Superclass-Guided Representation Disentanglement for Spurious Correlation Mitigation
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866918398638686208 |
|---|---|
| author | Liu, Chenruo Liu, Hongjun Lai, Zeyu Shen, Yiqiu Zhao, Chen Lei, Qi |
| author_facet | Liu, Chenruo Liu, Hongjun Lai, Zeyu Shen, Yiqiu Zhao, Chen Lei, Qi |
| contents | To enhance group robustness to spurious correlations, prior work often relies on auxiliary group annotations and assumes identical sets of groups across training and test domains. To overcome these limitations, we propose to leverage superclasses -- categories that lie higher in the semantic hierarchy than the task's actual labels -- as a more intrinsic signal than group labels for discerning spurious correlations. Our model incorporates superclass guidance from a pretrained vision-language model via gradient-based attention alignment, and then integrates feature disentanglement with a theoretically supported minimax-optimal feature-usage strategy. As a result, our approach attains robustness to more complex group structures and spurious correlations, without the need to annotate any training samples. Experiments across diverse domain generalization tasks show that our method significantly outperforms strong baselines and goes well beyond the vision-language model's guidance, with clear improvements in both quantitative metrics and qualitative visualizations. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2508_08570 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Superclass-Guided Representation Disentanglement for Spurious Correlation Mitigation Liu, Chenruo Liu, Hongjun Lai, Zeyu Shen, Yiqiu Zhao, Chen Lei, Qi Computer Vision and Pattern Recognition Artificial Intelligence Machine Learning To enhance group robustness to spurious correlations, prior work often relies on auxiliary group annotations and assumes identical sets of groups across training and test domains. To overcome these limitations, we propose to leverage superclasses -- categories that lie higher in the semantic hierarchy than the task's actual labels -- as a more intrinsic signal than group labels for discerning spurious correlations. Our model incorporates superclass guidance from a pretrained vision-language model via gradient-based attention alignment, and then integrates feature disentanglement with a theoretically supported minimax-optimal feature-usage strategy. As a result, our approach attains robustness to more complex group structures and spurious correlations, without the need to annotate any training samples. Experiments across diverse domain generalization tasks show that our method significantly outperforms strong baselines and goes well beyond the vision-language model's guidance, with clear improvements in both quantitative metrics and qualitative visualizations. |
| title | Superclass-Guided Representation Disentanglement for Spurious Correlation Mitigation |
| topic | Computer Vision and Pattern Recognition Artificial Intelligence Machine Learning |
| url | https://arxiv.org/abs/2508.08570 |