AI Paper Atlas

Research map / Learning theory and algorithms / Fairness

Approximate Heavily-Constrained Learning with Lagrange Multiplier Models

Harikrishna Narasimhan, Andrew Cotter, Yichen Zhou, Serena Wang, Wenshuo Guo

NeurIPS 2020 · Learning theory and algorithms › Fairness · approach: Fair allocation and selection · 5 citations (OpenAlex)

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
increase multiplier model capacity from common multiplier to 3 hidden layers (ablation)Lagrange multiplier model · Communities & Crime, test95th percentile violation: 0.11 ± 0.015 -> 0.05 ± 0.017Improved
increase multiplier capacity from common multiplier to 3 hidden layers (ablation)Lagrange multiplier model · Communities & Crime, testerror rate: 0.17 ± 0.006 -> 0.21 ± 0.009Hurt
add constraints (3-hidden layer multiplier vs unconstrained)linear classifier · Communities & Crime, test95th percentile violation: 0.09 ± 0.012 -> 0.05 ± 0.017Improved
add constraints (3-hidden layer multiplier vs unconstrained)linear classifier · Communities & Crime, testerror rate: 0.15 ± 0.003 -> 0.21 ± 0.009Hurt
use multiplier model instead of brute-force oracle FairFictitiousPlay3-hidden-layer multiplier · Communities & Crime, test95th percentile violation: 0.05 ± 0.017 vs 0.04 ± 0.008No clear effect
use linear multiplier instead of common multiplier (ablation)Lagrange multiplier model · Communities & Crime, test95th percentile violation: 0.11 ± 0.015 -> 0.14 ± 0.004Hurt
proposed method vs SoftAssign at high noise 0.5fairness with noisy groups · Adult, noise 0.5test error rate: 0.218 ± 0.002 -> 0.189 ± 0.006Improved
proposed method vs SoftAssign at low noise 0.1fairness with noisy groups · Adult, noise 0.1equal opportunity violation: -0.048 ± 0.002 -> 0.015 ± 0.035Hurt
proposed method vs DRO at noise 0.5fairness with noisy groups · Adult, noise 0.5test error rate: 0.219 ± 0.012 -> 0.189 ± 0.006Improved
proposed method vs DRO at noise 0.3fairness with noisy groups · Adult, noise 0.3test error rate: 0.216 ± 0.010 -> 0.165 ± 0.003Improved
per-query constraints with multiplier model vs average constraint of Narasimhan et al.ranking model · MSLR-WEB10K, test90th percentile violation: 0.246 ± 0.025 -> 0.195 ± 0.023Improved
per-query constraints with multiplier model vs Narasimhan et al.ranking model · MSLR-WEB10K, testpairwise error: 0.400 ± 0.014 -> 0.412 ± 0.011Hurt

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