Visual nonlinear discriminant analysis for classifier design

We present a new method for analyzing classifiers by visualization, which we call visual nonlinear discriminant analysis. Classifiers that output posterior probabilities are visualized by embedding samples and classes so as to approximate posterior probabilities using parametric embedding. The visualization provides a better intuitive understanding of such classifier characteristics as separability and generalization ability than conventional methods. We evaluate our method by visualizing classifiers for an artificial data set.