Abstract
One of the fundamental problems in pattern recognition is the curse of dimensionality in data representation. Many algorithms have been proposed to find a compact representation of data as well as to facilitate the recognition task. In this paper, we propose a novel Dimensionality Reduction technique called Marginality Preserving Embedding (MPE). Unlike Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA) which projects data in a global sense, MPE seeks for local structure in the manifold. This is similar to other subspace learning techniques but the difference with them is that MPE preserves marginality in local reconstruction. Experimental results show that the proposed method provides better representation in low dimensional space and achieves lower error rates in face recognition.