Use of neural networks to analyze pulse shape data in low-background detectors
- 104 Downloads
Pacific Northwest National Laboratory has accumulated years of data with ultra-low-background proportional counters collected in an on-site shallow underground laboratory. This large dataset of events is exploited to study the impact of using neural networks for data analysis compared to simple pulse shape discrimination (PSD). The PSD method can introduce false positives for overlapping event distributions; however, a neural network can separate and correctly classify these events. This paper describes the training, testing, and validation of a neural network, analysis of challenge datasets, and a comparison between the standard PSD approach and a dense, fully-connected neural network.
KeywordsPulse shape discrimination Neural network Classification Low-background Gas proportional counter Machine learning
We would like to thank Dr. Nathan Hodas and Dr. Court Corley with the Data Science group at PNNL for providing advice and support for this work. The research described in this paper is part of the Agile Deep Science Initiative at Pacific Northwest National Laboratory and was conducted under the Laboratory Directed Research and Development (LDRD) Program. This work was performed by Pacific Northwest National Laboratory under award number DE-AC05-76RL01830. The research presented in this paper utilized the PNNL Institutional Computing (PIC) resources at Pacific Northwest National Laboratory. Information Release Number: PNNL-SA-133676.
- 2.Seifert A, Aalseth CE, Day AR, Fuller ES, Hoppe EW, Keillor ME, Mace EK, Overman CT, Warren GA (2013) The design, construction, and initial characterization of an ultra-low-background gas-proportional counting system. J Radioanal Nucl Chem 296(2):915–921. https://doi.org/10.1007/s10967-012-2059-5 CrossRefGoogle Scholar
- 7.Hoppe EW, Aalseth CE, Farmer OT, Hossbach TW, Liezers M, Miley HS, Overman NR, Reeves JH (2014) Reduction of radioactive backgrounds in electroformed copper for ultra-sensitive radiation detectors. Nucl Instrum Methods Phys Res Sect A 764:116–121. https://doi.org/10.1016/j.nima.2014.06.082 CrossRefGoogle Scholar
- 8.Mace EK, Aalseth CE, Bonicalzi RM, Day AR, Hoppe EW, Keillor ME, Myers AW, Overman CT, Seifert A (2013) Controlling low-rate signal path microdischarge for an ultra-low-background proportional counter. J Radioanal Nucl Chem 296(2):753–758. https://doi.org/10.1007/s10967-012-2042-1 CrossRefGoogle Scholar
- 10.Hennig W, Chu YX, Tan H, Fallu-Labruyere A, Warburton WK (2007) The DGF pixie-4 spectrometer: compact digital readout electronics for HPGe clover detectors. Nuclear Instrum Methods Phys Res Sect B Beam Interact Mater Atoms 263(1):175–178. https://doi.org/10.1016/j.nimb.2007.04.079 CrossRefGoogle Scholar
- 14.Hagan MT, Demuth HB, Beale MH (1996) Neural network design, vol 20. Pws Pub, BostonGoogle Scholar
- 15.Haykin S (1994) Neural networks, A comprehensive Foundation. Macmilan, BasingstokeGoogle Scholar
- 18.Goodfellow I, Bengio Y, Courville A (2016) Deep learning. The MIT Press, CambridgeGoogle Scholar
- 19.Chollet F (2015) Keras documentation. https://keras.io
- 20.Abadi M, Agarwal A, Barham P, Brevdo E, Chen Z, Citro C, Corrado GS, Davis A, Dean J, Devin M, Ghemawat S, Goodfellow I, Harp A, Irving G, Isard M, Jozefowicz R, Jia Y, Kaiser L, Kudlur M, Levenberg J, Mané D, Schuster M, Monga R, Moore S, Murray D, Olah C, Shlens J, Steine B, Sutskever I, Talwar K, Tucker P, Vanhoucke V, Vasudevan V, Viégas F, Vinyals O, Warden P, Wattenberg M, Wicke M, Yu Y, Zheng X (2015) TensorFlow: large-scale machine learning on heterogeneous systems. www.tensorflow.org
- 21.Kingma DP, Ba J (2014) Adam: A method for stochastic optimization. arXiv preprint arXiv:14126980