University of Otago logo. Computer and Information Science Seminars

Seminar Homepage


Hanhe Lin, Department of Information Science


Multi-instance incremental and decremental one-class support vector machine and its applications


Owheo G34 - 1:00 pm, Friday 25 September


In recent years, One-Class Support Vector Machine (OCSVM) has achieved promising performance in anomaly detection, clustering, etc. However, the conventional OCSVM model is limited to batch mode and cannot be directly applied to video stream due to the challenges such as large volumes, stream fashion, and concept drift. In this talk, I will introduce our recent work, referred to as Multi-instance Incremental and Decremental One-Class Support Vector Machine (MID-OCSVM). By updating the OCSVM model incrementally and decrementally, MID-OCSVM addresses aforementioned limitations while keeping the advantages of OCSVM. Applications on abnormal event detection in video surveillance as well as shot boundary detection validate the advantages of MID-OCSVM.

Last modified: Wednesday, 23-Sep-2015 08:15:08 NZST

This page is maintained by the seminar list administrator.