Applications of Support Vector Regression to Sensor Network Localisation and Network Intrusion Detection
- Investigators
- Staff:
Marimuthu Palaniswami;
Post Doctoral Research Fellow: Alistair Shilton, Daniel T. H. Lai. - Student: Bharat Sundaram, Sophia Kaplantzis.
- Collaborations
- Description
- Introduction: The primary focus of our research is the extension of support vector machine methods for non-real target regression and directly multi-class classification.
- Significance: Our recent work has included the extension of the standard SVR model to division algebraic targets. Applications of this work include equalisation in non-linear communications channels and geometric regression problems in 2 and 3 dimensions.
- Applications: The applications include support vector regression and density estimation to sensor network localisation, anomaly detection and network intrusion detection.
- Challenges: In addition to this we have been studying other extensions of SVM methods including iterative fuzzy approaches that allow one to implement arbitrary empirical risk functions.
- Publication
