Focusing Image Generation to Mitigate Spurious Correlations
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2024
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866915081143451648 |
|---|---|
| author | Li, Xuewei Nie, Zhenzhen Yu, Mei Zhang, Zijian Gao, Jie Xu, Tianyi Liu, Zhiqiang |
| author_facet | Li, Xuewei Nie, Zhenzhen Yu, Mei Zhang, Zijian Gao, Jie Xu, Tianyi Liu, Zhiqiang |
| contents | Instance features in images exhibit spurious correlations with background features, affecting the training process of deep neural classifiers. This leads to insufficient attention to instance features by the classifier, resulting in erroneous classification outcomes. In this paper, we propose a data augmentation method called Spurious Correlations Guided Synthesis (SCGS) that mitigates spurious correlations through image generation model. This approach does not require expensive spurious attribute (group) labels for the training data and can be widely applied to other debiasing methods. Specifically, SCGS first identifies the incorrect attention regions of a pre-trained classifier on the training images, and then uses an image generation model to generate new training data based on these incorrect attended regions. SCGS increases the diversity and scale of the dataset to reduce the impact of spurious correlations on classifiers. Changes in the classifier's attention regions and experimental results on three different domain datasets demonstrate that this method is effective in reducing the classifier's reliance on spurious correlations. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2412_19457 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Focusing Image Generation to Mitigate Spurious Correlations Li, Xuewei Nie, Zhenzhen Yu, Mei Zhang, Zijian Gao, Jie Xu, Tianyi Liu, Zhiqiang Computer Vision and Pattern Recognition Instance features in images exhibit spurious correlations with background features, affecting the training process of deep neural classifiers. This leads to insufficient attention to instance features by the classifier, resulting in erroneous classification outcomes. In this paper, we propose a data augmentation method called Spurious Correlations Guided Synthesis (SCGS) that mitigates spurious correlations through image generation model. This approach does not require expensive spurious attribute (group) labels for the training data and can be widely applied to other debiasing methods. Specifically, SCGS first identifies the incorrect attention regions of a pre-trained classifier on the training images, and then uses an image generation model to generate new training data based on these incorrect attended regions. SCGS increases the diversity and scale of the dataset to reduce the impact of spurious correlations on classifiers. Changes in the classifier's attention regions and experimental results on three different domain datasets demonstrate that this method is effective in reducing the classifier's reliance on spurious correlations. |
| title | Focusing Image Generation to Mitigate Spurious Correlations |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2412.19457 |