报告题目：Understanding Unfairness via Training Concept Influence

报告人：刘扬

报告时间：2023年7月14日，上午10-11点。

报告地点：综合楼601会议室。

腾讯会议号：498-558-725

报告人简介：

刘扬是字节跳动人工智能实验室的Responsible AI团队的负责人，同时还是加州大学圣克鲁兹分校计算机科学与工程系助理教授。曾在哈佛大学从事博士后研究（2016 - 2018）。2015年，他从密歇根大学安娜堡分校EECS系获得博士学位。主要研究方向是开发公平、稳健的机器学习算法，以应对有偏差和偏移数据的挑战。刘扬是美国国家科学基金会CAREER奖的获得者，曾被选中参与多个备受瞩目的项目，包括NSF-Amazon Fairness in AI、DARPA SCORE和IARPA HFC。他的研究成果已在FICO和亚马逊进行了部署，最近的工作在相关研讨会上获得了四个最佳论文奖。

摘要：Knowing the causes of a model's unfairness helps practitioners better understand their data and algorithms. This is an important yet relatively unexplored task. We look into this problem through the lens of the training data -- one of the major sources of unfairness. We ask the following questions: how would a model's fairness performance change if, in its training data, some samples (1) were collected from a different (demographic) group, (2) were labeled differently, or (3) some features were changed? In other words, we quantify the fairness influence of training samples by counterfactually intervening and changing samples based on predefined concepts, i.e., data attributes such as features ($X$), labels ($Y$), or sensitive attributes ($A$). To calculate a training sample's influence on the model's unfairness w.r.t a concept, we first generate counterfactual samples based on the concept, i.e., the counterfactual versions of the sample if the concept were changed. We then calculate the resulting impact on the unfairness, via influence function, if the counterfactual samples were used in training. Our framework not only helps practitioners understand the observed unfairness and repair their training data, but also leads to many other applications, e.g., detecting mislabeling, fixing imbalanced representations, and detecting fairness-targeted poisoning attacks.