Approximation of Functions Over Manifolds by Moving Least SquaresCP10

We present an algorithm for approximating a function defined over a manifold utilizing only noisy function values at locations sampled from the manifold with noise. To produce the approximation we do not require any knowledge regarding the manifold other than its dimension. The approximation scheme is based upon the Manifold Moving Least-Squares (MMLS). We compare, using numerical experiments, the presented algorithm to state-of-the-art algorithms for regression over manifolds and show its resistans to noise.

This presentation is part of Contributed Presentation “CP10 - Contributed session 10

Yariv Aizenbud (Tel Aviv University)
Barak Sober (Tel-Aviv University)
computer graphics, computer vision, machine learning, manifold learning, regression