Multi-view Data Clustering Through Consensus Graph and Data Representation Learning

Fadi Dornaika () and Sally El Hajjar
Additional contact information
Fadi Dornaika: University of the Basque Country
Sally El Hajjar: University of the Basque Country

Chapter Chapter 6 in Advances in Data Clustering, 2024, pp 95-116 from Springer

Abstract: Abstract In the domain of multi-view clustering, existing methodologies can be categorized into subspace multi-view clustering algorithms, multi-view kernel approaches, matrix factorization approaches, and spectral clustering algorithms. However, a common limitation among these approaches is their dependence on combining predefined individual similarity matrices from multiple views. This susceptibility to noisy original similarity matrices, along with the integration of various spectral projection matrices, often affects their overall performance. To address these limitations, we introduce a novel approach named multi-view clustering with consensus graph learning and spectral representation (MCGLSR). In contrast to the traditional practice of directly integrating similarity matrices from different views, which may introduce noise, our proposed method simultaneously generates similarity graphs for each view and their shared similarity matrix (graph matrix) through a unified global objective function. This unified objective function ensures that the similarity matrices from different views are compelled to be sufficiently similar, effectively mitigating the impact of noise and promoting a more coherent unified data structure. Moreover, our approach facilitates the recovery of the common spectral projection and soft cluster assignments based on the shared graph structure. Crucially, MCGLSR operates on a kernelized representation of the views’ features, producing individual graphs, a common graph, a common spectral representation, and cluster assignments directly. This eliminates the need for an external clustering algorithm in the final stage. To validate the efficacy of our technique, we conduct experiments on several real-world datasets, demonstrating its robust performance and addressing the identified shortcomings in existing approaches.

Keywords: Multi-view clustering; Consensus graph learning; Consensus spectral representation matrix; Cluster assignment matrix (search for similar items in EconPapers)
Date: 2024
References: Add references at CitEc
Citations:

There are no downloads for this item, see the EconPapers FAQ for hints about obtaining it.

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:spr:sprchp:978-981-97-7679-5_6

Ordering information: This item can be ordered from
http://www.springer.com/9789819776795

DOI: 10.1007/978-981-97-7679-5_6

Access Statistics for this chapter

More chapters in Springer Books from Springer
Bibliographic data for series maintained by Sonal Shukla () and Springer Nature Abstracting and Indexing ().