[Ml-stat-talks] [stat seminar]Tracy Ke, Fri. April 8th, 12:30pm

xtong at princeton.edu xtong at princeton.edu
Thu Apr 7 21:14:07 EDT 2011

 Regression with coefficient homogeneity
 Tracy Ke, Princeton University
 Friday April 8th, 12:30pm
 ORFE, statistics Lab


 In high dimensional linear regression, we assume homogeneity among
covariates: they form sub-groups and share the same regression
 coefficient within each group. Compared to the usual "sparsity"
assumption, homogeneity assumes another type of low dimensional
 structure. In application area such as gene analysis, homogeneous
subgroups are frequently observed.

 Based on this assumption, we develop a procedure aiming to efficiently
recover the group structure and achieve high estimation power. A
 revision of this procedure can deal with both sparsity and homogeneity.
We provide theoretical results and simulating examples.

 This is joint work with Prof. Jianqing Fan and Prof. Yichao Wu.


[1] mailto:ml-stat-talks at lists.cs.princeton.edu
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.cs.princeton.edu/pipermail/ml-stat-talks/attachments/20110407/1d666ae4/attachment.htm>

More information about the Ml-stat-talks mailing list