Abrupt change detection with One-Class Time-Adaptive Support Vector Machines

• We introduce and analyze a non-stationary extension of the One-Class SVM method.
• We use this new extension as an improved abrupt change detection method.
• We compare our procedure with previous methods, on artificial and real world data.

We recently introduced an algorithm for training a sequence of coupled Support Vector Machines which shows promising results in the field of non-stationary classification problems Grinblat, Uzal, Ceccatto, and Granitto (2011). In this paper we analyze its application to the abrupt change detection problem. With this goal, we first introduce and analyze an extension of it to deal with the One-Class Support Vector Machine (OC-SVM) problem, and then discuss its use as an improved abrupt change detection method. Finally, we apply the proposed procedure to artificial and real-world examples, and demonstrate that it is competitive by comparison against other abrupt change detection methods.