Detecting Binary Black Hole Mergers with Effective Machine Learning Infrastructure

Deep learning models have quickly become a popular alternative to traditional matched filtering analyses for the identification of gravitational wave signals. The reduced computational cost and the potential for real-time, higher confidence detections have made these techniques an attractive avenue to explore; however, work remains in developing a network that can be effectively deployed on live data during a data collection run of the LIGO-Virgo-KAGRA detectors. We present here the preliminary results of a model for identifying binary black hole mergers, BBHNet, which has been built using libraries that address this gap, hermes and ml4gw. We demonstrate that our model is capable of event identification, and show that our design choices allow for rapid iteration and effective analysis of model performance.