AI and the World's Hard Problems · Physics · arXiv · 2026-10-26
Gamma Neutron Radioactive Source Identification in Water Cherenkov Detectors
The study demonstrates gamma-neutron discrimination in water Cherenkov detectors by combining statistical thresholding with machine learning. A 3σ-based threshold calibrates the charge spectrum to deposited energy, and a soft-voting ensemble (Bagging, CatBoost and an MLP) trained on pulse-shape data reached an accuracy of 0.816. Statistical thresholding gives a baseline across the full energy range, while machine learning improves classification at higher energies.