25
19
v1v2 (latest)

Instance Optimal Learning

Abstract

We consider the following basic learning task: given independent draws from an unknown distribution over a discrete support, output an approximation of the distribution that is as accurate as possible in 1\ell_1 distance (i.e. total variation or statistical distance). Perhaps surprisingly, it is often possible to "de-noise" the empirical distribution of the samples to return an approximation of the true distribution that is significantly more accurate than the empirical distribution, without relying on any prior assumptions on the distribution. We present an instance optimal learning algorithm which optimally performs this de-noising for every distribution for which such a de-noising is possible. More formally, given nn independent draws from a distribution pp, our algorithm returns a labelled vector whose expected distance from pp is equal to the minimum possible expected error that could be obtained by any algorithm that knows the true unlabeled vector of probabilities of distribution pp and simply needs to assign labels, up to an additive subconstant term that is independent of pp and goes to zero as nn gets large. One conceptual implication of this result is that for large samples, Bayesian assumptions on the "shape" or bounds on the tail probabilities of a distribution over discrete support are not helpful for the task of learning the distribution. As a consequence of our techniques, we also show that given a set of nn samples from an arbitrary distribution, one can accurately estimate the expected number of distinct elements that will be observed in a sample of any size up to nlognn \log n. This sort of extrapolation is practically relevant, particularly to domains such as genomics where it is important to understand how much more might be discovered given larger sample sizes, and we are optimistic that our approach is practically viable.

View on arXiv
Comments on this paper