AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

Fairness-aware Vision Transformer via Debiased Self-Attention

Yao Qiang, Chengyin Li, Prashant Khanduri, Dongxiao Zhu

ECCV 2024 · Architectures and efficiency › Attention and transformers · approach: Efficient and robust vision transformers · 10 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
apply DSA instead of vanilla ViT trainingViT-S/16 · CelebA (Hair Color, Gender)DBA: 0.0422 -> 0.0031Improved
apply DSA instead of vanilla ViT trainingViT-S/16 · WaterbirdsAccuracy (%): 62.36 -> 69.58Improved
use CNN-designed debiasing (MMD, MFD, DANN) on ViTViT-S/16 · WaterbirdsEO: 0.0209 -> 0.0617 (MMD)Hurt
add CE loss on original images to DSA objective (ablation)DSA · CelebAAccuracy (%): 88.32 -> 90.95Improved
add CE loss on adversarial examples to DSA objective (ablation)DSA · CelebAAccuracy (%): 88.54 -> 90.95Improved
add attention weights alignment regularizer (ablation)DSA · CelebADP: 0.2865 -> 0.2337Improved
add attention weights alignment regularizer (ablation)DSA · WaterbirdsEO: 0.0285 -> 0.0185Improved
perturb 5 patches instead of 3 (ablation)DSA · CelebAAccuracy (%): 90.95 -> 88.55Hurt
perturb only 1 patch instead of 3 (ablation)DSA · CelebADP: 0.2337 -> 0.3075Hurt
use AT discrepancy instead of MSE for attention alignment (ablation)DSA · CelebAEO: 0.2706 -> 0.2558Improved
use AT instead of KL-Div for attention alignment (ablation)DSA · CelebABalanced Accuracy: 83.26 -> 82.92Mixed
use attention masking from bias-only model instead of adversarial attack (AM vs DSA) (ablation)DSA · CelebADBA: 0.0256 -> 0.0031Improved

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