Novelty detection in passive SONAR systems using support vector machines

In naval warfare operations, several techniques have been developed for passive sonar signal detection and classification. Sonar systems operate over very noisy conditions and, eventually have to be identified new classes without loosing much efficiency to classes for which both, the sonar operator (OS) and a given decision support system have been trained on. Single-Class support vector machines (SVMs) are supervised learning models with associated kernel algorithms that analyse data and recognize patterns in high order dimensions. This paper proposes the use of Single-Class SVM to obtain a Novelty Detector which encapsulate passive sonar system data in underwater environments.