DocumentCode :
77822
Title :
Analysis of Sparse Regularization Based Robust Regression Approaches
Author :
Mitra, Kaushik ; Veeraraghavan, Ashok ; Chellappa, Rama
Author_Institution :
Dept. of Electr. & Comput. Eng., Rice Univ., Houston, TX, USA
Volume :
61
Issue :
5
fYear :
2013
fDate :
1-Mar-13
Firstpage :
1249
Lastpage :
1257
Abstract :
Regression in the presence of outliers is an inherently combinatorial problem. However, compressive sensing theory suggests that certain combinatorial optimization problems can be exactly solved using polynomial-time algorithms. Motivated by this connection, several research groups have proposed polynomial-time algorithms for robust regression. In this paper we specifically address the traditional robust regression problem, where the number of observations is more than the number of unknown regression parameters and the structure of the regressor matrix is defined by the training dataset (and hence it may not satisfy properties such as Restricted Isometry Property or incoherence). We derive the precise conditions under which the sparse regularization (l0 and l1-norm) approaches solve the robust regression problem. We show that the smallest principal angle between the regressor subspace and all k-dimensional outlier subspaces is the fundamental quantity that determines the performance of these algorithms. In terms of this angle we provide an estimate of the number of outliers the sparse regularization based approaches can handle. We then empirically evaluate the sparse (l1-norm) regularization approach against other traditional robust regression algorithms to identify accurate and efficient algorithms for high-dimensional regression problems.
Keywords :
combinatorial mathematics; optimisation; polynomials; regression analysis; signal reconstruction; combinatorial optimization problem; compressive sensing theory; high-dimensional regression problem; k-dimensional outlier subspace; polynomial-time algorithm; restricted isometry property; robust regression approach; sparse regularization analysis; unknown regression parameter; Algorithm design and analysis; Educational institutions; Optimization; Robustness; Signal processing algorithms; Sparse matrices; Training; Compressive sensing; robust regression; sparse representation;
fLanguage :
English
Journal_Title :
Signal Processing, IEEE Transactions on
Publisher :
ieee
ISSN :
1053-587X
Type :
jour
DOI :
10.1109/TSP.2012.2229992
Filename :
6362268
Link To Document :
بازگشت