Competitive Mixture of Deformable

Following the success of applying deformable models to feature extraction, a natural next step is to apply such models to pattern classiication. Recently, we have cast a deformable model under a Bayesian framework for classiication, giving promising results. However, deformable model methods are computation-ally expensive due to the required iterative optimization process. The problem is even more severe when there are a large number of models (e.g., for character recognition), because each of them has to deform and match with the input data before a nal classiication can be derived. In this paper, we propose to combine the deformable models into a mixture, in which the individual models compete with each other to survive the matching process during classiication. Models that do not compete well are eliminated early, thus allowing substantial savings in computation. This process of competition-elimination has been applied to handwritten digit recognition in which signiicant speedup can be achieved without sacriicing recognition accuracy.