2024/05/23 by Chun-Fu Chen, Chen, Chun-Fu, Bill Moriarty +11
Computer Science · Decision Sciences · #Biometric Identification and Security #Data Quality and Management #FOS: Computer and information sciences #Image Processing and 3D Reconstruction #Machine Learning (cs.LG)
paper · pdf · doi:10.48550/arxiv.2405.15062
openalex publication_date 2024/05/23 · openalex created_date 2025/11/01 · openalex updated_date 2026/07/28
The recent rapid advancements in both sensing and machine learning\ntechnologies have given rise to the universal collection and utilization of\npeople's biometrics, such as fingerprints, voices, retina/facial scans, or\ngait/motion/gestures data, enabling a wide range of applications including\nauthentication, health monitoring, or much more sophisticated analytics. While\nproviding better user experiences and deeper business insights, the use of\nbiometrics has raised serious privacy concerns due to their intrinsic sensitive\nnature and the accompanying high risk of leaking sensitive information such as\nidentity or medical conditions.\n In this paper, we propose a novel modality-agnostic data transformation\nframework that is capable of anonymizing biometric data by suppressing its\nsensitive attributes and retaining features relevant to downstream machine\nlearning-based analyses that are of research and business values. We carried\nout a thorough experimental evaluation using publicly available facial, voice,\nand motion datasets. Results show that our proposed framework can achieve a\n highlighthigh suppression level for sensitive information, while at the same\ntime retain underlying data utility such that subsequent analyses on the\nanonymized biometric data could still be carried out to yield satisfactory\naccuracy.\n