Nonlinear discriminant mapping using the Laplacian of a graph

In this paper, an algorithm for nonlinear discriminant mapping (NDM) is presented, which elegantly integrates the ideas of both linear discriminant analysis (LDA) and Isomap by using the Laplacian of a graph. The objective of NDM is to find a linear subspace project of nonlinear data set, which preserves maximum difference between between-class scatter and within-class scatter.