Nearest-neighbour ensembles in lasso feature subspaces

• Published in: IET computer vision Vol. 4; no. 4; pp. 306 - 319
• Main Authors: HE, X, BEAUSEROY, P, SMOLARZ, A
• Format: Journal Article
• Language: English
• Published: Stevenage Institution of Engineering and Technology 01.12.2010; John Wiley & Sons, Inc; IET
• Subjects: Algorithms; Applied sciences; Classification; Computer Science; Computer vision; Computer Vision and Pattern Recognition; Exact sciences and technology; Image processing; Image segmentation; Information, signal and communications theory; Miscellaneous; Operators; Preserves; Representations; Signal and communications theory; Signal processing; Signal representation. Spectral analysis; Signal, noise; Subspaces; Telecommunications and information theory; Performance evaluation; Nearest neighbour; Image texture; Automatic classification; Multistage method; Image processing; Subspace method; Algorithm; Signal classification; Selection problem; Image segmentation; Signal processing; Feature extraction
• ISSN: 1751-9632; 1751-9640
• DOI: 10.1049/iet-cvi.2009.0056

The least absolute shrinkage and selection operator (lasso) is a promising feature selection technique. However, it has traditionally not been a focus of research in ensemble classification methods. In this study, the authors propose a robust classification algorithm that makes use of an ensemble of classifiers in lasso feature subspaces. The algorithm consists of two stages: the first is a lasso-based multiple feature subsets selection cycle, which tries to find a number of relevant and diverse feature subspaces; the second is an ensemble-based decision system that intends to preserve the classification performance in case of abrupt changes in the representation space. The experimental results on the two-class textured image segmentation problem prove the effectiveness of the proposed classification method.