报告题目：Generative adversarial mediation network: A novel generative learning approach to causal mediation analysis

报 告 人：宋心远 教授

报告时间：2024年9月6日 10：50-11：50

报告地点： 格物楼数学研究中心528室

报告摘要：Casual mediation analysis (CMA) plays an essential role in various scientific fields. However, traditional models have restrictive parametric settings and strong distribution assumptions, which may not hold due to general nonlinearity, heterogeneity, and complex noise effects in many applications. Motivated by the similarities between CMA and image-to-image translation, this study proposes a novel prototype called the Generative Adversarial Mediation Network (GAMN) to explore the generative learning approach in the context of CMA. Thanks to a new encoding scheme for random terms, carefully designed partially linear architecture, and inherent advantages of the generative learning framework, GAMN can flexibly handle nonlinear covariate effects and effectively model complex noise and heterogeneous mediating mechanisms with minimal model assumptions. Thus, when encountering intricate data patterns, the counterfactuals relating to treatment effects in CMA can be efficiently inferred, providing highly reliable mediation results. Experiments conducted on synthetic and realistic datasets demonstrate that, compared with state-of-the-art approaches, GAMN can achieve notably more accurate estimations of out-of-sample predictions and treatment/mediation effects, further illustrating our method's utility and advantages. With the novel reinterpretations and solid theoretical results, this study also substantially broadens insights into developing mediation models from a machine-learning perspective.

报告人简介：宋心远教授是香港中文大学统计系主任、教育部长江学者特聘讲座教授，主要研究方向是潜变量模型、贝叶斯方法、统计计算和生存分析等。宋心远教授是多个著名国际期刊的副主编或编委（包括JASA、Biometrics、Statistics in Medicine、CSDA等).