Denoising gravitational-wave signals from binary black holes with dilated convolutional autoencoder

TL;DR

This paper presents a convolutional autoencoder neural network designed to denoise gravitational-wave signals from binary black hole mergers, improving the extraction of astrophysical information from noisy detector data.

Contribution

It introduces a novel deep learning denoising method trained on simulated and real data, enhancing gravitational-wave signal recovery in non-stationary noise environments.

Findings

Effective removal of noise artifacts from real gravitational-wave data

Improved signal clarity in LIGO O1 and O2 events

Potential for real-time gravitational-wave detection

Abstract

Broadband frequency output of gravitational-wave detectors is a non-stationary and non-Gaussian time series data stream dominated by noise populated by local disturbances and transient artifacts, which evolve on the same timescale as the gravitational-wave signals and may corrupt the astrophysical information. We study a denoising algorithm dedicated to expose the astrophysical signals by employing a convolutional neural network in the encoder-decoder configuration, i.e. apply the denoising procedure of coalescing binary black hole signals in the publicly available LIGO O1 time series strain data. The denoising convolutional autoencoder neural network is trained on a dataset of simulated astrophysical signals injected into the real detector's noise and a dataset of detector noise artifacts ("glitches"), and its fidelity is tested on real gravitational-wave events from O1 and O2 LIGO-Virgo observing runs.