Neural network approach for classification using features extracted by a mapping

Abstract A neural network approach for classification using features extracted by a mapping is presented. When the number of sample dimensions is much larger than the number of classes and no deviations are given but the means of classes, a mapping from class space to a new one whose dimensions is exactly equal to the number of classes is proposed. The vectors in the new space are considered as the feature vectors to be inputted to a neural network for classification. The property that the mapping does not change the separabilility of the original classification problem is given. Simulation results for object recognition are presented.