Generating transient noise artefacts in gravitational-wave detector data with generative adversarial networks

Jade Powell; Ling Sun; Katinka Gereb; Paul D. Lasky; Markus Dollmann

doi:10.1088/1361-6382/acb038

Generating transient noise artefacts in gravitational-wave detector data with generative adversarial networks

Jade Powell, Ling Sun, Katinka Gereb, Paul D. Lasky, Markus Dollmann

School of Physics and Astronomy

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

13 Citations (Scopus)

Abstract

Transient noise glitches in gravitational-wave detector data limit the sensitivity of searches and contaminate detected signals. In this paper, we show how glitches can be simulated using generative adversarial networks (GANs). We produce hundreds of synthetic images for the 22 most common types of glitches seen in the LIGO, KAGRA, and Virgo detectors. We show how our GAN-generated images can easily be converted to time series, which would allow us to use GAN-generated glitches in simulations and mock data challenges to improve the robustness of gravitational-wave searches and parameter-estimation algorithms. We perform a neural network classification to show that our artificial glitches are an excellent match for real glitches, with an average classification accuracy across all 22 glitch types of 99.0%.

Original language	English
Article number	035006
Number of pages	14
Journal	Classical and Quantum Gravity
Volume	40
Issue number	3
DOIs	https://doi.org/10.1088/1361-6382/acb038
Publication status	Published - 2 Feb 2023

Keywords

detector characterisation
gravitational waves
machine learning

Access to Document

10.1088/1361-6382/acb038Licence: CC BY

Cite this

@article{1494b384cf7b4304a0d29ab75a096dbc,

title = "Generating transient noise artefacts in gravitational-wave detector data with generative adversarial networks",

abstract = "Transient noise glitches in gravitational-wave detector data limit the sensitivity of searches and contaminate detected signals. In this paper, we show how glitches can be simulated using generative adversarial networks (GANs). We produce hundreds of synthetic images for the 22 most common types of glitches seen in the LIGO, KAGRA, and Virgo detectors. We show how our GAN-generated images can easily be converted to time series, which would allow us to use GAN-generated glitches in simulations and mock data challenges to improve the robustness of gravitational-wave searches and parameter-estimation algorithms. We perform a neural network classification to show that our artificial glitches are an excellent match for real glitches, with an average classification accuracy across all 22 glitch types of 99.0%.",

keywords = "detector characterisation, gravitational waves, machine learning",

author = "Jade Powell and Ling Sun and Katinka Gereb and Lasky, {Paul D.} and Markus Dollmann",

note = "Funding Information: The authors are supported by the Australian Research Council{\textquoteright}s (ARC) Centre of Excellence for Gravitational Wave Discovery (OzGrav) through project number CE170100004 and through OzGrav Translational Seed Funding. The authors also acknowledge support through ARC LIEF Grant LE210100107. J P is supported by Discovery Early Career Research Award (DECRA) Project Number DE210101050. P D L is supported by ARC Discovery Project DP220101610. This work was performed on the OzSTAR supercomputer which is funded by Swinburne University of Technology and the National Collaborative Research Infrastructure Strategy (NCRIS). We thank the Gravity Spy team for making their data publicly available. This material is based upon work supported by NSF{\textquoteright}s LIGO Laboratory which is a major facility fully funded by the National Science Foundation. Publisher Copyright: {\textcopyright} 2023 IOP Publishing Ltd.",

year = "2023",

month = feb,

day = "2",

doi = "10.1088/1361-6382/acb038",

language = "English",

volume = "40",

journal = "Classical and Quantum Gravity",

issn = "0264-9381",

publisher = "IOP Publishing",

number = "3",

}

TY - JOUR

T1 - Generating transient noise artefacts in gravitational-wave detector data with generative adversarial networks

AU - Powell, Jade

AU - Sun, Ling

AU - Gereb, Katinka

AU - Lasky, Paul D.

AU - Dollmann, Markus

N1 - Funding Information: The authors are supported by the Australian Research Council’s (ARC) Centre of Excellence for Gravitational Wave Discovery (OzGrav) through project number CE170100004 and through OzGrav Translational Seed Funding. The authors also acknowledge support through ARC LIEF Grant LE210100107. J P is supported by Discovery Early Career Research Award (DECRA) Project Number DE210101050. P D L is supported by ARC Discovery Project DP220101610. This work was performed on the OzSTAR supercomputer which is funded by Swinburne University of Technology and the National Collaborative Research Infrastructure Strategy (NCRIS). We thank the Gravity Spy team for making their data publicly available. This material is based upon work supported by NSF’s LIGO Laboratory which is a major facility fully funded by the National Science Foundation. Publisher Copyright: © 2023 IOP Publishing Ltd.

PY - 2023/2/2

Y1 - 2023/2/2

N2 - Transient noise glitches in gravitational-wave detector data limit the sensitivity of searches and contaminate detected signals. In this paper, we show how glitches can be simulated using generative adversarial networks (GANs). We produce hundreds of synthetic images for the 22 most common types of glitches seen in the LIGO, KAGRA, and Virgo detectors. We show how our GAN-generated images can easily be converted to time series, which would allow us to use GAN-generated glitches in simulations and mock data challenges to improve the robustness of gravitational-wave searches and parameter-estimation algorithms. We perform a neural network classification to show that our artificial glitches are an excellent match for real glitches, with an average classification accuracy across all 22 glitch types of 99.0%.

AB - Transient noise glitches in gravitational-wave detector data limit the sensitivity of searches and contaminate detected signals. In this paper, we show how glitches can be simulated using generative adversarial networks (GANs). We produce hundreds of synthetic images for the 22 most common types of glitches seen in the LIGO, KAGRA, and Virgo detectors. We show how our GAN-generated images can easily be converted to time series, which would allow us to use GAN-generated glitches in simulations and mock data challenges to improve the robustness of gravitational-wave searches and parameter-estimation algorithms. We perform a neural network classification to show that our artificial glitches are an excellent match for real glitches, with an average classification accuracy across all 22 glitch types of 99.0%.

KW - detector characterisation

KW - gravitational waves

KW - machine learning

UR - http://www.scopus.com/inward/record.url?scp=85146463345&partnerID=8YFLogxK

U2 - 10.1088/1361-6382/acb038

DO - 10.1088/1361-6382/acb038

M3 - Article

AN - SCOPUS:85146463345

SN - 0264-9381

VL - 40

JO - Classical and Quantum Gravity

JF - Classical and Quantum Gravity

IS - 3

M1 - 035006

ER -

Generating transient noise artefacts in gravitational-wave detector data with generative adversarial networks

Abstract

Keywords

Access to Document

Other files and links

A next generation fast radio burst detector for Australia

Precision cosmic expansion in the era of gravitational-wave astronomy

ARC Centre of Excellence for Gravitational Wave Discovery

Cite this

Generating transient noise artefacts in gravitational-wave detector data with generative adversarial networks

Abstract

Keywords

Access to Document

Other files and links

Projects

A next generation fast radio burst detector for Australia

Precision cosmic expansion in the era of gravitational-wave astronomy

ARC Centre of Excellence for Gravitational Wave Discovery

Cite this