We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.LG

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Machine Learning

Title: Deep Fair Clustering via Maximizing and Minimizing Mutual Information

Abstract: Fair clustering aims to divide data into distinct clusters, while preventing sensitive attributes (e.g., gender, race, RNA sequencing technique) from dominating the clustering. Although a number of works have been conducted and achieved huge success in recent, most of them are heuristical, and there lacks a unified theory for algorithm design. In this work, we fill this blank by developing a mutual information theory for deep fair clustering and accordingly designing a novel algorithm, dubbed FCMI. In brief, through maximizing and minimizing mutual information, FCMI is designed to achieve four characteristics highly expected by deep fair clustering, i.e., compact, balanced, and fair clusters, as well as informative features. Besides the contributions to theory and algorithm, another contribution of this work is proposing a novel fair clustering metric built upon information theory as well. Unlike existing evaluation metrics, our metric measures the clustering quality and fairness in a whole instead of separate manner. To verify the effectiveness of the proposed FCMI, we carry out experiments on six benchmarks including a single-cell RNA-seq atlas compared with 11 state-of-the-art methods in terms of five metrics. Code will be released after the acceptance.
Subjects: Machine Learning (cs.LG); Computers and Society (cs.CY)
Cite as: arXiv:2209.12396 [cs.LG]
  (or arXiv:2209.12396v1 [cs.LG] for this version)

Submission history

From: Pengxin Zeng [view email]
[v1] Mon, 26 Sep 2022 03:38:48 GMT (11970kb,D)

Link back to: arXiv, form interface, contact.