报告人：刘旭 教授（上海财经大学）

时间：2024年12月27日 10:30--

地点：数统学院LD718

摘要：One of the most fundamental tasks in statistics and artificial intelligence (AI) is to learn how the explanatory variable X affects the response variable Y, which can be naturally characterized as learning the conditional distribution P of Y given X. To completely understand how X affects Y, the conditional distribution P is essential. Our approach simultaneously estimates the high-dimensional regression parameter and the conditional generator using a generative learning framework, where the conditional generator is a function that can generate samples from a conditional distribution. We establish the non-asymptotic upper bound for estimation errors of generator and regression coefficients. Extensive simulation studies are conducted to show the good performance in the finite samples. A case study is also analyzed to demonstrate the well-performed applications.

邀请人: 周国立

欢迎广大师生积极参与！