Inverse matrix-free incremental proximal support vector machine

Publisher:
Elsevier
Publication Type:
Journal Article
Citation:
Decision Support Systems, 2012, 53 (3), pp. 395 - 405
Issue Date:
2012-01
Full metadata record
Files in This Item:
Filename Description Size
Thumbnail2011007990OK.pdfPublished Version1.02 MB
Adobe PDF
Traditional Support Vector Machines (SVMs) have been commonly regarded as strong classi?ers for many learning tasks [13,33]. As our data-gathering ability continuously grows, in many applications, large-scale or continuous stream data make the standard SVMs inef?- cient (or even inapplicable) for decision making [35]. To solve these problems, a number of variations are proposed [8,16,18,20, 21,24,27,28]. Among them, the proximal SVM (PSVM) and least squares SVM (LS-SVM) are simple yet effective algorithms [8,27], due to the essence of regularized least squares problems which merely need to solve linear equations [31]. In comparison, LS-SVM needs to calculate about n variables or parameters whereas PSVM has d variables for the corresponding linear equations, where n and d are the numbers of samples and dimensions, respectively
Please use this identifier to cite or link to this item: