English

Optimal quantisation of probability measures using maximum mean discrepancy

Machine Learning 2021-02-15 v4 Machine Learning Computation Methodology

Abstract

Several researchers have proposed minimisation of maximum mean discrepancy (MMD) as a method to quantise probability measures, i.e., to approximate a target distribution by a representative point set. We consider sequential algorithms that greedily minimise MMD over a discrete candidate set. We propose a novel non-myopic algorithm and, in order to both improve statistical efficiency and reduce computational cost, we investigate a variant that applies this technique to a mini-batch of the candidate set at each iteration. When the candidate points are sampled from the target, the consistency of these new algorithm - and their mini-batch variants - is established. We demonstrate the algorithms on a range of important computational problems, including optimisation of nodes in Bayesian cubature and the thinning of Markov chain output.

Keywords

Cite

@article{arxiv.2010.07064,
  title  = {Optimal quantisation of probability measures using maximum mean discrepancy},
  author = {Onur Teymur and Jackson Gorham and Marina Riabiz and Chris. J. Oates},
  journal= {arXiv preprint arXiv:2010.07064},
  year   = {2021}
}
R2 v1 2026-06-23T19:20:36.564Z