A split and merge strategy for multi-objective clustering algorithms

Azevedo, Beatriz Flamia; Rocha, Ana Maria A.C.; Pereira, Ana I.

http://hdl.handle.net/10198/36881

Utilize este identificador para referenciar este registo.

Nome:	Descrição:	Tamanho:	Formato:
A Split and Merge.pdf		24.96 MB	Adobe PDF	Ver/Abrir

Contacte-nos

Autores

Azevedo, Beatriz Flamia

Rocha, Ana Maria A.C.

Pereira, Ana I.

Resumo(s)

Complex real-world problems require advanced models for large datasets; combining optimization and machine learning methods can enhance solution effectiveness and efficiency. This work presents an automatic bio-inspired clustering algorithm named Multi-objective Clustering Algorithm II. Through an optimization process, the algorithm autonomously determines the number of clusters, their centroids, and the optimal distribution of their elements. Furthermore, the paper also presents a split and merge strategy for clustering algorithms, with a special focus on multi-objective ones. The proposed algorithms were executed on 10 benchmark datasets, yielding satisfactory results by accurately estimating the optimal number of clusters and providing appropriate dataset partitions. These results outstand the k-means and DBSCAN algorithms results, which were used as a comparison.

Palavras-chave

Collaborative system Data analysis Multi-objective clustering Partitioning

URI

http://hdl.handle.net/10198/36881

Citação

Azevedo, Beatriz Flamia; Rocha, Ana Maria A.C.; Pereira, Ana I. (2025). A Split and Merge Strategy for Multi-objective Clustering Algorithms. SN Computer Science. ISSN 2662995X. 6:6, p. 1-21