Display Abstract

Title Complexity estimates of Orthogonal Matching Pursuit under RIP conditions

Name Grzegorz M Swirszcz
Country USA
Email swirszcz@us.ibm.com
Co-Author(s) Aurellie Lozano, Tomasz Nowicki
Submit Time 2014-02-14 16:25:20
Session
Special Session 7: Topological and combinatorial dynamics
Contents
The study of complexity of algorithms can be very difficult and pose significant scientific challenge. We present a study of complexity of the Orthogonal Matching Pursuit algorithm under Restricted Isometry Property conditions using the estimates of convergence rate of iterations of a family of functions. Restricted Isometry Property was introduced by Terrence Tao et. al. and is considered a very important topic in compressed sensing and Machine Learning. Orthogonal Matching Pursuit is a classical feature-selection algorithm in Machine Learning.