Evaluating the Fairness of Discriminative Foundation Models in Computer Vision
arXiv:2310.11867 · doi:10.1145/3600211.3604720
Abstract
We propose a novel taxonomy for bias evaluation of discriminative foundation models, such as Contrastive Language-Pretraining (CLIP), that are used for labeling tasks. We then systematically evaluate existing methods for mitigating bias in these models with respect to our taxonomy. Specifically, we evaluate OpenAI's CLIP and OpenCLIP models for key applications, such as zero-shot classification, image retrieval and image captioning. We categorize desired behaviors based around three axes: (i) if the task concerns humans; (ii) how subjective the task is (i.e., how likely it is that people from a diverse range of backgrounds would agree on a labeling); and (iii) the intended purpose of the task and if fairness is better served by impartiality (i.e., making decisions independent of the protected attributes) or representation (i.e., making decisions to maximize diversity). Finally, we provide quantitative fairness evaluations for both binary-valued and multi-valued protected attributes over ten diverse datasets. We find that fair PCA, a post-processing method for fair representations, works very well for debiasing in most of the aforementioned tasks while incurring only minor loss of performance. However, different debiasing approaches vary in their effectiveness depending on the task. Hence, one should choose the debiasing approach depending on the specific use case.
Accepted at AIES'23
References in corpus (17)
- Learning Transferable Visual Models From Natural Language Supervision
- On the Opportunities and Risks of Foundation Models
- Equality of Opportunity in Supervised Learning
- Cross-lingual Language Model Pretraining
- Zero-Shot Text-to-Image Generation
- Fairness Beyond Disparate Treatment & Disparate Impact: Learning Classification without Disparate Mistreatment
- Reproducible scaling laws for contrastive language-image learning
- LAION-400M: Open Dataset of CLIP-Filtered 400 Million Image-Text Pairs
- Multimodal datasets: misogyny, pornography, and malignant stereotypes
- Mind the Gap: Understanding the Modality Gap in Multi-modal Contrastive Representation Learning
- A Step Toward More Inclusive People Annotations for Fairness
- Assaying Out-Of-Distribution Generalization in Transfer Learning
- Debiasing Vision-Language Models via Biased Prompts
- A Prompt Array Keeps the Bias Away: Debiasing Vision-Language Models with Adversarial Learning
- Benchmarking Robustness of Multimodal Image-Text Models under Distribution Shift
- Assessing Multilingual Fairness in Pre-trained Multimodal Representations
- Efficient fair PCA for fair representation learning