Speaker
传晖 郝
(清华大学)
Description
Liquid-scintillator neutrino detectors made significant contributions in the discoveries of neutrino physics. Reconstruction of neutrino energy and position demands point-source response function (probe) of the photoelectrons reception on photomultiplier tubes. We model the probe with inhomogeneous Poisson process for calibration from Monte Carlo simulated data. To achieve higher precision and ensure continuity and differentiability, we investigate several advanced machine learning and statisical methods and find the Generalized Additive Model demonstrates the best performance among the other studied approaches including neural networks and boosted decision tree.
Author
传晖 郝
(清华大学)
Co-author
Benda Xu
(Tsinghua University)
