Paper Info
Title
A regularized correntropy framework for robust pattern recognition
Abstract
This letter proposes a new multiple linear regression model using regularized correntropy for robust pattern recognition. First, we motivate the use of correntropy to improve the robustness of the classical mean square error (MSE) criterion that is sensitive to outliers. Then an l1 regularization scheme is imposed on the correntropy to learn robust and sparse representations. Based on the half-quadratic optimization technique, we propose a novel algorithm to solve the nonlinear optimization problem. Second, we develop a new correntropy-based classifier based on the learned regularization scheme for robust object recognition. Extensive experiments over several applications confirm that the correntropy-based l1 regularization can improve recognition accuracy and receiver operator characteristic curves under noise corruption and occlusion.
Year
DOI
Venue
2011
10.1162/NECO_a_00155
Neural Computation
Keywords
Field
DocType
robust object recognition,recognition accuracy,half-quadratic optimization technique,robust pattern recognition,regularized correntropy,regularized correntropy framework,new correntropy-based classifier,regularization scheme,l1 regularization scheme,l1 regularization,new multiple linear regression,nonlinear optimization,pattern recognition,sparse representation,object recognition,receiver operating characteristic curve,mean square error,multiple linear regression
Facial recognition system,Pattern recognition,Computer science,Sparse approximation,Outlier,Mean squared error,Robustness (computer science),Regularization (mathematics),Artificial intelligence,Linear discriminant analysis,Machine learning,Cognitive neuroscience of visual object recognition
Journal
Volume
Issue
ISSN
23
8
0899-7667
Citations 
PageRank 
References 
55
1.54
45
Authors
4
Name
Order
Citations
PageRank
Ran He11790108.39
Wei-Shi Zheng22915140.63
Hu Bao-Gang3138683.23
Xiang-Wei Kong421215.09