Denoising Gravitational Waves with Enhanced Deep Recurrent Denoising Auto-encoders

Hongyu Shen, Daniel George, Eliu A. Huerta, Zhizhen Zhao

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Denoising of time domain data is a crucial task for many applications such as communication, translation, virtual assistants etc. For this task, a combination of a recurrent neural net (RNNs) with a Denoising Auto-Encoder (DAEs) has shown promising results. However, this combined model is challenged when operating with low signal-to-noise ratio (SNR) data embedded in non-Gaussian and non-stationary noise. To address this issue, we design a novel model, referred to as 'Enhanced Deep Recurrent Denoising Auto-Encoder' (EDR-DAE), that incorporates a signal amplifier layer, and applies curriculum learning by first denoising high SNR signals, before gradually decreasing the SNR until the signals become noise dominated. We showcase the performance of EDR-DAE using time-series data that describes gravitational waves embedded in very noisy backgrounds. In addition, we show that EDRDAE can accurately denoise signals whose topology is significantly more complex than those used for training, demonstrating that our model generalizes to new classes of gravitational waves that are beyond the scope of established denoising algorithms.

Original languageEnglish (US)
Title of host publication2019 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019 - Proceedings
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages3237-3241
Number of pages5
ISBN (Electronic)9781479981311
DOIs
StatePublished - May 2019
Event44th IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019 - Brighton, United Kingdom
Duration: May 12 2019May 17 2019

Publication series

NameICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings
Volume2019-May
ISSN (Print)1520-6149

Conference

Conference44th IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019
CountryUnited Kingdom
CityBrighton
Period5/12/195/17/19

Fingerprint

Gravity waves
Signal to noise ratio
Curricula
Time series
Topology
Neural networks
Communication

Keywords

  • RNN
  • denoising algorithm
  • gravitational waves
  • time series

ASJC Scopus subject areas

  • Software
  • Signal Processing
  • Electrical and Electronic Engineering

Cite this

Shen, H., George, D., Huerta, E. A., & Zhao, Z. (2019). Denoising Gravitational Waves with Enhanced Deep Recurrent Denoising Auto-encoders. In 2019 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019 - Proceedings (pp. 3237-3241). [8683061] (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings; Vol. 2019-May). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ICASSP.2019.8683061

Denoising Gravitational Waves with Enhanced Deep Recurrent Denoising Auto-encoders. / Shen, Hongyu; George, Daniel; Huerta, Eliu A.; Zhao, Zhizhen.

2019 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019 - Proceedings. Institute of Electrical and Electronics Engineers Inc., 2019. p. 3237-3241 8683061 (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings; Vol. 2019-May).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Shen, H, George, D, Huerta, EA & Zhao, Z 2019, Denoising Gravitational Waves with Enhanced Deep Recurrent Denoising Auto-encoders. in 2019 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019 - Proceedings., 8683061, ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings, vol. 2019-May, Institute of Electrical and Electronics Engineers Inc., pp. 3237-3241, 44th IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019, Brighton, United Kingdom, 5/12/19. https://doi.org/10.1109/ICASSP.2019.8683061
Shen H, George D, Huerta EA, Zhao Z. Denoising Gravitational Waves with Enhanced Deep Recurrent Denoising Auto-encoders. In 2019 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019 - Proceedings. Institute of Electrical and Electronics Engineers Inc. 2019. p. 3237-3241. 8683061. (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings). https://doi.org/10.1109/ICASSP.2019.8683061
Shen, Hongyu ; George, Daniel ; Huerta, Eliu A. ; Zhao, Zhizhen. / Denoising Gravitational Waves with Enhanced Deep Recurrent Denoising Auto-encoders. 2019 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019 - Proceedings. Institute of Electrical and Electronics Engineers Inc., 2019. pp. 3237-3241 (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings).
@inproceedings{be5d62b2f04f45378b2f3c4d34841455,
title = "Denoising Gravitational Waves with Enhanced Deep Recurrent Denoising Auto-encoders",
abstract = "Denoising of time domain data is a crucial task for many applications such as communication, translation, virtual assistants etc. For this task, a combination of a recurrent neural net (RNNs) with a Denoising Auto-Encoder (DAEs) has shown promising results. However, this combined model is challenged when operating with low signal-to-noise ratio (SNR) data embedded in non-Gaussian and non-stationary noise. To address this issue, we design a novel model, referred to as 'Enhanced Deep Recurrent Denoising Auto-Encoder' (EDR-DAE), that incorporates a signal amplifier layer, and applies curriculum learning by first denoising high SNR signals, before gradually decreasing the SNR until the signals become noise dominated. We showcase the performance of EDR-DAE using time-series data that describes gravitational waves embedded in very noisy backgrounds. In addition, we show that EDRDAE can accurately denoise signals whose topology is significantly more complex than those used for training, demonstrating that our model generalizes to new classes of gravitational waves that are beyond the scope of established denoising algorithms.",
keywords = "RNN, denoising algorithm, gravitational waves, time series",
author = "Hongyu Shen and Daniel George and Huerta, {Eliu A.} and Zhizhen Zhao",
year = "2019",
month = "5",
doi = "10.1109/ICASSP.2019.8683061",
language = "English (US)",
series = "ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
pages = "3237--3241",
booktitle = "2019 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019 - Proceedings",
address = "United States",

