Toward real-time detection of unmodeled gravitational wave transients using convolutional neural networks

Convolutional neural networks (CNNs) have demonstrated potential for the real-time analysis of data from gravitational wave detector networks for the specific case of signals from coalescing compact-object binaries such as black-hole binaries. Unfortunately, CNNs presented to date have required a precise model of the target signal for training. Such CNNs are therefore not applicable to detecting generic gravitational wave transients from unknown sources, and may be unreliable for anticipated sources such as core-collapse supernovae and long gamma-ray bursts, where unknown physics or computational limitations prevent the development of robust, accurate signal models. We demonstrate for the first time a CNN analysis pipeline with the ability to detect generic signals-those without a precise model-with sensitivity across a wide parameter space and with useful significance. Our CNN has a novel structure that uses not only the network strain data but also the Pearson cross-correlation between detectors to distinguish correlated gravitational wave signals from uncorrelated noise transients. We demonstrate the efficacy of our CNN using data from the second LIGO-Virgo observing run. We show that it has sensitivity approaching that of the "goldstandard" unmodeled transient searches currently used by LIGO-Virgo, at extremely low (order of 1 s) latency and using only a fraction of the computing power required by existing searches, allowing our models the possibility of true real-time detection of gravitational wave transients associated with gamma-ray bursts, core-collapse supernovae, and other relativistic astrophysical phenomena.