Abstract
As more complex predictive models are used for gamma-ray spectral analysis, methods are needed to probe and understand their predictions and behavior. Recent work has begun to bring the latest techniques from the field of Explainable Artificial Intelligence (XAI) into the applications of gamma-ray spectroscopy, including the introduction of gradient-based methods like saliency mapping and Gradient-weighted Class Activation Mapping (Grad-CAM), and black box methods like Local Interpretable Model-agnostic Explanations (LIME) and SHapley Additive exPlanations (SHAP). In addition, new sources of synthetic radiological data are becoming available, and these new data sets present opportunities to train models using more data than ever before. In this work, we use a neural network model trained on synthetic NaI(Tl) urban search data to compare some of these explanation methods and identify modifications that need to be applied to adapt the methods to gamma-ray spectral data. We find that the black box methods LIME and SHAP are especially accurate in their results, and recommend SHAP since it requires little hyperparameter tuning. We also propose and demonstrate a technique for generating counterfactual explanations using orthogonal projections of LIME and SHAP explanations.
Funder
Office of Defense Nuclear Nonproliferation
U.S. Department of Energy by Lawrence Berkeley National Laboratory
U.S. Department of Energy
National Nuclear Security Administration
US Department of Energy
DOE Public Access Plan
Publisher
Public Library of Science (PLoS)
Reference48 articles.
1. Detecting Illicit Nuclear Materials;R Kouzes;American Scientist,2005
2. The Photon Haystack and Emerging Radiation Detection Technology;RC Runkle;Journal of Applied Physics,2009
3. A new approach to automatic radiation spectrum analysis;P Olmos;IEEE Transactions on Nuclear Science,1991
4. Application of neural network techniques in gamma spectroscopy;P Olmos;Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment,1992
5. Keller PE, Kouzes RT. Gamma spectral analysis via neural networks. In: Proceedings of 1994 IEEE Nuclear Science Symposium—NSS’94. vol. 1; 1994. p. 341–345 vol.1.