| |
Abstract:
Developments of nonlinear latent variable models based on
radial basis functions are discussed: the first is a re-sampling
approach that makes more effective use of latent samples in
evaluating the likelihood. Also, the use of priors or constraints
on allowable models is considered as a means of preserving data
structure on low-dimensional representations for visualisation
purposes. The former development is illustrated on both a simulated
data set and radar range profiles of ships.
|
