统计学系系列短课程

时　间:

5月10日（星期五）8:30-11:30

5月14日（星期二）8:30-11:30

5月17日（星期五）8:30-11:30

地　点：史带楼303室

主持人：朱仲义教授 复旦大学管理学院统计学系

主　题：Three vignettes in data science

主讲人：Ph.D. Yuekai Sun　Department of Statistics　University of Michigan

课程简介:

1. Introduction to stochastic optimization

In this lecture, we study algorithms for solving stochastic optimization problems of the from  

The methods we consider for this problem are simple first-order methods that converge more slowly than most advanced methods methods (e.g. Newton-type methods) for solving deterministic optimization problems. However, they are robust to noise in the optimization problem, which makes them particularly suitable for stochastic optimization problems. This lecture covers subgradient methods, mirror descent, and adaptive methods.

2. Introduction to multiple testing

Modern science and engineering routinely produce thousands and sometimes millions of parallel data sets, each with its own estimation or testing problem. In this lecture, we study basic methods to cope with multiple testing problems that arise in science and engineering. This lecture covers methods that control the family-wise error rate and the false discovery rate and a comparison of the power of the methods in a sandbox environment to elucidate their differences.

3. Introduction to causality

In this lecture, we study the mathematical foundations of causal inference. In causal inference, the goal is to study how a system changes under certain interventions (e.g. in gene knock-out experiments). These questions are not questions about a (probability) distribution and cannot be answered by statistical methods. This lecture covers structural causal models (SCM), interventional distributions, causal effects, and how to compute causal effects in a SCM.                       

 统计学系 

2019-4-16