HAMMING distance; FEATURE selection; TRANSFER functions; SWARM intelligence; ANIMAL sexual behavior; BIOLOGICALLY inspired computing
- Publication
Neural Computing & Applications, 2024, Vol 36, Issue 16, p9567
- ISSN
0941-0643
- Publication type
Article
- DOI
10.1007/s00521-024-09581-6