Gravitational-wave parameter inference using Deep Learning

We explore machine learning methods to detect gravitational waves (GW) from binary black hole (BBH) mergers using deep learning (DL) algorithms. The DL networks are trained with gravitational waveforms obtained from BBH mergers with component masses randomly sampled in the range from 5 to 100 solar masses and luminosity distances from 100Mpc to, at least, 2000Mpc. The GW signal waveforms are injected in public data from the O2 run of the Advanced LIGO and Advanced Virgo detectors, in time windows that do not coincide with those of known detected signals, and the data from each detector in the Advanced LIGO and Advanced Virgo network is combined into a unique RGB image. We show that a classifier network can be trained in order to detect the presence of GW signal with high accuracy. Furthermore, we train a regression network to perform parameter inference on BBH spectrogram data. Without significant optimization of our algorithms we manage to corroborate most of the BBH detections in the GWTC-1 and GWTC-2 catalogs, and obtain parameter inference results that are mostly consistent with published results by the LIGO-Virgo Collaboration in GWTC-1. In particular, our predictions for the chirp mass are compatible (up to 3 sigma) with the official values for 90% of events.