Gravity seminar : Vasileios Skliris, Cardiff University — Using machine learning ​to detect unmodelled GW signals fast​. MLy Pipeline and O5.

The search for unmodeled gravitational wave (GW) transients, or “bursts,” presents unique challenges. In a morphology-agnostic approach, any signal that cannot be identified as a glitch could be of astrophysical origin. However, to date, no non-CBC (compact binary coalescence) signals have been detected by coherent WaveBurst (cWB) , and their detection probability remains significantly lower than that of CBCs. The MLy pipeline is a novel machine learning-based search framework designed to identify unmodelled candidates, prioritizing latency over efficiency. First deployed in December 2024, its primary goal is to rapidly identify and localize signals that are well-suited for electromagnetic follow-up. In this talk, we will explore the structure of this search, the current challenges we face, and our plans for O5.