Survey of Dimension Reduction Techniques

摘要

In this paper. we review traditional and current state-of-the-art dimension reduction methods published in the statistics. signal processing and machine learning literature. There are numerous books and articles in the statistical literature on tcchniques for analyzing multivariate datasets. Earlier survey papers review several methods, including principal components and analysis, projection pursuit, principal curves, self-organizing maps, as well as provides neural network implementations of some of the reviewed statistical methods.