Search NASA⌕ Search

SEARCH · Search NASA

Results for “Bayesian network”

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 361 records · Page 20

Probabilistic Nanomagnetic Memories for Uncertain and Robust Machine Learning

This project evaluated the use of emerging spintronic memory devices for robust and efficient variational inference schemes. Variational inference (VI) schemes, which constrain the distribution for each weight to be a Gaussian distribution with a mean and standard deviation, are a tractable method for calculating posterior distributions of weights in a Bayesian neural network such that this neural network can also be trained using the powerful backpropagation algorithm. Our project focuses on domain-wall magnetic tunnel junctions (DW-MTJs), a powerful multi-functional spintronic synapse design that can achieve low power switching while also opening the pathway towards repeatable, analog operation using fabricated notches. Our initial efforts to employ DW-MTJs as an all-in-one stochastic synapse with both a mean and standard deviation didn’t end up meeting the quality metrics for hardware-friendly VI. In the future, new device stacks and methods for expressive anisotropy modification may make this idea still possible. However, as a fall back that immediately satisfies our requirements, we invented and detailed how the combination of a DW-MTJ synapse encoding the mean and a probabilistic Bayes-MTJ device, programmed via a ferroelectric or ionically modifiable layer, can robustly and expressively implement VI. This design includes a physics-informed small circuit model, that was scaled up to perform and demonstrate rigorous uncertainty quantification applications, up to and including small convolutional networks on a grayscale image classification task, and larger (Residual) networks implementing multi-channel image classification. Lastly, as these results and ideas all depend upon the idea of an inference application where weights (spintronic memory states) remain non-volatile, the retention of these synapses for the notched case was further interrogated. These investigations revealed and emphasized the importance of both notch geometry and anisotropy modification in order to further enhance the endurance of written spintronic states. In the near future, these results will be mapped to effective predictions for room temperature and elevated operation DW-MTJ memory retention, and experimentally verified when devices become available.

77 NANOSCIENCE AND NANOTECHNOLOGY↗

Comparing Automated Posterior Estimation Techniques for Modeling Strong Lenses In Ground-based Survey Data

Current and future ground-based cosmological surveys, such as the Dark Energy Survey (DES), and the Vera Rubin Observatory Legacy Survey of Space and Time (LSST), are predicted to discover thousands to tens of thousands of strong gravitational lenses. The large number of strong lenses discoverable in future surveys will make strong lensing a highly competitive and complementary cosmic probe. However, conventional lens modeling techniques are unable to scale up to the sheer number of lenses that will be discovered through upcoming surveys. Therefore, the use of automated lens analysis techniques is necessary. We demonstrate that machine learning methods can be used to automate the inference of informative model posteriors of strong lensing systems in ground-based surveys with credible uncertainty estimation. We present two Simulation-Based Inference (SBI) approaches for lens parameter estimation of galaxy-galaxy lenses. We demonstrate applications of Neural Posteriors Estima tors (NPEs) and Bayesian Neural Network (BNNs) to automate the inference of a 12-parameter lensing system for DES-like ground-based imaging data. We apply a suite of diagnostics (e.g., posterior coverage and SBC) to validate the performance of our methods. We find that NPEs outperform the BNN, producing posterior distributions that are for the most part both more accurate and more precise; in particular, several source-light model parameters are systematically biased in the BNN implementation.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Top Quark Produced Through the Electroweak Force: Discovery Using the Matrix Element Analysis and Search for Heavy Gauge Bosons Using Boosted Decision Trees

