AI Paper Atlas

Research map / Learning theory and algorithms / Fairness

FairGB: A Fair Granular-Ball Generation Method for Data Classification

Qifen Yang, Yuhui Deng, Jiande Huang, Peng Zhou, Xiwen Lu, Lin Cui

ICML 2026 · Learning theory and algorithms › Fairness · approach: Fairness-constrained classification

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 Neighborhood Mixup data augmentation to balance sensitive groups (ablation)FairGBFC · Six fairness datasetsDP and EO differenceImproved
add data augmentation (vs w/o Aug.) (ablation)FairGBFC · Law (Race)DP and EO differenceImproved
use FairGBG fair clustering instead of conventional GB generationgranular-ball generation · Law (Gender)average LDP: 0.012 -> 0.000Improved
use FairGBG instead of runner-up GB generationgranular-ball generation · Approval, Por, German, Adultaverage LDP: 97.0%, 95.0%, 97.4%, and 79.0% reductionImproved
use FairGBG on Default where baseline GBs contain a single sensitive groupgranular-ball generation · Default (Sex)average LDP: — -> 0.008Improved
FairGBFC instead of best GB-based baselineGB-KNN classifiers · German (Gender)DP difference: 94.2% and 92.2% reduction in ΔDP and ΔEOImproved
FairGBFC instead of best GB-based baselineGB-KNN classifiers · Adult (Sex)DP and EO difference: 52.2% and 44.7% reductionImproved
FairGBFC instead of runner-up fairness-aware methodMLP fairness baselines · Approval (Gender)DP and EO difference: 16.2% and 53.2% reductionImproved
FairGBFC instead of runner-up fairness-aware methodMLP fairness baselines · German (Gender)DP and EO difference: 50.0% and 55.9% reductionImproved
FairGBFC instead of runner-up on multi-valued attributefairness-aware baselines · Adult (Race)DP and EO difference: 34.6% and 56.2% reductionImproved
FairGBFC vs f-FERM on utilityfairness-aware baselines · Por (Sex)accuracy: 0.878 -> 0.858Hurt
FairGBFC vs GBKNN on runtimeGB-KNN classifiers · Adult (Sex)runtime (s): 20.40 -> 4.215Improved

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