NNETFIX: A neural network to 'fix' Gravitational Wave signals overlapping with short-duration glitches in LIGO-Virgo data.

With improved Advanced LIGO and Virgo detector sensitivity, it is increasingly likely that astrophysical Gravitational Wave (GW) signals overlap with short-duration noise transients. This may affect the initial parameter estimation and sky localization of the glitch-affected GW triggers, resulting in inaccurate low-latency alerts to observatories for Electromagnetic follow-up. Here we introduce NNETFIX: A multi-layered perceptron-based neural network algorithm that 'fixes' GW signals coincident with short-duration glitches. NNETFIX operates by gating the glitch and identifying the features of the GW signal to reconstruct it in the portion of the data affected by the glitch, improving upon the signal-to-noise ratio and recovering the signal parameters effectively. We present an application of NNETFIX on real Advanced LIGO data and propose plans to incorporate it into the low-latency LIGO and Virgo analysis framework.