W-GBC: An Adaptive Weighted Clustering Method Based on Granular-Ball Structure

Existing weighted clustering algorithms often heavily rely on specific parameters. Specifically, in addition to the number of clusters (k), several other parameters need to be manually tuned, which greatly limits their practical applicability. The fundamental issue lies in the fact that most weighted clustering methods derive feature weights through global iterations. To address this challenge, this paper introduces a novel weighted granular-ball structure, continually optimizing weights during the ball splitting process and restricting the calculation of local data point weights to the corresponding weighted granular-ball. We employ local iterations within this structure as an approximation to global weight calculations. This method eliminates the need for parameter tuning during the weight calculation process and incidentally addresses the “curse of dimensionality” in traditional granular-ball computing model. When applied to complex real-world datasets, this method accurately represents high-dimensional data, thereby improving clustering precision and extending the adaptability of the granular-ball computing model in high-dimensional spaces. Comprehensive experimental analysis demonstrates that our W-GBC algorithm performs well in terms of clustering results and competes strongly with baseline algorithms. The code has been released and is now available at https://github.com/xjnine/W-GBC.

I. Introduction

Clustering is a fundamental data analysis technique widely employed in various fields, including marketing research, data mining, bioinformatics, image processing, and pattern recognition [1]–[5]. While the K-means [6] algorithm is widely popular due to its intuitive and efficient characteristics, it does have limitations, such as sensitivity to noise and irrelevant features. The fundamental concept behind the K-means algorithm is to iteratively refine clusters by updating cluster centers. Although the K- means algorithm finds extensive application in clustering analysis, it has an inherent constraint: it assigns equal weight to each variable during the clustering process. However, in real-world scenarios, true clustering often occurs only within specific subsets of dimensions. Consequently, contemporary research leans toward adapting the algorithm to incorporate feature weights. This adaptation entails considering not only the significance of individual features but also the interplay among them. Such an approach enables a more comprehensive understanding and exploitation of the underlying data structure, ultimately enhancing the clustering algorithm's performance.