APPLIED MATH SEMINAR

When/where: Monday, April 4th , 4:15 PM, AKW 200 (please note the unusual day)

Speaker: Mauro Maggioni, Mathematics and Computer Science, Duke University.

Title: Multiscale geometric methods for the analysis of point clouds

Abstract:
We discuss techniques for the geometric multiscale analysis of intrinsically low-dimensional point clouds. We present a geometric multiscale transform, based on what we call geometric wavelets, that leads to novel dictionary learning methods for data sets, with guarantees depending on geometric properties of the data, and associated fast transforms. We also show how to apply multiscale geometric techniques to model estimation when points are sampled from a measure supported on a union of an unknown number of unknown planes of unknown dimensions.