报告人：夏应存（新加坡国立大学统计与应用概率系教授）
报告题目：A kernel estimation of the mutual information
报告人介绍：夏应存教授现为新加坡国立大学统计与应用概率系教授，博士生导师，主要从事非线性时间序列，半参数模型，非线性降维方法等方面的研究。夏应存教授在The Annals of Statistics，JRSSB，JASA，Biometrika，The American Naturalist和Journal of Econometrics等世界顶级学术杂志发表论文20余篇。其研究成果在世界上产生了重要影响，其中一篇论文被《Nature News》等多家学术媒体专题报道，三篇论文被Journal of the Royal Statistical Society (series B)等杂志列为重点讨论文章。夏应存教授现为统计学顶级杂志The Annals of Statistics副主编，SCI杂志Computational Statistics副主编,以及JRSSB，JASA，Biometrika，Journal of Econometrics等国际著名学术杂志审稿人。
报告摘要：Quantifying the strength of dependence between two random variables is a fundamental issue in data analysis.  Recent research has been focused on the mutual information (MI) after Reshef et al. (Science, 2011).“Unfortunately, reliably estimating mutual information from finite continuous data remains a significant and unresolved problem”(Kinney and Atwal, PNAS, 2014).We consider in this paper a kernel estimation of MI and find that bandwidths involved need be unified. We propose a jackknifed approach and estimate MI by the maximum value which is shown to be unique with respect to the bandwidth. The estimator is consistent in detecting the dependence, while it has an“oracle”statistical efficiency. We also compare our estimator with existing methods by simulations.

报告时间：2017年4月14日（星期五）下午15：00-16：00.
报告地点：科技楼南楼702