🤖 AI Summary
This work addresses algorithmic bias in AI models arising from implicit sensitive attributes—such as gender, age, and race—in image datasets. To this end, we propose an end-to-end fairness auditing framework. Methodologically: (1) we design a lightweight CNN dedicated to extracting sensitive attribute features; (2) we introduce a novel robust statistical test that integrates prediction uncertainty to quantify distributional disparities across demographic subgroups; and (3) we adopt a weakly supervised paradigm, requiring only a small number of human annotations to fully automate the auditing pipeline. Experiments on two facial benchmark datasets demonstrate that our approach significantly improves both the accuracy and robustness of bias detection. The framework provides a reproducible, scalable, and technically grounded solution for regulatory compliance auditing of AI systems, enabling rigorous, data-driven fairness assessment without extensive labeling effort.
📝 Abstract
Auditing involves verifying the proper implementation of a given policy. As such, auditing is essential for ensuring compliance with the principles of fairness, equity, and transparency mandated by the European Union's AI Act. Moreover, biases present during the training phase of a learning system can persist in the modeling process and result in discrimination against certain subgroups of individuals when the model is deployed in production. Assessing bias in image datasets is a particularly complex task, as it first requires a feature extraction step, then to consider the extraction's quality in the statistical tests. This paper proposes a robust methodology for auditing image datasets based on so-called"sensitive"features, such as gender, age, and ethnicity. The proposed methodology consists of both a feature extraction phase and a statistical analysis phase. The first phase introduces a novel convolutional neural network (CNN) architecture specifically designed for extracting sensitive features with a limited number of manual annotations. The second phase compares the distributions of sensitive features across subgroups using a novel statistical test that accounts for the imprecision of the feature extraction model. Our pipeline constitutes a comprehensive and fully automated methodology for dataset auditing. We illustrate our approach using two manually annotated datasets.