What is it about?
Natural Evolution Strategies are known to work very well in optimization, because they apply to any domain, as compared to other strategies that rely on the Euclidean space. The paper explains how the functional domain of Natural Evolution Strategies can be extended to sampling.
Featured Image
Photo by Jonathan Greenaway on Unsplash
Why is it important?
The paper is important, mainly because of its generalization of the search domain to the Riemannian manifold of distributions.
Perspectives
Whereas Natural Evolution strategies have been keenly researched under optimization, their use in the sampling realm has been underestimated. This paper encourages and hopefully redirects research towards the perspective of sampling.
Nixon Ronoh
Vrije Universiteit Brussel
Read the Original
This page is a summary of: Natural gradient evolution strategies for adaptive sampling, July 2022, ACM (Association for Computing Machinery),
DOI: 10.1145/3520304.3533939.
You can read the full text:
Contributors
The following have contributed to this page