The top quark produced through the electroweak channel provides a direct measurement of the V tb element in the CKM matrix which can be viewed as a transition rate of a top quark to a bottom quark. This production channel of top quark is also sensitive to different theories beyond the Standard Model such as heavy charged gauged bosons termed W'. This thesis measures the cross section of the electroweak produced top quark using a technique based on using the matrix elements of the processes under consideration. The technique is applied to 2.3 fb -1 of data from the D0 detector. From a comparison of the matrix element discriminants between data and the signal and background model using Bayesian statistics, we measure the cross section of the top quark produced through the electroweak mechanism σ(p$\bar{p}$ → tb + X, tqb + X) = 4.30 -1.20 +0.98 pb. The measured result corresponds to a 4.9σ Gaussian-equivalent significance. By combining this analysis with other analyses based on the Bayesian Neural Network (BNN) and Boosted Decision Tree (BDT) method, the measured cross section is 3.94 ± 0.88 pb with a significance of 5.0σ, resulting in the discovery of electroweak produced top quarks. Using this measured cross section and constraining |V tb | < 1, the 95% confidence level (C.L.) lower limit is |V tb | > 0.78. Additionally, a search is made for the production of W' using the same samples from the electroweak produced top quark. An analysis based on the BDT method is used to separate the signal from expected backgrounds. No significant excess is found and 95% C.L. upper limits on the production cross section are set for W' with masses within 600-950 GeV. For four general models of W{prime} boson production using decay channel W' → t$\bar{p}$, the lower mass limits are the following: M(W' L with SM couplings) > 840 GeV; M(W' R ) > 880 GeV or 890 GeV if the right-handed neutrino is lighter or heavier than W{prime}{sub R}; and M(W' L+R ) > 915 GeV.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Utility of Social Modeling for Proliferation Assessment - Enhancing a Facility-Level Model for Proliferation Resistance Assessment of a Nuclear Enegry System

The Utility of Social Modeling for Proliferation Assessment project (PL09-UtilSocial) investigates the use of social and cultural information to improve nuclear proliferation assessments, including nonproliferation assessments, Proliferation Resistance (PR) assessments, safeguards assessments, and other related studies. These assessments often use and create technical information about a host State and its posture towards proliferation, the vulnerability of a nuclear energy system (NES) to an undesired event, and the effectiveness of safeguards. This objective of this project is to find and integrate social and technical information by explicitly considering the role of cultural, social, and behavioral factors relevant to proliferation; and to describe and demonstrate if and how social science modeling has utility in proliferation assessment. This report describes a modeling approach and how it might be used to support a location-specific assessment of the PR assessment of a particular NES. The report demonstrates the use of social modeling to enhance an existing assessment process that relies on primarily technical factors. This effort builds on a literature review and preliminary assessment performed as the first stage of the project and compiled in PNNL-18438. [ T his report describes an effort to answer questions about whether it is possible to incorporate social modeling into a PR assessment in such a way that we can determine the effects of social factors on a primarily technical assessment. This report provides: 1. background information about relevant social factors literature; 2. background information about a particular PR assessment approach relevant to this particular demonstration; 3. a discussion of social modeling undertaken to find and characterize social factors that are relevant to the PR assessment of a nuclear facility in a specific location; 4. description of an enhancement concept that integrates social factors into an existing, technically based nuclear facility assessment; 5. a discussion of a way to engage with the owners of the PR assessment methodology to assess and improve the enhancement concept; 6. a discussion of implementation of the proposed approach, including a discussion of functionality and potential users; and 7. conclusions from the research. This report represents technical deliverables for the NA-22 Simulations, Algorithms, and Modeling program. Specifically this report is the Task 2 and 3 deliverables for project PL09-UtilSocial.

29 ENERGY PLANNING, POLICY, AND ECONOMY↗

Comparative Analysis of Machine Learning Models for Day-Ahead Photovoltaic Power Production Forecasting

A main challenge for integrating the intermittent photovoltaic (PV) power generation remains the accuracy of day-ahead forecasts and the establishment of robust performing methods. The purpose of this work is to address these technological challenges by evaluating the day-ahead PV production forecasting performance of different machine learning models under different supervised learning regimes and minimal input features. Specifically, the day-ahead forecasting capability of Bayesian neural network (BNN), support vector regression (SVR), and regression tree (RT) models was investigated by employing the same dataset for training and performance verification, thus enabling a valid comparison. The training regime analysis demonstrated that the performance of the investigated models was strongly dependent on the timeframe of the train set, training data sequence, and application of irradiance condition filters. Furthermore, accurate results were obtained utilizing only the measured power output and other calculated parameters for training. Consequently, useful information is provided for establishing a robust day-ahead forecasting methodology that utilizes calculated input parameters and an optimal supervised learning approach. Finally, the obtained results demonstrated that the optimally constructed BNN outperformed all other machine learning models achieving forecasting accuracies lower than 5%.

