人类审核员检测误报的能力取决于人脸识别算法及其安全配置
原标题：Detection of false positives by human reviewers depends on the face recognition algorithm and its security configuration
AI 摘要
研究团队使用基于四款商用面部识别算法误报构建的困难面部图像比对测试集（DFICT），测量了801名人类审核员在三种安全配置下检测算法误报的能力。结果显示，人类审核员能够减少算法误报，但在更严格的安全阈值下表现下降；且审核员表现因算法而异，表明部分算法更适合人工监督。作者建议生物识别系统设计应考虑审核员效能差异，尤其在高安全场景中。
正文节选
[Page 1] Detection of false positives by human reviewers depends on the face recognition algorithm and its security configuration Esha Edakoth, John J. Howard, Isabelle M. Shuggi*, Yevgeniy B. Sirotin, Robert Wood Authors listed alphabetically. *Corresponding author: ishuggi@idslabs.org ABSTRACT Integration of artificial intelligence (AI) technology into identity workflows, such as automated face recognition (FR), has enhanced operational capabilities, with FR algorithms achieving s