Colloquium Mathematics, Prof.dr. Elio Mineo

 

Join us for coffee and tea at 15.45 p.m.

 

 

Date: Tuesday, April 10^th 2012

Speaker: Prof.dr. Elio Mineo
                University of Palermo

Room: 5161.0267 (Bernoulliborg)

Time: 16.15

 

 

 

 

Title: Sparse Inference in Generalized Linear Models: theoretical and computational 
          aspects of dgLARS

 

 

Abstract:

  

Sparsity is an essential feature of many contemporary data problems. Remote sensing, various forms of automated screening and other high-throughput measurement devices collect a lot of information, typically about few independent statistical subjects or units. In certain cases it is reasonable to assume that the underlying process generating the data is itself sparse, in the sense that only a few of the measured variables are involved in the process. In this talk we present a path following algorithm for generalized linear models that can be considered a generalization of least angle regression (LARS). In our approach we use di erential geometry to generalize the equiangular condition, on which the LARS algorithm is based. For this reason we call this method dgLARS. Although the geometry involves the Fisher information in a way that is not obvious in the simple regression setting, the equiangular condition turns out to be equivalent with an intuitive condition imposed on the Rao score test statistics. Although the computation of the solution paths is not trivial, the method compares favorably to other path following algorithms.

 

 

 

 

 

 

 

Colloquium coordinators are Prof.dr. A.C.D. van Enter (e-mail : A.C.D.van.Enter@rug.nl) and

Dr. A.V. Kiselev (e-mail: a.v.kiselev@rug.nl)