Efficient and Effective Optimal Transport-Based Biclustering

Abstract

Bipartite graphs can be used to model a wide variety of dyadic information such as user-rating, document-term, and gene-disorder pairs. Biclustering is an extension of clustering to the underlying bipartite graph induced from this kind of data. In this paper, we leverage optimal transport (OT) which has gained momentum in the machine learning community to propose a novel and scalable biclustering model that generalizes several classical biclustering approaches. We perform extensive experimentation to show the validity of our approach compared to other OT biclustering algorithms along both dimensions of the dyadic datasets.

PDF NeurIPS OpenReview Semantic Scholar

Cite

Text

Fettal et al. "Efficient and Effective Optimal Transport-Based Biclustering." Neural Information Processing Systems, 2022.

Markdown

[Fettal et al. "Efficient and Effective Optimal Transport-Based Biclustering." Neural Information Processing Systems, 2022.](https://mlanthology.org/neurips/2022/fettal2022neurips-efficient/)

BibTeX

@inproceedings{fettal2022neurips-efficient,
  title     = {{Efficient and Effective Optimal Transport-Based Biclustering}},
  author    = {Fettal, Chakib and Labiod, Lazhar and Nadif, Mohamed},
  booktitle = {Neural Information Processing Systems},
  year      = {2022},
  url       = {https://mlanthology.org/neurips/2022/fettal2022neurips-efficient/}
}