Faculty of Engineering and Natural Sciences
Multiple Classifier Fusion
Centre for Vision, Speech and Signal Processing
University of Surrey
Guildford GU2 7XH, UK
The merits of combining multiple classifier systems are now well recognised. The talk will provide an overview of the methodology of multiple classifier fusion. The various fusion paradigms will be briefly reviewed and the underlying theory of multimodal and intramodal fusion presented. The various mechanisms of intramodal fusion will then be discussed, including Bayesian marginalisation techniques and margin manipulation techniques. Some of the resulting fusion strategies will be illustrated on selected applications.
April 21, 2006, 13:40, G032