An Efficient One-Class SVM for Novelty Detection in IoT
Video
Team Information
Team Members
Samory Kpotufe, Associate Professor, Department of Statistics, Columbia University
Abstract
One-Class Support Vector Machines (OCSVM) have been applied in Internet of Things (IoT) for novelty detection, due to their flexibility in fitting complex nonlinear boundaries between normal and novel data.
Conventional OCSVMs introduce significant memory requirements and are computationally expensive at prediction time as the size of the train set grows. This work extends so-called Nystrom and (Gaussian) Sketching approaches to OCSVM, by combining these methods with clustering and Gaussian mixture models to achieve significant speedups in prediction time and space in IoT settings, without sacrificing detection accuracy.
Contact this Team
Contact: Samory Kpotufe (use form to send email)
