Multivariate density estimation and its applications
主 题: Multivariate density estimation and its applications
报告人: 王永雄（COPSS奖获得者，美国科学院院士，中央研究院院士，大阳城2138统计科学中心顾问委员会主任）
时 间: 2013-07-11 9:00-10:00
地 点: 北京国际数学研究中心报告厅（未名湖北岸82号甲乙丙，全斋西边）
We will examine the problem of density estimation in k-dimensional space when the data dimension is moderate (k=3 to 5) and the sample size is large (n=10^3 to 10^6). The key to solving this problem is the learning of a partition of the underlying space so that the density is well approximated by simple functions supported on that partition. In this talk recent theoretical and algorithmic advances on this problem will be reviewed. We argue that a good density estimator is a building block that enables the development of statistical methods for more complex problems. We will illustrate this point by examples from data compression, flow cytometry and image analysis.
