Scholarlink

<p>One of the key challenges in the field of data clustering is the identification of cluster centroid positions to form compact and well-separated clusters. To address this challenge, this work proposes the hybrid opposition-based improved hiking optimization algorithm (HOB-iHOA) for partitional data clustering. The proposed algorithm employs a hybrid approach for initializing cluster centroids by integrating random estimation, quasi-opposition-based learning, and K-Means to improve exploration capability, ensuring diversity and identifying various promising regions in the feature space. Moreover, this algorithm balances exploration and exploitation using momentum-driven updates combined with logistic map chaos, promoting stable convergence of cluster centroids for improved clustering solutions. The performance of HOB-iHOA is evaluated on 23 benchmark functions, comprising unimodal, multimodal, and fixed-dimension multimodal functions, against seven state-of-the-art (SoA) algorithms. Furthermore, its applicability is tested on three real-world engineering design problems, demonstrating its reliability. Finally, the effectiveness of HOB-iHOA in data clustering problem is evaluated on thirteen real-world datasets from the UCI Machine Learning Repository and the results obtained with HOB-iHOA are compared with existing metaheuristic-based SoA clustering algorithms using both internal and external validation metrics. The HOB-iHOA improves the performance of data clustering by reducing the sum of intra-cluster distances and the Davies–Bouldin index and by increasing the silhouette coefficient and adjusted rand index scores on benchmark datasets. 3D scatter plots, convergence curve analysis, ablation studies, and runtime comparisons are conducted to show HOB-iHOA suitability for practical deployment. Additionally, scalability, statistical robustness and parameter sensitivity analyses of HOB-iHOA have been performed. The results demonstrate that HOB-iHOA effectively discovers optimal cluster centroid positions, forming compact and well-separated clusters that provide deeper insights into the datasets.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Dynamic centroids discovery using a hybrid approach for improved data clustering

  • 期刊论文

  • 发表日期：

    2026年5月15日

查看全文
  • Jaswinder Singh,
  • Damanpreet Singh

摘要

One of the key challenges in the field of data clustering is the identification of cluster centroid positions to form compact and well-separated clusters. To address this challenge, this work proposes the hybrid opposition-based improved hiking optimization algorithm (HOB-iHOA) for partitional data clustering. The proposed algorithm employs a hybrid approach for initializing cluster centroids by integrating random estimation, quasi-opposition-based learning, and K-Means to improve exploration capability, ensuring diversity and identifying various promising regions in the feature space. Moreover, this algorithm balances exploration and exploitation using momentum-driven updates combined with logistic map chaos, promoting stable convergence of cluster centroids for improved clustering solutions. The performance of HOB-iHOA is evaluated on 23 benchmark functions, comprising unimodal, multimodal, and fixed-dimension multimodal functions, against seven state-of-the-art (SoA) algorithms. Furthermore, its applicability is tested on three real-world engineering design problems, demonstrating its reliability. Finally, the effectiveness of HOB-iHOA in data clustering problem is evaluated on thirteen real-world datasets from the UCI Machine Learning Repository and the results obtained with HOB-iHOA are compared with existing metaheuristic-based SoA clustering algorithms using both internal and external validation metrics. The HOB-iHOA improves the performance of data clustering by reducing the sum of intra-cluster distances and the Davies–Bouldin index and by increasing the silhouette coefficient and adjusted rand index scores on benchmark datasets. 3D scatter plots, convergence curve analysis, ablation studies, and runtime comparisons are conducted to show HOB-iHOA suitability for practical deployment. Additionally, scalability, statistical robustness and parameter sensitivity analyses of HOB-iHOA have been performed. The results demonstrate that HOB-iHOA effectively discovers optimal cluster centroid positions, forming compact and well-separated clusters that provide deeper insights into the datasets.