搜索
报告题目：Data-Driven Selection of Instrumental Variables for Additive Non-Parametric Causal Models
报 告 人：谢峰 北京工商大学
报告摘要：In this presentation, we address the testability of instrumental variables derived from observational data. While much of the existing literature on testable implications deals with discrete treatments or assumes constant causal effects, real-world applications often involve continuous treatments (e.g., drug dosages, nutritional levels) and varying effects. We propose an Auxiliary-based Independence Test (AIT) condition that can be used to assess the validity of an instrument in an additive nonlinear, non-constant effects model. We first show that if the candidate instrument is valid, then the AIT condition holds. Moreover, we illustrate the implications of the AIT condition and demonstrate that, in certain conditions, AIT conditions are necessary and sufficient to detect all invalid instruments. We also extend the AIT condition to incorporate covariates and present a practical algorithm for its implementation. Experimental evaluations using synthetic and real-world data sets demonstrate the effectiveness and robustness of our approach.
报告人简介：谢峰，北京工商大学 数学与统计学院副教授、应用统计系系主任，博士生导师。兼任全国工业统计学教学研究会理事，中国现场统计研究会因果推断分会理事，中国人工智能学会因果与不确定性人工智能专委会委员。研究领域涵盖因果推断与人工智能理论，特别是因果发现机制、隐变量因果表达学习及其在社会学、经济学和生物学中的应用。研究成果发表于人工智能、机器学习领域重要国际学术会议ICML、NeurIPS、ICLR、AAAI、IJCAI和重要学术期刊JMLR、TNNLS、Neurocomputing等。
报告时间：2025年4月18日 14：30-16：30
报告地点：文渊楼 B434
主办单位：数学与统计学院
学院宣传片
