DOI: 10.3724/SP.J.1004.2013.02077

Acta Automatica Sinica (自动化学报) 2013/39:12 PP.2077-2089

A New Supervised Manifold Learning Algorithm Based on MMC and LSE

In order to circumvent the two major shortcomings of the original local spline embedding (LSE) algorithm, i.e., out-of-sample and unsupervised learning, we proposed a novel feature extraction algorithm called orthogonal local spline discriminant projection (O-LSDP). By introducing an explicit linear mapping, constructing different translation and resealing models for different classes as well as orthogonality feature subspace, the O-LSDP not only inherits the advantages of LSE which uses local tangent space as a representation of the local geometry so as to preserve the local structure, but also makes full use of class information and orthogonal subspace to significantly improve the discriminant power. Experimental results on standard face databases and plant leaf data set demonstrate the feasibility and effectiveness of the proposed algorithm.

Key words:Local spline embedding (LSE), maximum margin criterion, feature extraction, manifold learning

ReleaseDate:2014-07-21 17:04:35

Funds:National Natural Science Foundation of China (61272333, 61273302, 61005010), Natural Science Foundation of Anhui Province (1208085MF94, 1208085MF98, 1308085MF84)