PASCAL - Pattern Analysis, Statistical Modelling and Computational Learning

Adaptive One-Class Support Vector Machine
Vanessa Gómez-Verdejo, Jerónimo Arenas-García, Miguel Lázaro-Gredilla and Angel Navia-Vazquez
IEEE Transactions on Signal Processing 2011. ISSN 1053-587X

Abstract

In this work we derive an on-line adaptive one-class Support Vector Machine which is suitable to solve on-line novelty detection problems. The machine structure is updated via growing and pruning mechanisms, and the weights are updated using Structural Risk Minimization principles underlying Support Vector Machines. Our approach leads to very compact machines compared to other state-of-the-art on-line kernel methods whose size, unless truncated, grows almost linearly with the number of observed patterns. The proposed method is on-line in the sense that every pattern is only presented once to the machine and there is no need to store past samples, and adaptive in the sense that it can forget past input patterns and adapt to the new characteristics of the incoming data.We illustrate its performance in a time series segmentation problem and we benchmark it against one state-of-the-art approach, obtaining favorable results in both accuracy and machine complexity.

PDF - PASCAL Members only - Requires Adobe Acrobat Reader or other PDF viewer.
EPrint Type:Article
Project Keyword:Project Keyword UNSPECIFIED
Subjects:Learning/Statistics & Optimisation
ID Code:7525
Deposited By:Angel Navia-Vazquez
Deposited On:17 March 2011