Adaptive distributed methods under communication constraints

Botond Szabó; Harry van Zanten

doi:10.1214/19-AOS1890

Adaptive distributed methods under communication constraints

Botond Szabó, Harry van Zanten

Mathematics

Research output: Contribution to Journal › Article › Academic › peer-review

Abstract

We study estimation methods under communication constraints in a distributed version of the nonparametric random design regression model. We derive minimax lower bounds and exhibit methods that attain those bounds. Moreover, we show that adaptive estimation is possible in this setting.

Original language	English
Pages (from-to)	2347-2380
Number of pages	34
Journal	Annals of Statistics
Volume	48
Issue number	4
DOIs	https://doi.org/10.1214/19-AOS1890
Publication status	Published - Aug 2020

Keywords

Adaptation
Communication constraints
Distributed computation
Minimax rates
Nonparametric regression

Access to Document

10.1214/19-AOS1890

Cite this

@article{243d69002fb64ddd88e8db379acaf69e,

title = "Adaptive distributed methods under communication constraints",

abstract = "We study estimation methods under communication constraints in a distributed version of the nonparametric random design regression model. We derive minimax lower bounds and exhibit methods that attain those bounds. Moreover, we show that adaptive estimation is possible in this setting.",

keywords = "Adaptation, Communication constraints, Distributed computation, Minimax rates, Nonparametric regression",

author = "Botond Szab{\'o} and {van Zanten}, Harry",

year = "2020",

month = aug,

doi = "10.1214/19-AOS1890",

language = "English",

volume = "48",

pages = "2347--2380",

journal = "Annals of Statistics",

issn = "0090-5364",

publisher = "Institute of Mathematical Statistics",

number = "4",

}

TY - JOUR

T1 - Adaptive distributed methods under communication constraints

AU - Szabó, Botond

AU - van Zanten, Harry

PY - 2020/8

Y1 - 2020/8

N2 - We study estimation methods under communication constraints in a distributed version of the nonparametric random design regression model. We derive minimax lower bounds and exhibit methods that attain those bounds. Moreover, we show that adaptive estimation is possible in this setting.

AB - We study estimation methods under communication constraints in a distributed version of the nonparametric random design regression model. We derive minimax lower bounds and exhibit methods that attain those bounds. Moreover, we show that adaptive estimation is possible in this setting.

KW - Adaptation

KW - Communication constraints

KW - Distributed computation

KW - Minimax rates

KW - Nonparametric regression

UR - http://www.scopus.com/inward/record.url?scp=85090476340&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85090476340&partnerID=8YFLogxK

U2 - 10.1214/19-AOS1890

DO - 10.1214/19-AOS1890

M3 - Article

AN - SCOPUS:85090476340

VL - 48

SP - 2347

EP - 2380

JO - Annals of Statistics

JF - Annals of Statistics

SN - 0090-5364

IS - 4

ER -

Adaptive distributed methods under communication constraints

Abstract

Keywords

Access to Document

Fingerprint

Cite this