146

Statistical-Computational Trade-offs for Density Estimation

Neural Information Processing Systems (NeurIPS), 2024
Main:8 Pages
2 Figures
Bibliography:2 Pages
2 Tables
Appendix:8 Pages
Abstract

We study the density estimation problem defined as follows: given kk distributions p1,,pkp_1, \ldots, p_k over a discrete domain [n][n], as well as a collection of samples chosen from a ``query'' distribution qq over [n][n], output pip_i that is ``close'' to qq. Recently~\cite{aamand2023data} gave the first and only known result that achieves sublinear bounds in {\em both} the sampling complexity and the query time while preserving polynomial data structure space. However, their improvement over linear samples and time is only by subpolynomial factors. Our main result is a lower bound showing that, for a broad class of data structures, their bounds cannot be significantly improved. In particular, if an algorithm uses O(n/logck)O(n/\log^c k) samples for some constant c>0c>0 and polynomial space, then the query time of the data structure must be at least k1O(1)/loglogkk^{1-O(1)/\log \log k}, i.e., close to linear in the number of distributions kk. This is a novel \emph{statistical-computational} trade-off for density estimation, demonstrating that any data structure must use close to a linear number of samples or take close to linear query time. The lower bound holds even in the realizable case where q=piq=p_i for some ii, and when the distributions are flat (specifically, all distributions are uniform over half of the domain [n][n]). We also give a simple data structure for our lower bound instance with asymptotically matching upper bounds. Experiments show that the data structure is quite efficient in practice.

View on arXiv
Comments on this paper