14 SOLAR ENERGY↗

Global Framework for Emulation of Nuclear Calculations

We introduce a hierarchical framework that combines ab initio many-body calculations with a Bayesian neural network, developing emulators capable of accurately predicting nuclear properties across isotopic chains simultaneously and being applicable to different regions of the nuclear chart. We benchmark our developments using the oxygen isotopic chain, achieving accurate results for ground-state energies and nuclear charge radii, while providing robust uncertainty quantification. Our framework enables global sensitivity analysis of nuclear binding energies and charge radii with respect to the low-energy constants that describe the nuclear force.

FOS: Computer and information sciences↗

SOCIAL MODELING IN ASSESSEMENT OF A STATE’S PROPENSITY FOR NUCLEAR PROLIFERATION

This paper presents approach for assessing a State’s propensity for nuclear weapons proliferation using social modeling. We supported this modeling by first reviewing primarily literature by social scientists on factors related to the propensity of a State to proliferation and by leveraging existing relevant data compiled by social scientists. We performed a number of validation tests on our model including one that incorporates use of benchmark data defining the proliferation status of countries in the years between 1945 and 2000. We exercise the BN model against a number of country cases representing different perceived levels of proliferation risk. We also describe how the BN model could be further refined to be a proliferation assessment tool for decision making.

98 NUCLEAR DISARMAMENT, SAFEGUARDS, AND PHYSICAL P↗

Strong Lensing Parameter Estimation on Ground-Based Imaging Data Using Simulation-Based Inference

Current ground-based cosmological surveys, such as the Dark Energy Survey (DES), are predicted to discover thousands of galaxy-scale strong lenses, while future surveys, such as the Vera Rubin Observatory Legacy Survey of Space and Time (LSST) will increase that number by 1-2 orders of magnitude. The large number of strong lenses discoverable in future surveys will make strong lensing a highly competitive and complementary cosmic probe. To leverage the increased statistical power of the lenses that will be discovered through upcoming surveys, automated lens analysis techniques are necessary. We present two Simulation-Based Inference (SBI) approaches for lens parameter estimation of galaxy-galaxy lenses. We demonstrate the successful application of Neural Posterior Estimation (NPE) to automate the inference of a 12-parameter lens mass model for DES-like ground-based imaging data. We compare our NPE constraints to a Bayesian Neural Network (BNN) and find that it outperforms the BNN, producing posterior distributions that are for the most part both more accurate and more precise; in particular, several source-light model parameters are systematically biased in the BNN implementation.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Finite-Test-Case Development to Demonstrate Logical Integrity of Software in DRPS

Reactor protection system (RPS) in nuclear power plant (NPP) is in the process of digitalization in response to advanced features of digital systems. In this context, software integrity in the RPS becomes an important issue since it may cause common-cause failure. Therefore, to guarantee the safety of NPPs, the integrity of the software need to be properly demonstrated or assessed. There are three typical approaches for software reliability quantification: software reliability growth model (SRGM), Bayesian belief network (BBN), and test-based method. Basically, SRGM is difficult to apply to the safety-critical software because not only sufficient failure sets are unavailable but consequence of system failure is also unacceptable. Then the result obtained through the BBN model may have large uncertainty, so it produces difficulties in integrity demonstration of safety-critical software. Thus, the BBN result need to be complemented or verified by test-based method. Currently, for the integrity demonstration of safety-critical software, black-box test which feeds inputs and examines just outputs whether they succeeded or failed is considered based on trajectory-form-of-input (a series of successive values for the input variables that occur during the operation of the software over time). Usually these trajectory-inputs are randomly sampled from operational profile. In this approach, however, there are some practical difficulties. Therefore, we proposed a new test method for safety-critical software in digitalized reactor protection system (DRPS). The main contents described in this paper are about basic concepts under development of finite-test-cases by which the integrity of the software logic can be verified. It should be noted that the influence of the environments, such as operating system and other hardware, are not considered since the method focuses on software logic only. (authors)

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

A Risk-Informed Approach to Trustworthiness Assessment in Digital Twins-Based Autonomous Control

