Superclass-Guided Representation Disentanglement for Spurious Correlation Mitigation

Chenruo Liu, Hongjun Liu, Zeyu Lai, Yiqiu Shen, Chen Zhao, Qi Lei
Conference on Parsimony and Learning, PMLR 328:756-794, 2026.

Abstract

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.

Cite this Paper

BibTeX
@InProceedings{pmlr-v328-liu26a, title = {Superclass-Guided Representation Disentanglement for Spurious Correlation Mitigation}, author = {Liu, Chenruo and Liu, Hongjun and Lai, Zeyu and Shen, Yiqiu and Zhao, Chen and Lei, Qi}, booktitle = {Conference on Parsimony and Learning}, pages = {756--794}, year = {2026}, editor = {Burkholz, Rebekka and Liu, Shiwei and Ravishankar, Saiprasad and Redman, William and Huang, Wei and Su, Weijie and Zhu, Zhihui}, volume = {328}, series = {Proceedings of Machine Learning Research}, month = {23--26 Mar}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v328/main/assets/liu26a/liu26a.pdf}, url = {https://proceedings.mlr.press/v328/liu26a.html}, abstract = {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.} }
Endnote
%0 Conference Paper %T Superclass-Guided Representation Disentanglement for Spurious Correlation Mitigation %A Chenruo Liu %A Hongjun Liu %A Zeyu Lai %A Yiqiu Shen %A Chen Zhao %A Qi Lei %B Conference on Parsimony and Learning %C Proceedings of Machine Learning Research %D 2026 %E Rebekka Burkholz %E Shiwei Liu %E Saiprasad Ravishankar %E William Redman %E Wei Huang %E Weijie Su %E Zhihui Zhu %F pmlr-v328-liu26a %I PMLR %P 756--794 %U https://proceedings.mlr.press/v328/liu26a.html %V 328 %X 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.
APA
Liu, C., Liu, H., Lai, Z., Shen, Y., Zhao, C. & Lei, Q.. (2026). Superclass-Guided Representation Disentanglement for Spurious Correlation Mitigation. Conference on Parsimony and Learning, in Proceedings of Machine Learning Research 328:756-794 Available from https://proceedings.mlr.press/v328/liu26a.html.

Related Material