AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

PASS: Protected Attribute Suppression System for Mitigating Bias in Face Recognition

Prithviraj Dhar, Joshua Gleason, Aniket Roy, Carlos D. Castillo, Rama Chellappa

ICCV 2021 · Label-efficient and robust learning › Recognition and long-tail · approach: Fair and quality-aware face recognition · 38 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
add OAT discriminator training instead of AET (ablation)PASS-g on Arcface · IJB-C gendergender classification accuracy: 81.84 -> 73.65Improved
add OAT discriminator training instead of AET (ablation)PASS-g on Crystalface · IJB-C gendergender classification accuracy: 86.42 -> 80.54Improved
add OAT instead of AET (ablation)PASS-g on Crystalface · IJB-C gender bias FPR 1e-5bias: 0.028 -> 0.002Improved
apply PASS-g to Arcface descriptorsArcface · IJB-C gendergender classification accuracy: 82.06 -> 73.65Improved
add race (second attribute) via MultiPASS on top of PASS-gPASS-g on Arcface · IJB-C gendergender classification accuracy: 73.65 -> 68.43Improved
apply PASS-s to Arcface descriptorsArcface · IJB-C skintone bias FPR 1e-3bias: 0.006 -> 0.000Improved
apply PASS-g to Arcface descriptorsArcface · IJB-C verification TPR% at FPR 1e-4TPR: 95.3 -> 94.6Hurt
apply PASS-s to Arcface descriptorsArcface · IJB-C verification TPR% at FPR 1e-4TPR: 95.3 -> 93.4Hurt
apply IVE(s) baseline to ArcfaceArcface · IJB-C skintoneskintone classification accuracy: 87.15 -> 88.23No clear effect
remove hair region (W/o hair) from Arcface inputArcface · IJB-C verification FPR 1e-5TPR: 0.929 -> 0.616Hurt
apply PASS-s to Crystalface descriptorsCrystalface · IJB-C skintoneskintone classification accuracy: 89.30 -> 83.84Improved
apply PASS-g to Crystalface descriptorsCrystalface · IJB-C gendergender classification accuracy: 86.73 -> 80.54Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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