}

TY - GEN

T1 - Denoising Gravitational Waves with Enhanced Deep Recurrent Denoising Auto-encoders

AU - Shen, Hongyu

AU - George, Daniel

AU - Huerta, Eliu A.

AU - Zhao, Zhizhen

PY - 2019/5

Y1 - 2019/5

N2 - Denoising of time domain data is a crucial task for many applications such as communication, translation, virtual assistants etc. For this task, a combination of a recurrent neural net (RNNs) with a Denoising Auto-Encoder (DAEs) has shown promising results. However, this combined model is challenged when operating with low signal-to-noise ratio (SNR) data embedded in non-Gaussian and non-stationary noise. To address this issue, we design a novel model, referred to as 'Enhanced Deep Recurrent Denoising Auto-Encoder' (EDR-DAE), that incorporates a signal amplifier layer, and applies curriculum learning by first denoising high SNR signals, before gradually decreasing the SNR until the signals become noise dominated. We showcase the performance of EDR-DAE using time-series data that describes gravitational waves embedded in very noisy backgrounds. In addition, we show that EDRDAE can accurately denoise signals whose topology is significantly more complex than those used for training, demonstrating that our model generalizes to new classes of gravitational waves that are beyond the scope of established denoising algorithms.

AB - Denoising of time domain data is a crucial task for many applications such as communication, translation, virtual assistants etc. For this task, a combination of a recurrent neural net (RNNs) with a Denoising Auto-Encoder (DAEs) has shown promising results. However, this combined model is challenged when operating with low signal-to-noise ratio (SNR) data embedded in non-Gaussian and non-stationary noise. To address this issue, we design a novel model, referred to as 'Enhanced Deep Recurrent Denoising Auto-Encoder' (EDR-DAE), that incorporates a signal amplifier layer, and applies curriculum learning by first denoising high SNR signals, before gradually decreasing the SNR until the signals become noise dominated. We showcase the performance of EDR-DAE using time-series data that describes gravitational waves embedded in very noisy backgrounds. In addition, we show that EDRDAE can accurately denoise signals whose topology is significantly more complex than those used for training, demonstrating that our model generalizes to new classes of gravitational waves that are beyond the scope of established denoising algorithms.

KW - RNN

KW - denoising algorithm

KW - gravitational waves

KW - time series

UR - http://www.scopus.com/inward/record.url?scp=85068996607&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85068996607&partnerID=8YFLogxK

U2 - 10.1109/ICASSP.2019.8683061

DO - 10.1109/ICASSP.2019.8683061

M3 - Conference contribution

AN - SCOPUS:85068996607

T3 - ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings

SP - 3237

EP - 3241

BT - 2019 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019 - Proceedings

PB - Institute of Electrical and Electronics Engineers Inc.

ER -