Automatic configuration of spectral dimensionality reduction methods

Featured Image

Why is it important?

This paper deals with the issue of automatic configuration of the various parameters of spectral dimensionality reduction methods (ISOMAP, Locally Linear Embedding, Laplacian Eigenmaps). The proposed framework performs better than previous solutions for all different dimensionality reduction methods. Such an approach clearly facilitates the usage of this family of spectral methods for all possible datasets.