Optimal Distributed Composite Testing in High-Dimensional Gaussian Models With 1-Bit Communication

Botond Szabo; Lasse Vuursteen; Harry Van Zanten

doi:10.1109/TIT.2022.3150599

Optimal Distributed Composite Testing in High-Dimensional Gaussian Models With 1-Bit Communication

Botond Szabo, Lasse Vuursteen^*, Harry Van Zanten

^*Corresponding author for this work

Mathematics

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

Abstract

In this paper we study the problem of signal detection in Gaussian noise in a distributed setting where the local machines in the star topology can communicate a single bit of information. We derive a lower bound on the Euclidian norm that the signal needs to have in order to be detectable. Moreover, we exhibit optimal distributed testing strategies that attain the lower bound.

Original language	English
Pages (from-to)	4070-4084
Number of pages	15
Journal	IEEE Transactions on Information Theory
Volume	68
Issue number	6
Early online date	9 Feb 2022
DOIs	https://doi.org/10.1109/TIT.2022.3150599
Publication status	Published - Jun 2022

Bibliographical note

Publisher Copyright:
© 1963-2012 IEEE.

Keywords

distributed algorithms
federated learning
Gaussian noise
hypothesis testing
minimax lower bounds
Testing

Access to Document

10.1109/TIT.2022.3150599

Persistent URL (handle)

Persistent link

Cite this

@article{c04386f7cf9e499db18933bb56b46c0c,

title = "Optimal Distributed Composite Testing in High-Dimensional Gaussian Models With 1-Bit Communication",

abstract = "In this paper we study the problem of signal detection in Gaussian noise in a distributed setting where the local machines in the star topology can communicate a single bit of information. We derive a lower bound on the Euclidian norm that the signal needs to have in order to be detectable. Moreover, we exhibit optimal distributed testing strategies that attain the lower bound. ",

keywords = "distributed algorithms, federated learning, Gaussian noise, hypothesis testing, minimax lower bounds, Testing",

author = "Botond Szabo and Lasse Vuursteen and {Van Zanten}, Harry",

note = "Publisher Copyright: {\textcopyright} 1963-2012 IEEE.",

year = "2022",

month = jun,

doi = "10.1109/TIT.2022.3150599",

language = "English",

volume = "68",

pages = "4070--4084",

journal = "IEEE Transactions on Information Theory",

issn = "0018-9448",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

number = "6",

}

TY - JOUR

T1 - Optimal Distributed Composite Testing in High-Dimensional Gaussian Models With 1-Bit Communication

AU - Szabo, Botond

AU - Vuursteen, Lasse

AU - Van Zanten, Harry

PY - 2022/6

Y1 - 2022/6

N2 - In this paper we study the problem of signal detection in Gaussian noise in a distributed setting where the local machines in the star topology can communicate a single bit of information. We derive a lower bound on the Euclidian norm that the signal needs to have in order to be detectable. Moreover, we exhibit optimal distributed testing strategies that attain the lower bound.

AB - In this paper we study the problem of signal detection in Gaussian noise in a distributed setting where the local machines in the star topology can communicate a single bit of information. We derive a lower bound on the Euclidian norm that the signal needs to have in order to be detectable. Moreover, we exhibit optimal distributed testing strategies that attain the lower bound.

KW - distributed algorithms

KW - federated learning

KW - Gaussian noise

KW - hypothesis testing

KW - minimax lower bounds

KW - Testing

UR - http://www.scopus.com/inward/record.url?scp=85124757390&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85124757390&partnerID=8YFLogxK

U2 - 10.1109/TIT.2022.3150599

DO - 10.1109/TIT.2022.3150599

M3 - Article

AN - SCOPUS:85124757390

SN - 0018-9448

VL - 68

SP - 4070

EP - 4084

JO - IEEE Transactions on Information Theory

JF - IEEE Transactions on Information Theory

IS - 6

ER -

Optimal Distributed Composite Testing in High-Dimensional Gaussian Models With 1-Bit Communication

Abstract

Bibliographical note

Keywords

Access to Document

Persistent URL (handle)

Other files and links

Fingerprint

Cite this