Local Topological Linear Discriminant Analysis
Hongwei Zhao · Journal of Information and Computational Science · 2013
Subspace learning has been widely applied to face recognition, data clustering and pattern analysis. It is particularly important to supervised learning methods. To deal with the problem of lacking local features in many supervised dimensionality reduction methods, we propose a new supervised dimensionality reduction method called Local Topological Linear Discriminant Analysis (LTLDA) We apply the local topological structure of within-class to the original LDA method. The experimental results show that our method is more efficient to LDA and Maximum Margin Criterion.