Fast Highdimensional Approximation with Sparse Occupancy Trees (Preprint)
Abstract
This paper is concerned with scattered data approximation in high dimensions. In this paper we presented algorithms for high dimensional approximation based on sparse occupancy trees which are well suited for large data sets and online learning. It was demonstrated that they offer a viable alternative to the k-nearest neighbor approximation. In particular the vertex algorithm seems to have further potential, since in some application it already outperforms classical approximation methods and there are several direction in which one can search for improvements, notably with regard to the construction of globally continuous approximants.
Document Details
- Document Type
- Technical Report
- Publication Date
- Feb 04, 2010
- Accession Number
- ADA640828
Entities
People
- Peter Binev
- Philipp Lamby
- Wolfgang Dahmen
Organizations
- University of South Carolina