Aarhus University Seal

Privacy-preserved Efficient Contrastive Multi-View Clustering (EC-MVC)

Multi-view Clustering (MVC) aims to find patterns and clusters across these multi-view data to achieve a better understanding of the entity, such as classifying cancers using multi-omics data from different hospitals.

   

Most solutions focus on centralized setups or ignore the privacy concerns of distributed setups. This project aims to enhance the privacy protection in distributed MVC without performance deterioration via incorporating contrastive learning to improve the intra-view consensus while introducing new distance metrics to maximize inter-cluster separability.


Partners


Principal Investigator from ECE, AU:

Peng Yuan Zhou

Tenure Track Assistant Professor

About the project:

Grant source:
AUFF NOVA, Aarhus Universitets Forskningsfond

Granted amount:
DKK 2.126.560

Project start:
01/07/2025