Article: A three-way density peak clustering algorithm based on sinusoidal fuzzy entropy Journal: International Journal of Business Intelligence and Data Mining (IJBIDM) 2024 Vol.25 No.2 pp.184 - 209 Abstract: The density peaks clustering (DPC) algorithm is an efficient and concise method for clustering that automatically detects density centres and noise points. However, it is vulnerable to inaccuracies in outlier identification due to the cutoff distance parameter. To address this drawback, this paper proposes a novel three-way density peak clustering algorithm called sinusoidal fuzzy entropy-based density peak clustering (SFE-DPC). An important aspect of SFE-DPC is determining the thresholds. To tackle this issue, a criterion using sinusoidal fuzzy entropy is developed, and genetic algorithms are utilised to search for the optimal thresholds. By employing the optimal thresholds determined through the proposed criterion, SFE-DPC divides each cluster into three sections, with points in the trivial region identified as outliers. To assess the performance of SFE-DPC, we evaluate it on UCI datasets that include outliers using three baseline measures (NMI, RI, and F1-score), and compare it to DPC, k-means, FCM, BIRCH, and SC methods. Experimental results confirm that SFE-DPC is more effective in outlier detection. Inderscience Publishers - linking academia, business and industry through research

Title: A three-way density peak clustering algorithm based on sinusoidal fuzzy entropy

Authors: Yudong Meng; Xin Xu; Leyuan Yan; Yuan Cao

Addresses: Department of Mathematics and Statistics, Shandong University of Technology, Zibo 255000, China ' Department of Mathematics and Statistics, Shandong University of Technology, Zibo 255000, China ' Department of Mathematics and Statistics, Shandong University of Technology, Zibo 255000, China ' Department of Mathematics and Statistics, Shandong University of Technology, Zibo 255000, China

Abstract: The density peaks clustering (DPC) algorithm is an efficient and concise method for clustering that automatically detects density centres and noise points. However, it is vulnerable to inaccuracies in outlier identification due to the cutoff distance parameter. To address this drawback, this paper proposes a novel three-way density peak clustering algorithm called sinusoidal fuzzy entropy-based density peak clustering (SFE-DPC). An important aspect of SFE-DPC is determining the thresholds. To tackle this issue, a criterion using sinusoidal fuzzy entropy is developed, and genetic algorithms are utilised to search for the optimal thresholds. By employing the optimal thresholds determined through the proposed criterion, SFE-DPC divides each cluster into three sections, with points in the trivial region identified as outliers. To assess the performance of SFE-DPC, we evaluate it on UCI datasets that include outliers using three baseline measures (NMI, RI, and F1-score), and compare it to DPC, k-means, FCM, BIRCH, and SC methods. Experimental results confirm that SFE-DPC is more effective in outlier detection.

Keywords: three-way clustering; 3WC; sinusoidal fuzzy entropy; SFE; outlier identification; genetic algorithm.

DOI: 10.1504/IJBIDM.2024.140239

International Journal of Business Intelligence and Data Mining, 2024 Vol.25 No.2, pp.184 - 209

Received: 28 May 2022
Accepted: 20 Jul 2023
Published online: 31 Jul 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article

Title: A three-way density peak clustering algorithm based on sinusoidal fuzzy entropy

Keep up-to-date