上 海 大 学 经 济 学 院
青年学者论坛第43期,总第252期
演讲题目：一个脸的匿名感知范式及其在网络约会上的应用
演 讲 人：周影辉博士 复旦大学
时 间：12月09日（周五）上午10:00－11:30
地 点：校本部东区经管大楼520会议室
主 办：威斯尼斯人5158cc
威斯尼斯人5158cc青联会
威斯尼斯人5158cc金融信息研究中心
讲座简介：
Faces play an important role in decision making processes in many contexts (e.g., security screening, hiring decisions, online dating); yet, there is also a strong need to protect facial privacy. The authors propose a face anonymity-perceptibility (FAP) paradigm as a general framework for solving three types of tradeoff problems (non-perception, component perception and holistic perception) between face anonymization and face perception. Four different methods can be applied to achieve particular type(s) of tradeoffs: feature reduction and feature replacement from the computer vision literature, and two face abstraction methods (local abstraction and global abstraction) proposed by the authors and grounded in the social psychology and neuropsychology literature. The authors empirically test the FAP paradigm by applying the local abstraction and global abstraction methods to the online dating industry, where Type III (i.e., holistic perception) tradeoff is desired. The results demonstrate that both methods are effective in balancing face anonymization and perception. When evaluated on their predictive performances as screening tools, the proposed face abstraction-based screening (FAS) models generate a median relative improvement of 36.1% over a model without abstracted facial information.
报告人简介：
周影辉博士，复旦大学管理学院工商管理博士后流动站在站博士后。研究兴趣包括：人工神入(图像和视频分析)，及图像和视频分析在营销上的应用；机器学习和贝叶斯统计及其在营销上的应用。他的研究成果多次发表在统计学、管理学和市场营销学的国内外期刊上，包括Journal of Consumer Marketing，Management and Organization Review，Communications in Statistics - Simulation and Computation，管理世界，中国管理科学，中国工业经济等。他最新的研究项目关注于网络约会网站上基于脸的推荐系统及用户隐私保护，复杂图像产品的偏好测量等。