Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2401.13213 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866909461834104832 |
|---|---|
| author | Zhang, Miao fryer, Zee Colman, Ben Shahriyari, Ali Bharaj, Gaurav |
| author_facet | Zhang, Miao fryer, Zee Colman, Ben Shahriyari, Ali Bharaj, Gaurav |
| contents | Machine learning model bias can arise from dataset composition: correlated sensitive features can distort the downstream classification model's decision boundary and lead to performance differences along these features. Existing de-biasing works tackle the most prominent bias features, such as colors of digits or background of animals. However, real-world datasets often include a large number of feature correlations that intrinsically manifest in the data as common sense information. Such spurious visual cues can further reduce model robustness. Thus, domain practitioners desire a comprehensive understanding of correlations and the flexibility to address relevant biases. To this end, we propose a novel framework to extract comprehensive biases in image datasets based on textual descriptions, a common sense-rich modality. Specifically, features are constructed by clustering noun phrase embeddings with similar semantics. The presence of each feature across the dataset is inferred, and their co-occurrence statistics are measured, with spurious correlations optionally examined by a human-in-the-loop module. Downstream experiments show that our method uncovers novel model biases in multiple image benchmark datasets. Furthermore, the discovered bias can be mitigated by simple data re-weighting to de-correlate the features, outperforming state-of-the-art unsupervised bias mitigation methods. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2401_13213 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Common-Sense Bias Modeling for Classification Tasks Zhang, Miao fryer, Zee Colman, Ben Shahriyari, Ali Bharaj, Gaurav Computer Vision and Pattern Recognition Machine learning model bias can arise from dataset composition: correlated sensitive features can distort the downstream classification model's decision boundary and lead to performance differences along these features. Existing de-biasing works tackle the most prominent bias features, such as colors of digits or background of animals. However, real-world datasets often include a large number of feature correlations that intrinsically manifest in the data as common sense information. Such spurious visual cues can further reduce model robustness. Thus, domain practitioners desire a comprehensive understanding of correlations and the flexibility to address relevant biases. To this end, we propose a novel framework to extract comprehensive biases in image datasets based on textual descriptions, a common sense-rich modality. Specifically, features are constructed by clustering noun phrase embeddings with similar semantics. The presence of each feature across the dataset is inferred, and their co-occurrence statistics are measured, with spurious correlations optionally examined by a human-in-the-loop module. Downstream experiments show that our method uncovers novel model biases in multiple image benchmark datasets. Furthermore, the discovered bias can be mitigated by simple data re-weighting to de-correlate the features, outperforming state-of-the-art unsupervised bias mitigation methods. |
| title | Common-Sense Bias Modeling for Classification Tasks |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2401.13213 |