In *ISIT*, pages 2704–2708, 2019. IEEE.

bibtex

[BHÖ19] Provides general lower bounds for parameter estimation (and, in some cases, nonparametric density estimation) under $\ell_2$ loss for interactive protocols (blackboard model) under communication constraints, and instantiate it to obtain tight bounds for various statistical models such as Gaussian, product Bernoulli, and discrete distribution estimation. At its core, the lower bounds rely on a Cramér–Rao/van Trees-type approach, which leaves the (trace of) the Fisher information matrix as quantity to analyze to get lower bounds.

bibtex

@inproceedings{BHO19, author = {Leighton P. Barnes and Yanjun Han and Ayfer {\"{O}}zg{\"{u}}r}, title = {Fisher Information for Distributed Estimation under a Blackboard Communication Protocol}, booktitle = {{ISIT}}, pages = {2704--2708}, publisher = {{IEEE}}, year = {2019}, bibbase_note = {<div class="well well-small bibbase"><span class="bluecite">[BHÖ19]</span> Provides general lower bounds for parameter estimation (and, in some cases, nonparametric density estimation) under $\ell_2$ loss for interactive protocols (blackboard model) under communication constraints, and instantiate it to obtain tight bounds for various statistical models such as Gaussian, product Bernoulli, and discrete distribution estimation. At its core, the lower bounds rely on a Cramér–Rao/van Trees-type approach, which leaves the (trace of) the Fisher information matrix as quantity to analyze to get lower bounds.</div>} }

Downloads: 0