Multi-view biclustering via non-negative matrix tri-factorisation

TL;DR

This paper introduces ResNMTF, a novel non-negative matrix tri-factorisation method for multi-view biclustering that effectively discovers overlapping clusters without prior knowledge of their number, and extends silhouette scores for better evaluation.

Contribution

The paper presents ResNMTF, a new multi-view biclustering algorithm capable of identifying overlapping clusters without pre-specified numbers, and introduces the bisilhouette score for improved cluster evaluation.

Findings

ResNMTF accurately identifies overlapping biclusters.

The bisilhouette score correlates well with external validation measures.

ResNMTF works effectively on both synthetic and real datasets.

Abstract

Multi-view data is ever more apparent as methods for production, collection and storage of data become more feasible both practically and fiscally. However, not all features are relevant to describe the patterns for all individuals. Multi-view biclustering aims to simultaneously cluster both rows and columns, discovering clusters of rows as well as their view-specific identifying features. A novel multi-view biclustering approach based on non-negative matrix factorisation is proposed named ResNMTF. Demonstrated through extensive experiments on both synthetic and real datasets, ResNMTF successfully identifies both overlapping and non-exhaustive biclusters, without pre-existing knowledge of the number of biclusters present, and is able to incorporate any combination of shared dimensions across views. Further, to address the lack of a suitable bicluster-specific intrinsic measure, the popular silhouette score is extended to the bisilhouette score. The bisilhouette score is demonstrated to align well with known extrinsic measures, and proves useful as a tool for hyperparameter tuning as well as visualisation.