English

Memory Efficient And Minimax Distribution Estimation Under Wasserstein Distance Using Bayesian Histograms

Statistics Theory 2023-07-20 v1 Computation Machine Learning Statistics Theory

Abstract

We study Bayesian histograms for distribution estimation on [0,1]d[0,1]^d under the Wasserstein Wv,1v<W_v, 1 \leq v < \infty distance in the i.i.d sampling regime. We newly show that when d<2vd < 2v, histograms possess a special \textit{memory efficiency} property, whereby in reference to the sample size nn, order nd/2vn^{d/2v} bins are needed to obtain minimax rate optimality. This result holds for the posterior mean histogram and with respect to posterior contraction: under the class of Borel probability measures and some classes of smooth densities. The attained memory footprint overcomes existing minimax optimal procedures by a polynomial factor in nn; for example an n1d/2vn^{1 - d/2v} factor reduction in the footprint when compared to the empirical measure, a minimax estimator in the Borel probability measure class. Additionally constructing both the posterior mean histogram and the posterior itself can be done super--linearly in nn. Due to the popularity of the W1,W2W_1,W_2 metrics and the coverage provided by the d<2vd < 2v case, our results are of most practical interest in the (d=1,v=1,2),(d=2,v=2),(d=3,v=2)(d=1,v =1,2), (d=2,v=2), (d=3,v=2) settings and we provide simulations demonstrating the theory in several of these instances.

Keywords

Cite

@article{arxiv.2307.10099,
  title  = {Memory Efficient And Minimax Distribution Estimation Under Wasserstein Distance Using Bayesian Histograms},
  author = {Peter Matthew Jacobs and Lekha Patel and Anirban Bhattacharya and Debdeep Pati},
  journal= {arXiv preprint arXiv:2307.10099},
  year   = {2023}
}
R2 v1 2026-06-28T11:34:50.201Z