Abstract
In many clustering problems, we have access to multiple sources of data representing different aspects of the problem. Each of these data separately represents an association between entities. Multi-view clustering involves integrating clustering information from these heterogeneous sources of data and has been shown to improve results over a single-view clustering. On the other hand, co-clustering has been widely used as a technique to improve clustering results on a single view by exploiting the duality between objects and their attributes. In this paper, we propose a multi-view clustering setting in the context of a co-clustering framework. Our underlying assumption is that similarity values generated from the individual data can be transferred from one view to the other(s) resulting in a better clustering of the data. We provide empirical evidence to show that this framework results in a better clustering accuracy than those obtained from any of the single views, tested on different datasets.
Original language | English |
---|---|
Pages (from-to) | 545-570 |
Number of pages | 26 |
Journal | Knowledge and Information Systems |
Volume | 47 |
Issue number | 3 |
DOIs | |
Publication status | Published - 1 Jun 2016 |
Bibliographical note
Publisher Copyright:© 2015, Springer-Verlag London.
Keywords
- Co-clustering
- Ensemble clustering
- Multi-view clustering
- Similarity measure
- Transfer learning
ASJC Scopus subject areas
- Software
- Information Systems
- Human-Computer Interaction
- Hardware and Architecture
- Artificial Intelligence