Statistical Engineering Division SeminarNonlinear Exploratory Latent Structure Analysis
Haonan Wang Abstract In this talk we discuss the use of a recent dimension reduction technique called Locally Linear Embedding, introduced by Roweis and Saul, for performing an exploratory latent structure analysis. The coordinate variables from the locally linear embedding, describing the manifold on which the data reside, serve as the latent variable scores.  We propose the use of semi-parametric penalized spline methods for reconstruction of the manifold equations that approximate the data space. We also discuss a cross-validation strategy that can guide in selecting an appropriate number of latent variables. Synthetic as well as real data sets are used to illustrate the proposed approach. A nonlinear latent structure representation of a dataset also serves as a data visualization tool. This is joint work with Hari Iyer. NIST Contact: Antonio Possolo, (301) 975-2853.
Date created: 9/24/2007 |