Research map / Label-efficient and robust learning / Recognition and long-tail
Discover the Unknown Biased Attribute of an Image Classifier
Zhiheng Li, Chenliang Xu
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add orthogonalization penalty (ablation) | total variation loss method · SmallNORB | Delta cos: -0.02 -> 0.12 | Improved |
| add orthogonalization penalty (ablation) | total variation loss method · dSprites | Delta cos: -0.01 -> 0.05 | Improved |
| add orthogonalization penalty (ablation) | total variation loss method · SmallNORB | |cos(w_b_hat,w_t)|: 0.27 -> 0.10 | Improved |
| add orthogonalization penalty (ablation) | total variation loss method · dSprites | |cos(w_b_hat,w_t)|: 0.21 -> 0.13 | Improved |
| add orthogonalization penalty (ablation) | total variation loss method · SmallNORB | |cos(w_b_hat,w_b)|: 0.25 -> 0.23 | Hurt |
| add orthogonalization penalty (ablation) | total variation loss method · dSprites | |cos(w_b_hat,w_b)|: 0.20 -> 0.17 | Hurt |
| add Hessian Penalty to full method (ablation) | Ours · SmallNORB | Delta cos: 0.12 -> 0.10 | Hurt |
| add Hessian Penalty to full method (ablation) | Ours · dSprites | Delta cos: 0.05 -> 0.01 | Hurt |
| add Hessian Penalty to full method (ablation) | Ours · SmallNORB | |cos(w_b_hat,w_t)|: 0.10 -> 0.15 | Hurt |
| add Hessian Penalty to full method (ablation) | Ours · SmallNORB | |cos(w_b_hat,w_b)|: 0.23 -> 0.25 | Improved |
| use proposed method instead of Hessian Penalty | unsupervised hyperplane discovery · CelebA | Delta cos: 0.0005 -> 0.06 | Improved |
| use proposed method instead of Hessian Penalty | unsupervised hyperplane discovery · FFHQ | Delta cos: 0.03 -> 0.17 | Improved |
Closest papers by meaning
- Classifier-to-Bias: Toward Unsupervised Automatic Bias Detection for Visual Classifiers (CVPR 2025 · similarity 0.87)
- BIAS-ID: A Framework for Analyzing Transformation Biases in AI-Generated Image Detectors (NeurIPS 2026 · similarity 0.87)
- Discover and Mitigate Multiple Biased Subgroups in Image Classifiers (CVPR 2024 · similarity 0.86)
- Enhancing Intrinsic Features for Debiasing via Investigating Class-Discerning Common Attributes in Bias-Contrastive Pair (CVPR 2024 · similarity 0.86)
- Discovering and Mitigating Visual Biases through Keyword Explanation (CVPR 2024 · similarity 0.85)
- Disentangled Representation Learning via Modular Compositional Bias (NeurIPS 2025 · similarity 0.84)
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.
