A novel learning classification on pattern recognition

Abstract

In this paper we propose a novel approach to model pattern classification based on tangent distance within a statistical framework for classification. Statistical characteristics of classifier data are analyzed by the parametric estimates in the regression equation. We research on feature extraction for the training data in statistical pattern recognition by transformation and optimal Bayesian decision rule. Finally, we give our experiments on automatic speech recognition. The experimental results show the effectiveness of our approach. © 2012 Springer-Verlag.