5/30/2014
Post date: Jun 3, 2014 8:52:41 PM
Speaker: Fang Han, Dept of Biostatistics, JHU
Lecture: Theory for big data - Part II
Abstract: I briefly introduce some useful empirical process techniques for big data. These techniques fit to the ``large d'' scenario, can provide nonasymptotic analysis, and greatly help us understand the performance of many statistical methods. This starts from some simple probabilistic inequalities (Markov's inequality), and builds up through several stronger concentration results (Chernoff, Hoeffding, and McDiarmid), until we can give a complete analysis of the truncated normal mean estimator and Kendall's tau.