解耦偏置、对齐分布：面向深度伪造检测的协同公平性优化
Decoupling Bias, Aligning Distributions: Synergistic Fairness Optimization for Deepfake Detection
在模型架构层面解耦对人口统计属性敏感的通道、在特征层面对齐整体分布与各群体分布，这套「结构解耦+分布对齐」的双机制可直接迁移到人脸属性识别、人脸验证等易受群体偏置影响的视觉任务中，在不牺牲精度的情况下提升群体公平性。
原文依据 decoupling channels sensitive to demographic groups at the model architectural level
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论公平性是深度伪造检测模型可信部署的核心要素，在数字身份安全领域尤为如此。检测模型对性别、种族等不同人口统计群体存在偏置，可能导致系统性误判，进而加剧数字鸿沟与社会不平等。然而，现有的公平性增强检测器往往以牺牲检测精度为代价来提升公平性。为应对这一挑战，我们提出一种双机制协同优化框架。所提方法创新性地融合了结构性公平解耦与全局分布对齐：在模型架构层面解耦对人口统计群体敏感的通道，随后在特征层面缩小整体样本分布与各人口统计群体对应分布之间的距离。实验结果表明，与其他方法相比，我们的框架在跨域保持整体检测精度的同时，提升了组间与组内公平性。代码已发布于 https://github.com/ywh1093/Fairness-Optimization。
英文原文摘要 1143 字符
Fairness is a core element in the trustworthy deployment of deepfake detection models, especially in the field of digital identity security. Biases in detection models toward different demographic groups, such as gender and race, may lead to systemic misjudgments, exacerbating the digital divide and social inequities. However, current fairness-enhanced detectors often improve fairness at the cost of detection accuracy. To address this challenge, we propose a dual-mechanism collaborative optimization framework. Our proposed method innovatively integrates structural fairness decoupling and global distribution alignment: decoupling channels sensitive to demographic groups at the model architectural level, and subsequently reducing the distance between the overall sample distribution and the distributions corresponding to each demographic group at the feature level. Experimental results demonstrate that, compared with other methods, our framework improves both inter-group and intra-group fairness while maintaining overall detection accuracy across domains. The code is available at https://github.com/ywh1093/Fairness-Optimization.