Detection of data injection attacks in decentralized learning

Reinhard Gentz; Hoi To Wai; Anna Scaglione; Amir Leshem

doi:10.1109/ACSSC.2015.7421145

Detection of data injection attacks in decentralized learning

Reinhard Gentz, Hoi To Wai, Anna Scaglione, Amir Leshem

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

10 Scopus citations

Abstract

Gossip based optimization and learning are appealing methods that solve big data learning problems sharing computation and network resources when data are distributed. The main advantage these methods offer is that they are fault tolerant. Their flat architecture, however, expands the attack surface in the case of a data injection attack. We analyze the effects of data injection on the asymptotic behavior of the network and draw a parallel with the case of opinion dynamics in a network where zealots inject opinions to mislead a community. We further propose a possible decentralized detection of such attacks and analyze its performance.

Original language	English (US)
Title of host publication	Conference Record - Asilomar Conference on Signals, Systems and Computers
Publisher	IEEE Computer Society
Pages	350-354
Number of pages	5
Volume	2016-February
ISBN (Print)	9781467385763
DOIs	https://doi.org/10.1109/ACSSC.2015.7421145
State	Published - Feb 26 2016
Event	49th Asilomar Conference on Signals, Systems and Computers, ACSSC 2015 - Pacific Grove, United States Duration: Nov 8 2015 → Nov 11 2015

Other

Other	49th Asilomar Conference on Signals, Systems and Computers, ACSSC 2015
Country/Territory	United States
City	Pacific Grove
Period	11/8/15 → 11/11/15

Keywords

attack detection
data injection attack
decentralized learning
randomized gossip protocol

ASJC Scopus subject areas

Computer Networks and Communications
Signal Processing

Access to Document

10.1109/ACSSC.2015.7421145

Cite this

Gentz, R, Wai, HT, Scaglione, A & Leshem, A 2016, Detection of data injection attacks in decentralized learning. in Conference Record - Asilomar Conference on Signals, Systems and Computers. vol. 2016-February, 7421145, IEEE Computer Society, pp. 350-354, 49th Asilomar Conference on Signals, Systems and Computers, ACSSC 2015, Pacific Grove, United States, 11/8/15. https://doi.org/10.1109/ACSSC.2015.7421145

@inproceedings{dfbdd7023f744f26b1f52b9f4d057536,

title = "Detection of data injection attacks in decentralized learning",

abstract = "Gossip based optimization and learning are appealing methods that solve big data learning problems sharing computation and network resources when data are distributed. The main advantage these methods offer is that they are fault tolerant. Their flat architecture, however, expands the attack surface in the case of a data injection attack. We analyze the effects of data injection on the asymptotic behavior of the network and draw a parallel with the case of opinion dynamics in a network where zealots inject opinions to mislead a community. We further propose a possible decentralized detection of such attacks and analyze its performance.",

keywords = "attack detection, data injection attack, decentralized learning, randomized gossip protocol",

author = "Reinhard Gentz and Wai, {Hoi To} and Anna Scaglione and Amir Leshem",

year = "2016",

month = feb,

day = "26",

doi = "10.1109/ACSSC.2015.7421145",

language = "English (US)",

isbn = "9781467385763",

volume = "2016-February",

pages = "350--354",

booktitle = "Conference Record - Asilomar Conference on Signals, Systems and Computers",

publisher = "IEEE Computer Society",

note = "49th Asilomar Conference on Signals, Systems and Computers, ACSSC 2015 ; Conference date: 08-11-2015 Through 11-11-2015",

}

TY - GEN

T1 - Detection of data injection attacks in decentralized learning

AU - Gentz, Reinhard

AU - Wai, Hoi To

AU - Scaglione, Anna

AU - Leshem, Amir

PY - 2016/2/26

Y1 - 2016/2/26

N2 - Gossip based optimization and learning are appealing methods that solve big data learning problems sharing computation and network resources when data are distributed. The main advantage these methods offer is that they are fault tolerant. Their flat architecture, however, expands the attack surface in the case of a data injection attack. We analyze the effects of data injection on the asymptotic behavior of the network and draw a parallel with the case of opinion dynamics in a network where zealots inject opinions to mislead a community. We further propose a possible decentralized detection of such attacks and analyze its performance.

AB - Gossip based optimization and learning are appealing methods that solve big data learning problems sharing computation and network resources when data are distributed. The main advantage these methods offer is that they are fault tolerant. Their flat architecture, however, expands the attack surface in the case of a data injection attack. We analyze the effects of data injection on the asymptotic behavior of the network and draw a parallel with the case of opinion dynamics in a network where zealots inject opinions to mislead a community. We further propose a possible decentralized detection of such attacks and analyze its performance.

KW - attack detection

KW - data injection attack

KW - decentralized learning

KW - randomized gossip protocol

UR - http://www.scopus.com/inward/record.url?scp=84969836578&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84969836578&partnerID=8YFLogxK

U2 - 10.1109/ACSSC.2015.7421145

DO - 10.1109/ACSSC.2015.7421145

M3 - Conference contribution

AN - SCOPUS:84969836578

SN - 9781467385763

VL - 2016-February

SP - 350

EP - 354

BT - Conference Record - Asilomar Conference on Signals, Systems and Computers

PB - IEEE Computer Society

T2 - 49th Asilomar Conference on Signals, Systems and Computers, ACSSC 2015

Y2 - 8 November 2015 through 11 November 2015

ER -

Detection of data injection attacks in decentralized learning

Abstract

Other

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this