Proposes ReLiF framework for reliable Lipschitz fairness in MTL via fixed-δ auditing and violation-rate feedback control, claiming it reveals obscured utility-fairness trade-offs on clinical and NYUv2 benchmarks.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Is Fairness Truly Fair? Towards Reliable Lipschitz Fairness in Multi-Task Learning via Fixed-\texorpdfstring{$\delta$}{delta} Alignment
Proposes ReLiF framework for reliable Lipschitz fairness in MTL via fixed-δ auditing and violation-rate feedback control, claiming it reveals obscured utility-fairness trade-offs on clinical and NYUv2 benchmarks.