Sparsity oracle inequalities in high-dimensional statistical problems
University of Edinburgh
School of Mathematics
Statistics seminars, Spring 2010
Friday 26 February, 3:15pm, JCMB 5327 - Karim Lounici (University of
Cambridge)
"Sparsity oracle inequalities in high-dimensional statistical problems"
Abstract.
This talk is about statistical learning in high-dimension, that is when the number of parameters to estimate is larger than the sample size. In this context, the generally adopted assumption is that the number of active parameters is much smaller than the number of potential parameters. This assumption is called the "sparsity assumption". We study the statistical properties of two types of procedures: penalized risk minimization procedures with l_1-type penalty such as the Lasso or the group Lasso and exponential weights procedures.
From: Natalia Bochkina n.bochkina@ed.ac.uk
