Researchers have developed three deep learning models, including a spatiotemporal graph neural network and two Transformer-based models, to identify and suppress photon hits from carbon-14 decays in liquid scintillator detectors. These models aim to improve the energy resolution in neutrino experiments by distinguishing between carbon-14 signals and actual event signals, particularly when they overlap in space and time. The proposed models achieved carbon-14 recall rates between 25% and 48% while maintaining a misidentification rate below 1% for electron positron events. AI
IMPACT Potential to improve data quality and resolution in sensitive scientific experiments by enhancing signal-to-noise ratios.
RANK_REASON Academic paper detailing a new methodology for signal processing in scientific instrumentation. [lever_c_demoted from research: ic=1 ai=1.0]
- arXiv
- carbon-14
- Junle Li
- liquid scintillator detectors
- neutrino experiments
- spatiotemporal graph neural network
- Transformer-based Models
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
