Physics-aware data analytics effectively treat sparse sensor data

Jordan Stomps; Kenneth Dayman; Birdy Phathanapirom

doi:10.1007/s10967-025-10476-6

Physics-aware data analytics effectively treat sparse sensor data

Jordan Stomps
, Kenneth Dayman
, Birdy Phathanapirom

Data Science and Engineering for Nonprol

Research output: Contribution to journal › Article › peer-review

Abstract

Physics-informed data augmentations have been used to improve data analytic models where attributed calibration data are limited, a common scenario in nuclear nonproliferation. This study used a small dataset of labeled gamma-ray spectra spanning several classes of shielded radiological material transfers collected at a real multiuse nuclear facility. Augmentations are used to increase the number of labeled data available for training supervised models for classifying transfer types. When trained with augmented data, relatively low-capacity models can achieve benchmark test performance with severely limited initial labeled data. This work motivates developing comparable augmentations for other measurement modalities relevant to nuclear nonproliferation.

Original language	English
Journal	Journal of Radioanalytical and Nuclear Chemistry
DOIs	https://doi.org/10.1007/s10967-025-10476-6
State	Accepted/In press - 2025

Funding

This work was funded by the Office of Defense Nuclear Nonproliferation Research and Development (NA-22), within the US Department of Energy’s National Nuclear Security Administration. The authors are thankful for the support of the MINOS collaboration at Oak Ridge National Laboratory—particularly Daniel Archer, James Ghawaly, Andrew Nicholson, and Michael Willis—for collecting, organizing, and sharing the data used in this project. Methods and Applications of Radioanalytical Chemistry (MARC XIII). MARC XIII Assigned Log Number: 678. Notice: This manuscript has been authored by UT-Battelle LLC under contract DE-AC05–00OR22725 with the US Department of Energy (DOE). The US government retains and the publisher, by accepting the article for publication, acknowledges that the US government retains a nonexclusive, paid-up, irrevocable, worldwide license to publish or reproduce the published form of this manuscript, or allow others to do so, for US government purposes. DOE will provide public access to these results of federally sponsored research in accordance with the DOE Public Access Plan (https://www.energy.gov/doe-public-access-plan).

Keywords

Data augmentations
Gamma spectroscopy
Nuclear nonproliferation
Semi-supervised learning

Access to Document

10.1007/s10967-025-10476-6

Cite this

@article{d7bbb89dfcdd4486b926137b4545f031,

title = "Physics-aware data analytics effectively treat sparse sensor data",

abstract = "Physics-informed data augmentations have been used to improve data analytic models where attributed calibration data are limited, a common scenario in nuclear nonproliferation. This study used a small dataset of labeled gamma-ray spectra spanning several classes of shielded radiological material transfers collected at a real multiuse nuclear facility. Augmentations are used to increase the number of labeled data available for training supervised models for classifying transfer types. When trained with augmented data, relatively low-capacity models can achieve benchmark test performance with severely limited initial labeled data. This work motivates developing comparable augmentations for other measurement modalities relevant to nuclear nonproliferation.",

keywords = "Data augmentations, Gamma spectroscopy, Nuclear nonproliferation, Semi-supervised learning",

author = "Jordan Stomps and Kenneth Dayman and Birdy Phathanapirom",

note = "Publisher Copyright: {\textcopyright} UT-Battelle, LLC, under exclusive licence to Springer Nature Switzerland AG 2025.",

year = "2025",

doi = "10.1007/s10967-025-10476-6",

language = "English",

journal = "Journal of Radioanalytical and Nuclear Chemistry",

issn = "0236-5731",

publisher = "Akad{\'e}miai Kiad{\'o}",

}

TY - JOUR

T1 - Physics-aware data analytics effectively treat sparse sensor data

AU - Stomps, Jordan

AU - Dayman, Kenneth

AU - Phathanapirom, Birdy

N1 - Publisher Copyright: © UT-Battelle, LLC, under exclusive licence to Springer Nature Switzerland AG 2025.

PY - 2025

Y1 - 2025

N2 - Physics-informed data augmentations have been used to improve data analytic models where attributed calibration data are limited, a common scenario in nuclear nonproliferation. This study used a small dataset of labeled gamma-ray spectra spanning several classes of shielded radiological material transfers collected at a real multiuse nuclear facility. Augmentations are used to increase the number of labeled data available for training supervised models for classifying transfer types. When trained with augmented data, relatively low-capacity models can achieve benchmark test performance with severely limited initial labeled data. This work motivates developing comparable augmentations for other measurement modalities relevant to nuclear nonproliferation.

AB - Physics-informed data augmentations have been used to improve data analytic models where attributed calibration data are limited, a common scenario in nuclear nonproliferation. This study used a small dataset of labeled gamma-ray spectra spanning several classes of shielded radiological material transfers collected at a real multiuse nuclear facility. Augmentations are used to increase the number of labeled data available for training supervised models for classifying transfer types. When trained with augmented data, relatively low-capacity models can achieve benchmark test performance with severely limited initial labeled data. This work motivates developing comparable augmentations for other measurement modalities relevant to nuclear nonproliferation.

KW - Data augmentations

KW - Gamma spectroscopy

KW - Nuclear nonproliferation

KW - Semi-supervised learning

UR - https://www.scopus.com/pages/publications/105021822456

U2 - 10.1007/s10967-025-10476-6

DO - 10.1007/s10967-025-10476-6

M3 - Article

AN - SCOPUS:105021822456

SN - 0236-5731

JO - Journal of Radioanalytical and Nuclear Chemistry

JF - Journal of Radioanalytical and Nuclear Chemistry

ER -

Physics-aware data analytics effectively treat sparse sensor data

Abstract

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this