Application of Non-Linear Noise Regression in the Virgo Detector

TL;DR

This paper demonstrates the successful application of deep learning-based non-linear noise regression using DeepClean to Virgo gravitational wave data, improving detection range and signal recovery without bias.

Contribution

It introduces DeepClean for Virgo data, showing its effectiveness in modeling non-linear noise couplings and enhancing gravitational wave signal detection.

Findings

Up to 1.3 Mpc improvement in binary neutron star inspiral range

Average 1.7% increase in recovered SNR for binary black hole signals

No bias introduced in source parameter estimation

Abstract

This work presents the first demonstration of non-linear noise regression in the Virgo detector using deep learning techniques. We use DeepClean, a convolutional autoencoder previously shown to be effective in denoising LIGO data, as our tool for modeling and subtracting environmental and technical noise in Virgo. The method uses auxiliary witness channels to learn correlated noise features and remove them from the strain data. For this study, we apply DeepClean to Virgo O3b data, using 225 witness channels selected across 13 targeted frequency bands. Our analysis confirms the presence of non-linear couplings in the subtracted noise, highlighting the importance of DeepClean-like tools in capturing such effects. We observe up to a 1.3 Mpc improvement in the binary neutron star inspiral range (~2.5% gain), and an average increase of 1.7% in the recovered signal-to-noise ratio for injected binary black hole signals. Parameter estimation studies further confirm that DeepClean does not introduce bias in the recovery of source parameters. These results demonstrate the robustness of DeepClean on Virgo data and support its adoption in real-time noise subtraction frameworks for future observing runs.