Research map / Label-efficient and robust learning / Recognition and long-tail
Regulating Internal Evidence Flows for Robust Learning Under Spurious Correlations
Rajeev Ranjan Dwivedi, Mohammedkaif Mohammedrafiq Kalagond, Niramay M.Patel, Vinod K. Kurmi
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 confidence weighting pk(x) (ablation) | AGS · Waterbirds | worst-group accuracy: 73.9 -> 79.4 | Improved |
| add EMA smoothing of energies (ablation) | AGS · Waterbirds | worst-group accuracy: 75.2 -> 79.4 | Improved |
| use parameter-space alignment instead of EvA-style activation proxy (ablation) | AGS · Waterbirds | worst-group accuracy: 70.1 -> 79.4 | Improved |
| add confidence weighting to ERM (ablation) | ERM · Waterbirds | worst-group accuracy: 63.7 -> 68.2 | Improved |
| add EMA smoothing (ablation) | ERM + confidence weighting · Waterbirds | worst-group accuracy: 68.2 -> 72.1 | Improved |
| add percentile gating (ablation) | ERM + confidence weighting + EMA · Waterbirds | worst-group accuracy: 72.1 -> 80.9 | Improved |
| add percentile gating (ablation) | ERM + confidence weighting + EMA · Waterbirds | average accuracy: 95.4 -> 97.4 | Improved |
| add AGS to ERM | ERM · ImageNet-9 Mixed-Rand | accuracy: 80.17 -> 82.47 | Improved |
| add AGS to ERM | ERM · ImageNet-9 background gap | BG-Gap: 10.94 -> 11.84 | Hurt |
| add AGS to ERM | ERM · BAR | average accuracy: 60.51 -> 76.09 | Improved |
| add AGS to ERM | ERM · CelebA (bias-conflicting split) | accuracy: 52.52 -> 93.95 | Improved |
| add AGS to ERM | ERM · Waterbirds | worst-group accuracy: 63.74 -> 80.93 | Improved |
Closest papers by meaning
- Sigmoid Gating is More Sample Efficient than Softmax Gating in Mixture of Experts (NeurIPS 2024 · similarity 0.85)
- NeuronTune: Towards Self-Guided Spurious Bias Mitigation (ICML 2025 · similarity 0.84)
- The Evidence Contraction Issue in Deep Evidential Regression: Discussion and Solution (AAAI 2024 · similarity 0.83)
- Uncertainty Regularized Evidential Regression (AAAI 2024 · similarity 0.83)
- The Effectiveness of Random Forgetting for Robust Generalization (ICLR 2024 · similarity 0.83)
- Let Samples Speak: Mitigating Spurious Correlation by Exploiting the Clusterness of Samples (CVPR 2025 · similarity 0.83)
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.
