Adversarially Robust Streaming Algorithms via Differential Privacy

Avinatan Hassidim; Haim Kaplan; Yishay Mansour; Yossi Matias; Uri Stemmer

doi:10.1145/3556972

Adversarially Robust Streaming Algorithms via Differential Privacy

Avinatan Hassidim
, Haim Kaplan
, Yishay Mansour
, Yossi Matias
, Uri Stemmer

Department of Computer Science and Artificial Intelligence

Tel Aviv University
Alphabet Inc.

Research output: Contribution to journal › Article › peer-review

14 Scopus citations

Abstract

A streaming algorithm is said to be adversarially robust if its accuracy guarantees are maintained even when the data stream is chosen maliciously, by an adaptive adversary. We establish a connection between adversarial robustness of streaming algorithms and the notion of differential privacy. This connection allows us to design new adversarially robust streaming algorithms that outperform the current state-of-The-Art constructions for many interesting regimes of parameters.

Original language	English
Article number	42
Journal	Journal of the ACM
Volume	69
Issue number	6
DOIs	https://doi.org/10.1145/3556972
State	Published - 24 Nov 2022

Bibliographical note

Publisher Copyright:
© 2022 Copyright held by the owner/author(s).

Funding

Haim Kaplan was supported in part by the Israel Science Foundation (grant 1595/19) and by the Blavatnik family foundation. Yishay Mansour was supported in part by funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation program (grant 882396), by the Israel Science Foundation (grant 993/17), Tel Aviv University Center for AI and Data Science (TAD), and the Yandex Initiative for Machine Learning at Tel Aviv University. Uri Stemmer was supported in part by the Israel Science Foundation (grant 1871/19) and by the Blavatnik family foundation.

Funders	Funder number
Yandex Initiative for Machine Learning	1871/19
Horizon 2020 Framework Programme	882396, 993/17
Blavatnik Family Foundation
European Commission
Israel Science Foundation	1595/19
Tel Aviv University

Keywords

Streaming
differential privacy
robustness

Access to Document

10.1145/3556972

Cite this

@article{bc9c64c5a8d948ec856e0d4eede8eb51,

title = "Adversarially Robust Streaming Algorithms via Differential Privacy",

abstract = "A streaming algorithm is said to be adversarially robust if its accuracy guarantees are maintained even when the data stream is chosen maliciously, by an adaptive adversary. We establish a connection between adversarial robustness of streaming algorithms and the notion of differential privacy. This connection allows us to design new adversarially robust streaming algorithms that outperform the current state-of-The-Art constructions for many interesting regimes of parameters.",

keywords = "Streaming, differential privacy, robustness",

author = "Avinatan Hassidim and Haim Kaplan and Yishay Mansour and Yossi Matias and Uri Stemmer",

note = "Publisher Copyright: {\textcopyright} 2022 Copyright held by the owner/author(s).",

year = "2022",

month = nov,

day = "24",

doi = "10.1145/3556972",

language = "אנגלית",

volume = "69",

journal = "Journal of the ACM",

issn = "0004-5411",

publisher = "Association for Computing Machinery (ACM)",

number = "6",

}

TY - JOUR

T1 - Adversarially Robust Streaming Algorithms via Differential Privacy

AU - Hassidim, Avinatan

AU - Kaplan, Haim

AU - Mansour, Yishay

AU - Matias, Yossi

AU - Stemmer, Uri

PY - 2022/11/24

Y1 - 2022/11/24

N2 - A streaming algorithm is said to be adversarially robust if its accuracy guarantees are maintained even when the data stream is chosen maliciously, by an adaptive adversary. We establish a connection between adversarial robustness of streaming algorithms and the notion of differential privacy. This connection allows us to design new adversarially robust streaming algorithms that outperform the current state-of-The-Art constructions for many interesting regimes of parameters.

AB - A streaming algorithm is said to be adversarially robust if its accuracy guarantees are maintained even when the data stream is chosen maliciously, by an adaptive adversary. We establish a connection between adversarial robustness of streaming algorithms and the notion of differential privacy. This connection allows us to design new adversarially robust streaming algorithms that outperform the current state-of-The-Art constructions for many interesting regimes of parameters.

KW - Streaming

KW - differential privacy

KW - robustness

UR - https://www.scopus.com/pages/publications/85146365747

U2 - 10.1145/3556972

DO - 10.1145/3556972

M3 - ???researchoutput.researchoutputtypes.contributiontojournal.article???

AN - SCOPUS:85146365747

SN - 0004-5411

VL - 69

JO - Journal of the ACM

JF - Journal of the ACM

IS - 6

M1 - 42

ER -

Adversarially Robust Streaming Algorithms via Differential Privacy

Abstract

Bibliographical note

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this