题目：EM-test for finite mixture models

嘉宾：陈家骅教授（加拿大UBC）

时间：2016年6月13日下午4:30-5:30

地点：管理科研大楼EMBA第二教室

摘要：

In scientific investigations, a population is often suspected of containing several more homogeneous sub-populations. Such a population structure is most accurately described by a finite mixture model. The evidence for mixture is best examined through a rigorous statistical hypothesis test. Developing valid and effective statistical inference methods is an important

and challenging research problem. Classical procedures when applied to mixture models often have sophisticated asymptotic properties which render them useless in applications.

For many finite mixture models, we have successfully designed corresponding EM-tests whose limiting distributions are easier to derive mathematically and simpler for implementation in data analysis. This talk illustrates the ideas behind the EM-tests, their elegant asymptotic properties and other related issues.

陈家骅，加拿大不列颠哥伦比亚大学(University of British Columbia)统计系教授, 加拿大国家（一级）讲座教授（Canada Research Chair , Tier I, 2007-2013, 2014-2020)，云南大学“创新人才计划”教授。1982年于中国科学技术大学获得数学学士学位， 1985年于中国科学院系统科学研究所获得统计学硕士学位，1990年与美国威斯康辛大学统计学系获得博士学位（导师：吴建福教授, C.F. Jeff Wu）。2015年起云南大学“创新人才计划”教授。陈家骅教授在统计学诸多领域都作出了重要贡献：早期师从吴建福教授研究试验设计，其后从事混合模型、遗传统计学、抽样理论、经验似然和变量选择方面的研究。已在国际统计学顶级杂志如Annals of Statistics, JASA, JRSSB, Biometrika等上发表论文110余篇，其中至少52篇论文的被引次数超过10，引用次数超过100的论文有7篇。陈家骅教授是多个具有影响力的国际统计杂志的主编或者副主编，比如Canadian Journal of Statistics主编，以及Statistica Sinica，Quality Technology and Quality Management的副主编等。