Research map / Detection and segmentation / Open-set and few-shot detection
Salient ImageNet: How to discover spurious features in Deep Learning?
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use robust-model neural features, limited worker labels, and NAMs to generalize visual-feature annotations | Salient ImageNet feature annotation · ImageNet, spurious visual-attribute validation | majority agreement: 95% | Helped |
| Corrupt spurious regions rather than core regions | Standard ResNet-50 · ImageNet, drake | accuracy drop: 41.5% vs. 4.6% | Mixed |
| Corrupt a spurious visual feature with Gaussian noise | Standard ResNet-50 · ImageNet, wok | accuracy drop: -46.154% | Hurt |
| Corrupt a spurious visual feature with Gaussian noise | Standard ResNet-50 · ImageNet, band aid | accuracy drop: -41.538% | Hurt |
| Corrupt a spurious visual feature with Gaussian noise | Standard ResNet-50 · ImageNet, sorrel | accuracy drop: -12.308% | Hurt |
| Corrupt spurious regions rather than core regions | Standard pretrained models · ImageNet, multiple classes and corruption levels | accuracy drop | No clear effect |
| Use standard accuracy alone to assess model reliability | ImageNet classifiers · ImageNet, triumphal arch and ostrich | core accuracy: 100% standard accuracy for both; ≈ 40% vs. almost 0% core accuracy at σ=2.0 | Mixed |
Closest papers by meaning
- Hard ImageNet: Segmentations for Objects with Strong Spurious Cues (NeurIPS 2022 · similarity 0.87)
- Complexity Matters: Feature Learning in the Presence of Spurious Correlations (ICML 2024 · similarity 0.86)
- How Spurious Features are Memorized: Precise Analysis for Random and NTK Features (ICML 2024 · similarity 0.85)
- A Sober Look at the Robustness of CLIPs to Spurious Features (NeurIPS 2024 · similarity 0.85)
- Let Samples Speak: Mitigating Spurious Correlation by Exploiting the Clusterness of Samples (CVPR 2025 · similarity 0.85)
- Severing Spurious Correlations with Data Pruning (ICLR 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.