Unite and conquer approach for data clustering based on particle swarm optimization and moth flame optimization

Iranian Journal of Numerical Analysis and Optimization

مقالات آماده انتشار، پذیرفته شده، انتشار آنلاین از تاریخ 26 دی 1403

نوع مقاله: Research Article

شناسه دیجیتال (DOI): 10.22067/ijnao.2025.89754.1508

نویسندگان

Esra Mosavi¹؛ Seyed Abolfazl Shahzadeh Fazeli^{* 1}؛ Elham Abbasi²؛ Fatemeh Kaveh-Yazdy²

¹Parallel Processing Laboratory, Department of Computer Science, Yazd University, Yazd, Iran.

²Department of Computer Science, Yazd University, Yazd, Iran.

چکیده

Data clustering is a widely used technique in various domains to group data objects according to their similarity. Clustering molecules is a useful process where you can easily subdivide and manipulate large datasets to group compounds into smaller clusters with similar properties. To discover new molecules with optimal properties and desired biological activity can be used by comparing molecules and their similarities. A prominent clustering technique is the k-means algorithm, which assigns data objects to the nearest cluster center. However, this algorithm relies on the initial selection of the cluster centers, which can affect its convergence and quality. To address this issue, metaheuristic algorithms have been proposed as a type of approximate optimization algorithm capable of identifying almost optimal solutions. In this paper, a new meta-heuristic approach is proposed by combining two algorithms of particle swarm optimization (PSO) and moth flame optimization (MFO), following that, it is used to improve data clustering. The efficiency of the proposed approach is evaluated utilizing benchmark functions F1-F23. Its efficiency is evaluated with PSO and MFO algorithms on different datasets. Our experiential results show that the suggested approach exceeds the PSO and MFO algorithms with respect to speed of convergence and clustering quality.

کلیدواژه‌ها

Data clustering؛ k-means clustering؛ Metaheuristic optimization؛ Particle swarm optimization؛ Moth flame optimization