Reproduce “Kernel Dependence Regularizers and Gaussian Processes with Applications to Algorithmic Fairness” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Zhu Li, Adrian Perez-Suay, Gustau Camps-Valls, Dino Sejdinovic
The authors introduce a Gaussian-process method that penalizes dependence between predictions and prespecified sensitive variables, rather than removing those variables or their proxies. Its Hilbert-Schmidt Independence Criterion penalty targets fairness in expectation. In a synthetic biased-label example, penalty 200 lowers prediction–x3 correlation to 0.0139, from 0.3863 for standard GP, but reduces R2 on observed labels; real-task gains are not universal.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.