AI Paper Atlas

Research map / Learning theory and algorithms / Fairness

Training individually fair ML models with sensitive subspace robustness

Mikhail Yurochkin, Amanda Bower, Yuekai Sun

ICLR 2020 · Learning theory and algorithms › Fairness · approach: Fairness-constrained classification · 51 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
train with sensitive-subspace distributionally robust optimization instead of empirical risk minimizationone-layer neural network · Sentiment predictionRace gap: 7.01$\pm$.44 -> 0.30$\pm$.05Improved
train with sensitive-subspace distributionally robust optimization instead of empirical risk minimizationone-layer neural network · Sentiment predictionAcc.,%: 95$\pm$1 -> 94$\pm$1Hurt
use expert-provided evaluation names to define the sensitive subspaceSenSR · Sentiment predictionRace gap: 0.30$\pm$.05 -> 0.11$\pm$.04Improved
use expert-provided evaluation names to define the sensitive subspaceSenSR · Sentiment predictionCuis. gap: 0.23$\pm$.05 -> 1.11$\pm$.15Hurt
project inputs onto the orthogonal complement of the sensitive subspace before trainingone-layer neural network · Adult testS-Con.: .848 -> .868Improved
train with SenSR rather than the baseline100-unit single-hidden-layer neural network · Adult testS-Con.: .848 -> .934Improved
train with SenSR rather than the baseline100-unit single-hidden-layer neural network · Adult testB-Acc,%: 82.9 -> 78.9Hurt
train with SenSR rather than adversarial debiasing100-unit single-hidden-layer neural network · Adult testS-Con.: .807 -> .934Improved
train with adversarial debiasing rather than the baseline100-unit single-hidden-layer neural network · Adult testS-Con.: .848 -> .807Hurt
use expert-provided sensitive directions rather than directions learned from a larger names datasetSenSR · Sentiment predictionCuis. gap: 0.23$\pm$.05 -> 1.11$\pm$.15Hurt

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