Abstract
The Majorana Demonstrator is a leading experiment searching for neutrinoless double-beta decay with high purity germanium (HPGe) detectors. Machine learning provides a new way to maximize the amount of information provided by these detectors, but the data-driven nature makes it less interpretable compared to traditional analysis. An interpretability study reveals the machine's decision-making logic, allowing us to learn from the machine to feed back to the traditional analysis. In this work, we present the first machine learning analysis of the data from the Majorana Demonstrator; this is also the first interpretable machine learning analysis of any germanium detector experiment. Two gradient boosted decision tree models are trained to learn from the data, and a game-theory-based model interpretability study is conducted to understand the origin of the classification power. By learning from data, this analysis recognizes the correlations among reconstruction parameters to further enhance the background rejection performance. By learning from the machine, this analysis reveals the importance of new background categories to reciprocally benefit the standard Majorana analysis. This model is highly compatible with next-generation germanium detector experiments like LEGEND since it can be simultaneously trained on a large number of detectors.
| Original language | English |
|---|---|
| Article number | 014321 |
| Journal | Physical Review C |
| Volume | 107 |
| Issue number | 1 |
| DOIs | |
| State | Published - Jan 2023 |
Funding
This material is based upon work supported by the U.S. Department of Energy, Office of Science, Office of Nuclear Physics under Contracts/Awards No. DE-AC02-05CH11231, No. DE-AC05-00OR22725, No. DE-AC05-76RL0130, No. DE-FG02-97ER41020, No. DE-FG02-97ER41033, No. DE-FG02-97ER41041, No. DE-SC0012612, No. DE-SC0014445, No. DE-SC0018060, No. DE-SC0022339, and LANLEM77/LANLEM78. We acknowledge support from the Particle Astrophysics Program and Nuclear Physics Program of the National Science Foundation through Grants No. MRI-0923142, No. PHY-1003399, No. PHY-1102292, No. PHY-1206314, No. PHY-1614611, No. PHY-1812409, No. PHY-1812356, No. PHY-2111140, and No. PHY-2209530. We gratefully acknowledge the support of the Laboratory Directed Research & Development (LDRD) program at Lawrence Berkeley National Laboratory for this work. We gratefully acknowledge the support of the U.S. Department of Energy through the Los Alamos National Laboratory LDRD Program and through the Pacific Northwest National Laboratory LDRD Program for this work. We gratefully acknowledge the support of the South Dakota Board of Regents Competitive Research Grant. We acknowledge the support of the Natural Sciences and Engineering Research Council of Canada, funding Reference No. SAPIN-2017-00023, and from the Canada Foundation for Innovation John R. Evans Leaders Fund. This research used resources provided by the Oak Ridge Leadership Computing Facility at Oak Ridge National Laboratory and by the National Energy Research Scientific Computing Center (NERSC), a U.S. Department of Energy Office of Science User Facility at Lawrence Berkeley National Laboratory. We thank our hosts and colleagues at the Sanford Underground Research Facility for their support.