2000 character limit reached
On learning parametric distributions from quantized samples
Published 25 May 2021 in cs.IT, cs.LG, math.IT, math.ST, and stat.TH | (2105.12019v2)
Abstract: We consider the problem of learning parametric distributions from their quantized samples in a network. Specifically, agents or sensors observe independent samples of an unknown parametric distribution; and each of them uses bits to describe its observed sample to a central processor whose goal is to estimate the unknown distribution. First, we establish a generalization of the well-known van Trees inequality to general -norms, with , in terms of Generalized Fisher information. Then, we develop minimax lower bounds on the estimation error for two losses: general -norms and the related Wasserstein loss from optimal transport.
Paper Prompts
Sign up for free to create and run prompts on this paper.