Artificial intelligence model for gravitational wave search based on the waveform envelope
JournalPhysical Review D 107(6), 063029 (2023)
First public: Published online:
- Detection & denoising
- AI for science
Research summary
A two-stage gravitational-wave search uses deep learning to extract waveform-envelope information and then checks candidate coalescence times. Wavelet denoising and a dedicated training strategy support detection. Tests on observing-run data examine how the additional validation stage changes false alarms and recovery of confidently identified events relative to matched-filter searches.
Extended notes
The model separates candidate detection from a subsequent consistency test. The waveform envelope supplies timing information used to validate candidate events in LIGO–Virgo–KAGRA observing data.