Hidden Convexity of Fair PCA and Fast Solver via Eigenvalue Optimization

Junhui Shen, Aaron J. Davis, Ding Lu, Zhaojun Bai · 2025

Read Hidden Convexity of Fair PCA and Fast Solver via Eigenvalue Optimization by Junhui Shen, Aaron J. Davis, Ding Lu, Zhaojun Bai (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-00299.txt.

Description

Junhui Shen, Aaron Davis, Ding Lu, and Zhaojun Bai find a hidden convexity inside fair PCA and turn it into a fast solver, in this 2025 arXiv paper. Ordinary PCA can rebuild some groups' data worse than others, and the fair version evens that loss across groups. The standard fair solver leans on semidefinite relaxation, which is slow even for rough answers.

Later variants dodged the cost by changing what fairness means. This team kept the original fairness target and reframed the problem as eigenvalue optimization, where the hidden convexity makes it tractable. On real datasets their solver ran about eight times faster than the relaxation method and stayed close to plain PCA speed.

Catalog record

CREATORJunhui Shen, Aaron J. Davis, Ding Lu, Zhaojun Bai
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE47 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2503-00299

Browse the full library · About the Tower of Babel