AI Paper Atlas

Research map / Learning theory and algorithms / Fairness

Fair Multiple Decision Making Through Soft Interventions

Yaowei Hu, Yongkai Wu, Lu Zhang, Xintao Wu

NeurIPS 2020 · Learning theory and algorithms › Fairness · approach: Causal and policy fairness in decisions

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
train classifiers serially/jointly with soft-intervention post-deployment distributions instead of separatelyh2 · Synthetic testunfairness: 0.09 -> 0.03Improved
train serially instead of separatelyh2 · Adult testunfairness: 0.17 -> 0.10Improved
train serially instead of separatelyh2 · Synthetic testaccuracy (%): 77.06 -> 84.16Improved
train jointly instead of separatelyh2 · Synthetic testaccuracy (%): 77.06 -> 82.09Improved
train serially instead of separatelyh2 · Adult testaccuracy (%): 73.33 -> 68.91Hurt
train jointly instead of separatelyh2 · Adult testaccuracy (%): 73.33 -> 69.40Hurt
add fairness constraints versus unconstrained training (ablation)h2 · Synthetic testaccuracy (%): 89.95 -> 77.06Hurt
add fairness constraints versus unconstrained (ablation)h1 · Synthetic testunfairness: 0.15 -> 0.01Improved
separate training on grid-searched pairs fair in trainingh1,h2 pairs · Syntheticshare violating fairness: 71.43%Hurt
use serial or joint training on grid-searched pairsh1,h2 pairs · Syntheticfairness violationsImproved
joint instead of serial trainingfair classifiers · Synthetich2 accuracy (%): 84.16 vs 82.09No clear effect
add fairness constraint on h1 versus unconstrained (ablation)h1 · Adult testaccuracy (%): 55.63 -> 55.56No clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy