Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Synthetic Data”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 325 records · Page 18
All sourcesNASA NTRSDOE OSTI
NASA NTRS2024

Artificial Intelligence Medical Support for Long-Duration Space Missions

We envision an artificial intelligence (AI) based system that will provide support and recommendations to the crew medical officer (CMO) and ground flight surgeon during long-duration space missions. Such a system would be pretrained on the knowledgebase of clinical knowledge on Earth, minimizing the amount of Earth data that needs to be transferred into space. Then during deployment, the system would be constantly refined through active learning from diverse streams of data from sensors in the spacecraft, data collected daily from individual astronauts, and human-in-the-loop feedback from the crew. The model could be interrogated for predictions and recommendations on personalized crew health based on the overall status of the spacecraft, medicinal stores, and status of other crew members. Adaptation techniques would be used to incorporate spaceflight data that have very different distributions from the training data due to the extreme environment. Edge computing and the most advanced neuromorphic processing would enable computation in scenarios with low power and bandwidth, while dimensionality reduction would be employed to ensure that the input data streams from spaceflight are as small as possible. In order to realize this long-term vision, several hardware and software aspects need to be developed and assembled. First, models pretrained on Earth biomedical data would need to be evaluated for predictive accuracy, and the best one selected. That model would need to be adapted to learn from diverse, sparse, and inconsistently measured data streams, as well as human-in-the-loop feedback. A data integration, standardization, and dimensionality reduction methodology would need to be developed to handle all data types and feed them into the model. Once the software and data infrastructure is developed, it would need to be integrated with small footprint compute processors and tested in high-radiation, high-vibration, unregulated temperature situations. As a short-term goal, we recommend to focus on the development of the data and model software structure. Several large language models (LLM) already exist that have been trained on Earth biomedical and clinical knowledgebases, including BioMedLLM, Med-PaLM, SPOKE LLM, and Foresight. These models need to be evaluated for accuracy and the best one chosen for a proof-of-concept structure, while maintaining awareness of the accelerating AI field and incorporating any newly improved model architectures as needed. Then, we recommend to develop a database of synthetic data types to mimic the diverse data streams that are expected in a long-duration space mission. This should include environmental and microbial data from the spacecraft, non-invasive data from wearables and point-of-care devices employed by astronauts, and more invasive molecular and physiological monitoring of clinical and biomarker data from astronauts. The data standardization methodology should be developed, and these data streams used to refine the clinical LLM. Several scenarios should be developed that could plausibly come up in a long-duration space mission, and changes or aberrations introduced to the data at specific times to mimic these scenarios. Then, question and answer tasks should be designed to interrogate the model for predictions and recommendations, with acceptable answers already identified.

Artificial Intelligence↗
DOE OSTI2022

Detecting anomalous sequences in electronic health records using higher-order tensor networks

Detecting anomalous sequences is an integral part of building and protecting modern large-scale health information technology (HIT) systems. These HIT systems generate a large volume of records of patients’ state and significant events, which provide a valuable resource to help improve clinical decisions, patient care processes, and other issues. However, detecting anomalous sequences in electronic health records (EHR) remains a challenge in healthcare applications for several reasons, including imbalances in the data, complexity of relationships between events in the sequence, and the curse of dimensionality. Conventional anomaly detection methods use the finite sequence of events to discriminate sequences. They fail to incorporate salient event details under variable higher-order dependencies (e.g., duration between events) that can provide better discrimination of sequences in their models. To address this problem, we propose event sequence and subsequence anomaly detection algorithms that (1) use network-based representations of interactions in the data, (2) account for variable higher-order dependencies in the data, and (3) incorporate events duration for adequate discrimination of the data. The proposed approach identifies anomalies by monitoring the change in the graph after the test sequence is removed from the network. The change is quantified using graph distance metrics so that dramatic changes in the network can be attributed to the removed sequence. Furthermore, the proposed subsequence algorithm recommends plausible paths and salient information for the detected anomalous subsequences. Our results show that the proposed event sequence anomaly detection algorithm outperforms the baseline methods for both synthetic data and real-world EHR data.

96 KNOWLEDGE MANAGEMENT AND PRESERVATION↗
DOE OSTI2021

Data–Driven Velocity Model Evaluation Using K–Means Clustering

In this work, we develop a data-driven clustering method to evaluate a velocity model using surface wave velocity dispersion. This is done by first computing theoretical dispersion curves for 1-D velocity profiles of all the grid locations and then splitting the resulting dispersion curves into a certain number of groups via the K-means clustering. The observed dispersion curves are also clustered following the same procedure and the velocity model is assessed by comparing the spatial patterns obtained for the observed and synthetic data sets. The method is applied to evaluate two community velocity models in southern California, CVM-S4.26 and CVM-H15.1, using phase velocity maps derived for 3–16 s Rayleigh waves. We found a good correlation in the spatial distribution of clusters between the result of CVM-S4.26 and that of the observed data, suggesting that the CVM-S4.26 fits the observed dispersion maps better than the CVM-H15.1 in terms of features extracted from the clustering analysis.

58 GEOSCIENCES↗
DOE OSTI2025

Evaluating quenching in cosmological simulations of galaxy formation with spectral covariance in the optical window

ABSTRACT Cosmological hydrodynamical simulations provide valuable insights on galaxy evolution when coupled with observational data. Comparisons with real galaxies are typically performed via scaling relations of the observables. Here, we follow an alternative approach based on the spectral covariance in a model-independent way. We build upon previous work by Sharbaf et al. that studied the covariance of high-quality SDSS (Sloan Digital Sky Survey) continuum-subtracted spectra in a relatively narrow range of velocity dispersion ($\sigma \in [100,150]$ km s$^{-1}$). Here, the same analysis is applied to synthetic data from the eagle and IllustrisTNG100 simulations, to assess the ability of these runs to mimic real galaxies. The real and simulated spectra are consistent regarding spectral covariance, although with subtle differences that can inform the implementation of subgrid physics. Spectral fitting done a posteriori on stacks segregated with respect to latent space reveals that the first principal component (PC1) is predominantly influenced by the stellar age distribution, with an underlying age–metallicity degeneracy. Good agreement is found regarding star formation prescriptions but there is disagreement with active galactic nucleus (AGN) feedback, that also affects the subset of quiescent galaxies. We show a substantial difference in the implementation of the AGN subgrid prescriptions, regarding central black hole seeding, that could lead to the mismatch. Differences are manifest between these two simulations in the star formation histories stacked with respect to latent space. We emphasize that this methodology only relies on the spectral covariance to assess whether simulations provide a true representation of galaxy formation.

Sharbaf, Z. (ORCID:0009000450545946)↗
DOE OSTI2023

How reliable is distribution of relaxation times (DRT) analysis? A dual regression-classification perspective on DRT estimation, interpretation, and accuracy

The distribution of relaxation times (DRT) has gained increasing attention and adoption in recent years as a versatile method for analyzing electrochemical impedance spectroscopy (EIS) data obtained from complex devices like fuel cells, electrolyzers, and batteries. The DRT deconvolutes the impedance without a priori specification of a generative model, which is especially useful for interpretation and model selection when the governing principles of the system under study are not fully understood. However, DRT estimation is an ill-posed inversion problem that must be addressed with a subjective choice of regularization and tuning, which leaves substantial risk of misleading interpretations of EIS data. In this work, we suggest a new classification view of the DRT inversion to clarify DRT estimation and interpretation. We introduce a dual regression-classification framework that unifies the classification and regression views of the DRT inversion with wide-reaching implications for DRT analysis. The dual framework is employed to demonstrate a new kind of DRT inversion algorithm and develop novel evaluation metrics that capture previously ignored aspects of DRT accuracy. These approaches are applied to both synthetic data and experimental spectra collected from a protonic ceramic fuel cell and a lithium-ion battery to illustrate their broad utility. The dual inversion algorithm shows promising performance for accurate DRT estimation and autonomous model identification, while the dual evaluation approach produces metrics that meaningfully assess the strengths and risks of DRT algorithms. Here this work provides valuable insight for both practical application of the DRT to experimental data and further development of EIS analysis methods.

36 MATERIALS SCIENCE↗
NASA NTRS1990

Studies in astronomical time series analysis. IV - Modeling chaotic and random processes with linear filters

While chaos arises only in nonlinear systems, standard linear time series models are nevertheless useful for analyzing data from chaotic processes. This paper introduces such a model, the chaotic moving average. This time-domain model is based on the theorem that any chaotic process can be represented as the convolution of a linear filter with an uncorrelated process called the chaotic innovation. A technique, minimum phase-volume deconvolution, is introduced to estimate the filter and innovation. The algorithm measures the quality of a model using the volume covered by the phase-portrait of the innovation process. Experiments on synthetic data demonstrate that the algorithm accurately recovers the parameters of simple chaotic processes. Though tailored for chaos, the algorithm can detect both chaos and randomness, distinguish them from each other, and separate them if both are present. It can also recover nonminimum-delay pulse shapes in non-Gaussian processes, both random and chaotic.

Scargle, Jeffrey D.↗
NASA NTRS2003

Earth Science Imagery Registration

The study of global environmental changes involves the comparison, fusion, and integration of multiple types of remotely-sensed data at various temporal, radiometric, and spatial resolutions. Results of this integration may be utilized for global change analysis, as well as for the validation of new instruments or for new data analysis. Furthermore, future multiple satellite missions will include many different sensors carried on separate platforms, and the amount of remote sensing data to be combined is increasing tremendously. For all of these applications, the first required step is fast and automatic image registration, and as this need for automating registration techniques is being recognized, it becomes necessary to survey all the registration methods which may be applicable to Earth and space science problems and to evaluate their performances on a large variety of existing remote sensing data as well as on simulated data of soon-to-be-flown instruments. In this paper we present one of the first steps toward such an exhaustive quantitative evaluation. First, the different components of image registration algorithms are reviewed, and different choices for each of these components are described. Then, the results of the evaluation of the corresponding algorithms combining these components are presented o n several datasets. The algorithms are based on gray levels or wavelet features and compute rigid transformations (including scale, rotation, and shifts). Test datasets include synthetic data as well as data acquired over several EOS Land Validation Core Sites with the IKONOS and the Landsat-7 sensors.

LeMoigne, Jacqueline↗
DOE OSTI2021

Multi-fidelity Bayesian neural networks: Algorithms and applications

Here we propose a new class of Bayesian neural networks (BNNs) that can be trained using noisy data of variable fidelity, and we apply them to learn function approximations as well as to solve inverse problems based on partial differential equations (PDEs). These multi-fidelity BNNs consist of three neural networks: The first is a fully connected neural network, which is trained following the maximum a posteriori probability (MAP) method to fit the low-fidelity data; the second is a Bayesian neural network employed to capture the cross-correlation with uncertainty quantification between the low- and high-fidelity data; and the last one is the physics-informed neural network, which encodes the physical laws described by PDEs. For the training of the last two neural networks, we first employ the mean-field variational inference (VI) to maximize the evidence lower bound (ELBO) to obtain informative prior distributions for the hyperparameters in the BNNs, and subsequently we use the Hamiltonian Monte Carlo (HMC) method to estimate accurately the posterior distributions for the corresponding hyperparameters. We demonstrate the accuracy of the present method using synthetic data as well as real measurements. Specifically, we first approximate a one- and four-dimensional function, and then infer the reaction rates in one- and two-dimensional diffusion-reaction systems. Moreover, we infer the sea surface temperature (SST) in the Massachusetts and Cape Cod Bays using satellite images and in-situ measurements. Taken together, our results demonstrate that the present method can capture both linear and nonlinear correlation between the low- and high-fidelity data adaptively, identify unknown parameters in PDEs, and quantify uncertainties in predictions, given a few scattered noisy high-fidelity data. Finally, we demonstrate that we can effectively and efficiently reduce the uncertainties and hence enhance the prediction accuracy with an active learning approach, using as examples a specific one-dimensional function approximation and an inverse PDE problem.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2021

Process Anomaly Detection for Sparsely Labeled Events in Nuclear Power Plants

An essential aspect of online monitoring, subtle anomaly detection increases the detection lead time for equipment failure and enables a nuclear power plant (NPP) to mitigate unexpected partial or full outages, resulting in significant cost saving to the plant. Once an anomaly is detected by plant staff, its cause and severity are investigated. Because the vast majority of anomalies require some level of investigation, including some that require time-consuming examination, before they are passed over to the engineering organization for further analysis, plants are often equipped with tools to assist the staff in performing anomaly detection. Those tools operate as a black box and are often based on statistical methods that establish sensor correlations using preconfigured mathematical models and flag correlation deviations as anomalies. Due to the number of anomalies detected at a given NPP on a daily basis, a significant number of flagged anomalies usually await examination for days or weeks. A primary cause of this backlog is that the methods used by the tools generate many false positives. Though this is usually attributed to oversensitive model settings due to very narrow normal operation bands, it can also be associated with the model development being inadequate for the process being monitored, or with missing model inputs that could have explained misclassified positives. The performance of anomaly detection tools impacts their plant acceptance and utilization, especially when the effort to address false positives generated by the tool depletes the value or cost saved by using that tool. Thus, means to advance anomaly detection performance have been investigated by the Department of Energy’s Light Water Reactor Sustainability program. Previous and ongoing efforts have targeted unsupervised machine-learning (ML) methods, which do not require the labeling of any data fed into the ML model. By contrast, in supervised anomaly detection methods, every data point is labeled as either a normal or abnormal process condition, and the model is trained to replicate the classification process. Supervised methods usually outperform unsupervised methods, due to the added value in differentiating normal from anomalous states of the monitored process. An NPP’s corrective action program requires it to track and document, via a dedicated report, the resolution of any issues that occur within the plant. Once created, each report is reviewed by a plant screening committee, and several classifications and decisions are made. Recently, a collaborating NPP developed an artificial intelligence and ML-based classifier to categorize a condition report (CR) into classes that can serve to label the data as normal or anomalous. Applying CRs as labels represents a semi-supervised use case. Semi-supervised ML assumes that labels exist for some data points (i.e., labeled anomalies, in this case) but not for the rest. In this effort, semi-supervised ML methods were used to fuse data from CRs with anomaly detection methods in order to test the hypothesis that partially labeled anomalies would improve the accuracy of the anomaly detection methods. Specifically, two methods were used. The first is the deep Semi-supervised Anomaly Detection (deep SAD) method, which can handle labels ranging from fully unsupervised to fully supervised cases. The second is a newly designed ML method developed specifically for this effort and referred to as the high-order feature (HOF)-based method. To evaluate these two methods in controlled environments, synthetic data generators were developed and used. The first datasets used a spring-mass-damper (SMD) system simulator commonly found in mechanical engineering references. This was used to create two use cases: a one- and a three-mass system. Anomalies were introduced by changing the spring and damper coefficients while the system was actuated by random forces. The second datasets used the commercial Dymola-Modelica software to build a simplified nuclear reactor model. Anomalies were added in the form of corrupted sensor readings and/or control commands. The deep SAD method was tested using the SMD system, while the HOF method was tested using both datasets. Application of the deep SAD semi-supervised ML method demonstrated that labels can generate increased confidence in detecting true anomalies. This helped increase the number of true positives and decrease the number of false negatives—something that would aid in addressing the backlog of possible anomalies. Application of the HOF method demonstrated that labels can aid in down selecting from a candidate set of features to a more optimal subset in order to better differentiate between normal and anomalous conditions.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2022

Single Gaussian process method for arbitrary tokamak regimes with a statistical analysis

Abstract Gaussian process regression is a Bayesian method for inferring profiles based on input data. The technique is increasing in popularity in the fusion community due to its many advantages over traditional fitting techniques including intrinsic uncertainty quantification and robustness to over-fitting. This work investigates the use of a new method, the change-point method, for handling the varying length scales found in different tokamak regimes. The use of the Student’s t-distribution for the Bayesian likelihood probability is also investigated and shown to be advantageous in providing good fits in profiles with many outliers. To compare different methods, synthetic data generated from analytic profiles is used to create a database enabling a quantitative statistical comparison of which methods perform the best. Using a full Bayesian approach with the change-point method, Matérn kernel for the prior probability, and Student’s t-distribution for the likelihood is shown to give the best results.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2020

Determination of latent dimensionality in international trade flow

Currently, high-dimensional data is ubiquitous in data science, which necessitates the development of techniques to decompose and interpret such multidimensional (aka tensor) datasets. Finding a low dimensional representation of the data, that is, its inherent structure, is one of the approaches that can serve to understand the dynamics of low dimensional latent features hidden in the data. Moreover, decomposition methods with non-negative constraints are shown to extract more insightful factors. Nonnegative RESCAL is one such technique, particularly well suited to analyze self-relational data, such as dynamic networks found in international trade flows. Particularly, non-negative RESCAL computes a low dimensional tensor representation by finding the latent space containing multiple modalities. Furthermore, estimating the dimensionality of this latent space is crucial for extracting meaningful latent features. Here, to determine the dimensionality of the latent space with non-negative RESCAL, we propose a latent dimension determination method which is based on clustering of the solutions of multiple realizations of non-negative RESCAL decompositions. We demonstrate the performance of our model selection method on synthetic data. We then apply our method to decompose a network of international trade flows data from International Monetary Fund and shows that with a correct latent dimension determination, the resulting features are able to capture relevant empirical facts from economic literature.

97 MATHEMATICS AND COMPUTING↗
NASA NTRS2012

Can Selforganizing Maps Accurately Predict Photometric Redshifts?

We present an unsupervised machine-learning approach that can be employed for estimating photometric redshifts. The proposed method is based on a vector quantization called the self-organizing-map (SOM) approach. A variety of photometrically derived input values were utilized from the Sloan Digital Sky Survey's main galaxy sample, luminous red galaxy, and quasar samples, along with the PHAT0 data set from the Photo-z Accuracy Testing project. Regression results obtained with this new approach were evaluated in terms of root-mean-square error (RMSE) to estimate the accuracy of the photometric redshift estimates. The results demonstrate competitive RMSE and outlier percentages when compared with several other popular approaches, such as artificial neural networks and Gaussian process regression. SOM RMSE results (using delta(z) = z(sub phot) - z(sub spec)) are 0.023 for the main galaxy sample, 0.027 for the luminous red galaxy sample, 0.418 for quasars, and 0.022 for PHAT0 synthetic data. The results demonstrate that there are nonunique solutions for estimating SOM RMSEs. Further research is needed in order to find more robust estimation techniques using SOMs, but the results herein are a positive indication of their capabilities when compared with other well-known methods

Way, Michael J.↗
NASA NTRS2005

Radiative Studies of Planetary Atmospheres

Retrieval algorithms and associated software for application to CIRS infrared spectral data have been developed and coded. A general forward radiative transfer code has been written that runs efficiently on a Macintosh, even at high spectral resolution (0.5 per centimeter). It makes use of the correlated-k approach for representation of the gaseous absorption and can include those gases listed in the HITRAN and GEISA atlases, along with collision-induced absorption. Cloud effects are included as spectrally dependent absorbers. Provision has been made for future extension to include particle scattering in an n-stream approximation. The primary purpose of the code is to produce synthetic data and to serve as the forward calculating element in gas and cloud retrieval programs developed for the Mac as well as other platforms. Initial development of algorithms and production software suitable for application to CIRS data to be obtained from Jupiter, Saturn and Titan has been completed, and production versions of the software for application to the spectral data are in place. This includes temperature, gaseous constituent, and cloud opacity retrieval, algorithms that can be applied to both nadir and limb data. This work has been done as a cooperative effort between Conrath and Matcheva (Cornell), Achterberg (GSFWSSAI), and Flasar (GSFC).

Conrath, Barney J.↗
DOE OSTI2024

Robust unfolding of MeV x-ray spectra from filter stack spectrometer data

Here, we present an inversion method capable of robustly unfolding MeV x-ray spectra from filter stack spectrometer (FSS) data without requiring an a priori specification of a spectral shape or arbitrary termination of the algorithm. Our inversion method is based upon the perturbative minimization (PM) algorithm, which has previously been shown to be capable of unfolding x-ray transmission data, albeit for a limited regime in which the x-ray mass attenuation coefficient of the filter material increases monotonically with x-ray energy. Our inversion method improves upon the PM algorithm through regular smoothing of the candidate spectrum and by adding stochasticity to the search. With these additions, the inversion method does not require a physics model for an initial guess, fitting, or user-selected termination of the search. Instead, the only assumption made by the inversion method is that the x-ray spectrum should be near a smooth curve. Testing with synthetic data shows that the inversion method can successfully recover the primary large-scale features of MeV x-ray spectra, including the number of x-rays in energy bins of several-MeV widths to within 10%. Fine-scale features, however, are more difficult to recover accurately. Examples of unfolding experimental FSS data obtained at the Texas Petawatt Laser Facility and the OMEGA EP laser facility are also presented.

47 OTHER INSTRUMENTATION↗
DOE OSTI2022

Quasar Identification Using Multivariate Probability Density Estimated from Nonparametric Conditional Probabilities

Nonparametric estimation for a probability density function that describes multivariate data has typically been addressed by kernel density estimation (KDE). A novel density estimator recently developed by Farmer and Jacobs offers an alternative high-throughput automated approach to univariate nonparametric density estimation based on maximum entropy and order statistics, improving accuracy over univariate KDE. This article presents an extension of the single variable case to multiple variables. The univariate estimator is used to recursively calculate a product array of one-dimensional conditional probabilities. In combination with interpolation methods, a complete joint probability density estimate is generated for multiple variables. Good accuracy and speed performance in synthetic data are demonstrated by a numerical study using known distributions over a range of sample sizes from 100 to 10 6 for two to six variables. Performance in terms of speed and accuracy is compared to KDE. The multivariate density estimate developed here tends to perform better as the number of samples and/or variables increases. As an example application, measurements are analyzed over five filters of photometric data from the Sloan Digital Sky Survey Data Release 17. The multivariate estimation is used to form the basis for a binary classifier that distinguishes quasars from galaxies and stars with up to 94% accuracy.

79 ASTRONOMY AND ASTROPHYSICS↗
NASA NTRS1999

Using a Genetic Algorithm to Model Broadband Regional Waveforms for Crustal Structure in the Western United States

In this study, we analyze regional seismograms to obtain the crustal structure in the eastern Great Basin and western Colorado plateau. Adopting a for- ward-modeling approach, we develop a genetic algorithm (GA) based parameter search technique to constrain the one-dimensional crustal structure in these regions. The data are broadband three-component seismograms recorded at the 1994-95 IRIS PASSCAL Colorado Plateau to Great Basin experiment (CPGB) stations and supplemented by data from U.S. National Seismic Network (USNSN) stations in Utah and Nevada. We use the southwestern Wyoming mine collapse event (M(sub b) = 5.2) that occurred on 3 February 1995 as the seismic source. We model the regional seismograms using a four-layer crustal model with constant layer parameters. Timing of teleseismic receiver functions at CPGB stations are added as an additional constraint in the modeling. GA allows us to efficiently search the model space. A carefully chosen fitness function and a windowing scheme are added to the algorithm to prevent search stagnation. The technique is tested with synthetic data, both with and without random Gaussian noise added to it. Several separate model searches are carried out to estimate the variability of the model parameters. The average Colorado plateau crustal structure is characterized by a 40-km-thick crust with velocity increases at depths of about 10 and 25 km and a fast lower crust while the Great Basin has approximately 35- km-thick crust and a 2.9-km-thick sedimentary layer.

Bhattacharyya, Joydeep↗
DOE OSTI2022

Selective Dynamical Imaging of Interferometric Data

Recent developments in very long baseline interferometry (VLBI) have made it possible for the Event Horizon Telescope (EHT) to resolve the innermost accretion flows of the largest supermassive black holes on the sky. The sparse nature of the EHT’s (u, v)-coverage presents a challenge when attempting to resolve highly time-variable sources. We demonstrate that the changing (u, v)-coverage of the EHT can contain regions of time over the course of a single observation that facilitate dynamical imaging. These optimal time regions typically have projected baseline distributions that are approximately angularly isotropic and radially homogeneous. We derive a metric of coverage quality based on baseline isotropy and density that is capable of ranking array configurations by their ability to produce accurate dynamical reconstructions. We compare this metric to existing metrics in the literature and investigate their utility by performing dynamical reconstructions on synthetic data from simulated EHT observations of sources with simple orbital variability. We then use these results to make recommendations for imaging the 2017 EHT Sgr A* data set.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2024

Multilabel proportion prediction and out-of-distribution detection on gamma spectra of short-lived fission products

In the machine learning problem of multilabel classification, the objective is to determine for each test instance which classes the instance belongs to. In this work, we consider an extension of multilabel classification, called multilabel proportion prediction, in the context of radioisotope identification (RIID) using gamma spectra data. We aim to not only predict radioisotope proportions, but also identify out-of-distribution (OOD) spectra. We achieve this goal by viewing gamma spectra as discrete probability distributions, and based on this perspective, we develop a custom semi-supervised loss function that combines a traditional supervised loss with an unsupervised reconstruction error function. Our approach was motivated by its application to the analysis of short-lived fission products from spent nuclear fuel. In particular, we demonstrate that a neural network model trained with our loss function can successfully predict the relative proportions of 37 radioisotopes simultaneously. The model trained with synthetic data was then applied to measurements taken by Pacific Northwest National Laboratory (PNNL) to conduct analysis typically done by subject-matter experts. Here, we also extend our approach to successfully identify when measurements are OOD, and thus should not be trusted, whether due to the presence of a novel source or novel proportions.

Anomaly detection↗