报告时间: 6月19日（周一） 14:00-15:00
报告地点：维格堂319
摘 要：Conformal inference is a popular framework for producing distribution-free uncertainty quantification for machine learning prediction models. In this talk, I will introduce fundamental concepts and some recent developments in conformal inference. Then, I will discuss its interface with selective inference and multiple hypothesis testing, and how this can be applied to reliable discovery of new drugs with the aid of machine learning prediction models.
报告人简介：金滢 (Ying Jin)，斯坦福大学统计系博士，导师为Emmanuel Candes教授及Dominik Rothenhaeusler教授。 主要从事distribution-free inference, causal inference, multiple hypothesis testing, data-driven decision making等相关方面的研究，研究成果发表在PNAS, Biometrika, Journal of Machine Learning Research等国际知名期刊及ICML, NeurIPS, ISIT等国际会议。
邀请人：王 奎