AI Paper Atlas

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

FairCal: Fairness Calibration for Face Verification

Tiago Salvador, Stephanie Cairns, Vikram Voleti, Noah Marshall, Adam M Oberman

ICLR 2022 · Label-efficient and robust learning › Recognition and long-tail · approach: Fair and quality-aware face recognition · 3 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
Use FairCal’s feature-cluster-conditional calibration instead of cosine-score-only calibrationFaceNet face verification · RFWAUROC: 88.26 -> 90.58Improved
Use FairCal’s feature-cluster-conditional calibration instead of cosine-score-only calibrationFaceNet face verification · RFWmean KS: 6.37 -> 1.37Improved
Use FairCal’s feature-cluster-conditional calibration instead of cosine-score-only calibrationFaceNet face verification · BFWAUROC: 96.06 -> 96.90Improved
Use FairCal’s feature-cluster-conditional calibration instead of cosine-score-only calibrationFaceNet face verification · BFWmean KS: 6.77 -> 3.09Hurt
Use FairCal rather than the baselineArcFace face verification · BFWAUROC: 97.41 -> 97.44Improved
Use FairCal rather than the baselineArcFace face verification · BFWmean KS: 2.57 -> 2.49Improved
Use FairCal rather than FSNFaceNet face verification · RFW, 1% global FPRAAD of subgroup FPR: 0.37 -> 0.28Improved
Use FairCal rather than FSNFaceNet face verification · RFW, 1% global FPRAAD of subgroup FPR: 0.35 -> 0.29Improved
Use FairCal rather than the strongest listed baselineFace verification · BFW, 1% global FPR, FaceNet (Webface)AAD of subgroup FPR: 0.77 -> 0.80Hurt
Use FairCal rather than the strongest listed baselineArcFace face verification · BFW, 1% global FPRAAD of subgroup FPR: 0.44 -> 0.63Hurt
Use FairCal rather than the baselineFace verification · IJB-C, FaceNet (VGGFace2)AUROC: 92.72 -> 94.74Improved
Use FairCal rather than the baselineFace verification · IJB-C, FaceNet (Webface)AUROC: 89.23 -> 92.04Improved

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