Classifier-to-Bias: Toward Unsupervised Automatic Bias Detection for Visual Classifiers

2025-04-29

AI Summary

Existing methods for detecting bias in vision classifiers rely heavily on human-annotated data and are inaccessible to non-expert users. Method: We propose the first fully unsupervised, training-free, task-agnostic, and bias-type-agnostic framework for automatic bias discovery. Given only a task’s textual description, it leverages large language models to generate testable bias hypotheses and corresponding image prompts; retrieves relevant images via cross-modal search; and evaluates model performance disparities along bias dimensions in a zero-shot manner—requiring neither labeled data nor model fine-tuning. Contribution/Results: Our framework significantly lowers the barrier to bias auditing. On two public benchmarks, it achieves higher bias detection rates than state-of-the-art supervised methods and uncovers novel bias patterns absent in the original datasets, establishing a scalable, low-threshold, general-purpose tool for fairness assessment of vision models.

Technology Category

• Computer Vision: Bias, Fairness & Privacy
• Machine Learning: Ethics, Bias, and Fairness
• Natural Language Processing: Ethics — Bias, Fairness, Transparency & Privacy

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling
• Social Networks and Social Media: Fairness and bias in social network and social media analysis

Abstract

A person downloading a pre-trained model from the web should be aware of its biases. Existing approaches for bias identification rely on datasets containing labels for the task of interest, something that a non-expert may not have access to, or may not have the necessary resources to collect: this greatly limits the number of tasks where model biases can be identified. In this work, we present Classifier-to-Bias (C2B), the first bias discovery framework that works without access to any labeled data: it only relies on a textual description of the classification task to identify biases in the target classification model. This description is fed to a large language model to generate bias proposals and corresponding captions depicting biases together with task-specific target labels. A retrieval model collects images for those captions, which are then used to assess the accuracy of the model w.r.t. the given biases. C2B is training-free, does not require any annotations, has no constraints on the list of biases, and can be applied to any pre-trained model on any classification task. Experiments on two publicly available datasets show that C2B discovers biases beyond those of the original datasets and outperforms a recent state-of-the-art bias detection baseline that relies on task-specific annotations, being a promising first step toward addressing task-agnostic unsupervised bias detection.

Problem

• Detects biases in visual classifiers without labeled data
• Uses textual task descriptions to generate bias proposals
• Works on any pre-trained model without training or annotations

Innovation

• Uses large language model for bias proposals
• Retrieval model collects images for captions
• Training-free, annotation-free bias detection framework