发表机构
University of Zurich（苏黎世大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
该研究针对人脸识别系统的人口统计偏差问题，提出DeSCon损失方法，通过优化训练批次与样本对选择，在保持验证性能的同时提升了公平性。
AI 中文摘要
人脸识别系统已被证实对特定人口统计群体存在偏差，表现为不同性别、年龄或种族群体的错误率存在差异。尽管训练数据在这些人口统计维度上的不平衡是导致该偏差的原因之一，但对人工平衡的群体进行训练并不能完全缓解该问题。在实际部署中，人脸识别通常工作在允许极低错误匹配率的操作点，因此针对非匹配分数分布的尾部区域。类别平衡可改善这些分布的均值，而本研究方法的目标是通过解决尾部区域的行为来提升公平性。具体而言，我们提出了用于人脸识别的基于人口统计的监督对比损失（DeSCon），该方法依赖于精心设计的训练批次构成和人口统计感知的样本对选择。我们在带有人口统计标签的数据集及标准验证基准上开展的实验评估表明，DeSCon在保持具有竞争力的验证性能的同时，能够比平衡训练数据集更好地提升公平性。源代码可应要求提供。
英文摘要
Face recognition systems have been shown to be biased toward certain demographic groups by exhibiting different error rates across gender, age, or ethnicity. Though the imbalance of the training data with respect to these demographics is one cause of this bias, training on artificially balanced groups does not completely mitigate the problem. For deployment, face recognition typically works at operating points allowing very low false match rates and, hence, on the tail of the non-match score distribution. While class balancing can improve the means of these distributions, the aim of our approach is to improve fairness by addressing the behavior in the tail. Particularly, we propose the Demographic-based Supervised Contrastive loss (DeSCon) for face recognition, which relies on a well-designed composition of training batches and demographic-aware pair selection. Our experimental evaluation on both demographically-labeled datasets and standard verification benchmarks shows that DeSCon can improve fairness beyond balancing training datasets while maintaining competitive verification performance. Source code is available upon request.
Comments8 pages, 1 figure, 5 tables
