应用于数据降维的Laplacian 特征映射增量学习算法研究.pdfVIP

Out-of-sample algorithm of Laplacian Eigenmaps Applied to Dimensionality Reduction Jia Peng, Yin Junsong, Huang Xinsheng, Hu Dewen Department of Automatic Control, College of Mechatronics and Automation National University of Defense Technology Hunan, China (410073) E-mail ：dwhu@ Abstract The traditional nonlinear manifold learning methods have achieved great success in dimensionality reduction. However, when new samples are observed, the batch methods fail to learn them incrementally. This paper presents out-of-sample extension for Laplacian Eigenmaps, which computes the low-dimensional representation of data set by optimally preserving local neighborhood information in a certain sense. Two different incremental algorithms, the differential method and sub-manifold analysis method, are proposed. The algorithms are easy to be implemented and the computation procedure is simple. Simulation results testify the efficiency and accuracy of the proposed algorithm. Keywords: Laplacian eigenmaps, Manifold learning, Incremental learning, dimensionality reduction 1 Introduction In pattern recognition and machine learning, dimensionality reduction aims to transform the high-dimensional dataset into a low-dimensional space, while retaining most of the underlying structure in the data. It can be used to solve the curse of dimensionality [1], and to visualize the data set. Besides some linear algorithms, such as