Speaker
Description
The sensitivity of Gravitational Waves (GW) interferometers is limited by noise. Non-Gaussian transient noise artifacts, also known as glitches, are particularly challenging due to their similarity with astrophysical signals in the time and frequency domains; for this reason noise reduction and subtraction is one of the most important and challenging activities in GW research. Within the interTwin EU project we have developed GlitchFlow: a tool for modeling and generating glitches for GW interferometers using deep generative algorithms. GlitchFlow leverages the power of Convolutional Neural Networks (CNNs) to reconstruct the 2D spectrograms of glitches in the observational (strain) channel at Virgo, taking as input the spectrograms of the glitches in the data recorded by a small number of control sensors. Once a glitch has been generated, it is subtracted by the strain channel. The model has a noise-subtraction accuracy higher than 90% for Scattered Light glitches with Signal-To-Noise ratio of 6 or above. We aim to have the tool fully integrated into the Virgo data analysis pipelines by the next observing run.