In autonomous control systems, digital twins (DTs) are used to perform diagnostic and prognostic functions. The trustworthiness of these DTs is dependent on quality and coverage of the training data, model accuracy and integrity of sensor data. This work introduces a methodology to determine the trustworthiness of a DT system given faulty sensor data using a risk informed approach. Bayesian Belief Networks (BBNs) are used to propagate uncertainties and determine the probability of trustable recommendations. The decision to trust the control action provided by the DT is based on the DT output, expert opinion, and severity of problems. The performance of DTs is reliant on the data they are trained on. When they encounter out of distribution data, the trustworthiness of the recommendations decreases. To address this issue, we include an expert component that provides input on sensor degradation. For this, we utilize a generative artificial intelligence (AI) model, such as Generative Pretrained Transformer (GPT). The GPT functions as an expert with broad knowledge. The GPT is fine-tuned to understand and discriminate sensor degradation scenarios using manufactured data. This methodology is demonstrated through a case study on a Nearly Autonomous Management and Control System (NAMAC) during a steady state scenario. Various sensor degradation types with different severity levels are considered. Degraded sensor data is processed by the DT system and the fine-tuned GPT. Finally, using the BBN, we combine the GPT information and the DT output with its sources of uncertainty. This provides an output regarding the trustworthiness of the DT recommendation.

22 - GENERAL STUDIES OF NUCLEAR REACTORS↗

Search for flavor changing neutral currents in single top quark production using 2.3 fb$^-1$ of $p\bar{p}$ collisions

We present a search for flavor changing neutral currents via quark-gluon couplings in a sample of single top quark final states corresponding to 2.3 fb{sup -1} of integrated luminosity collected with the D0 detector at the Fermilab Tevatron Collider. We select events containing a single top quark candidates with an additional jet, and obtain separation between signal and background using Bayesian neural networks. We find consistency between background expectation and observed data, and set limits on avor changing neutral current gluon couplings of the top quark to up quarks (tgu) and charm quarks (tgc). The cross section limits at the 95% C.L. are {sigma}{sub tgu} < 0.20 pb and {sigma}{sub tgc} < 0.27 pb. These correspond to limits on the top quark decay branching fractions of B(t {yields} gu) < 2.0 x 10{sup -4} and B(t {yields} gc) < 3.9 x 10{sup -3}.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Bayesian Physics Informed Spatio-Temporal Network for Streamflow Data Imputation

Reliable reconstruction of incomplete streamflow records is critical for improving hydrological forecasting, flood preparedness, and water resource management. However, large observational gaps and uncertainties in governing physical parameters limit the accuracy of traditional statistical and machinelearning imputation frameworks. To address these challenges, we develop a Bayesian Physics-Informed Spatio-Temporal Network (BPI-STNet) that jointly captures spatial and temporal dependencies while enforcing hydrologic consistency through embedded physical constraints. The framework integrates a GraphSAGE-LSTM architecture to model spatial connectivity across gauges and temporal flow dynamics, coupled with a Bayesian update mechanism to estimate uncertain parameters in a simplified water-balance framework. Unlike conventional physics-informed networks that rely on sampling-based posterior estimation, BPI-STNet derives an analytic solution to the inverse problem, allowing closed-form Bayesian updates of uncertain parameters Λ={α,β,k} using Gaussian priors and likelihoods. Applied to daily observations from the Susquehanna River Basin (1980-2022), BPI-STNet achieves substantial improvements over a purely data-driven RGNN baseline, which reduced RMSE by 23 % and MAE by 9 %, and achieving an average NSE values up to 0.96. The results demonstrate that coupling Bayesian inference with physics-informed learning yields physically consistent, uncertainty-aware reconstructions that preserve the temporal persistence and statistical distribution of observed flows. The proposed framework establishes a generalizable paradigm for data-sparse hydrologic systems where both data fidelity and physical interpretability are essential.

Krishnan Kutty Ambika, Anukesh [ORNL] (ORCID:00000↗

Efficient Neural Network Approaches for Conditional Optimal Transport with Applications in Bayesian Inference

In this work, we present two neural network approaches that approximate the solutions of static and dynamic conditional optimal transport (COT) problems. Both approaches enable conditional sampling and conditional density estimation, which are core tasks in Bayesian inference—particularly in the simulation-based (“likelihood-free”) setting. Our methods represent the target conditional distribution as a transformation of a tractable reference distribution. Obtaining such a transformation, chosen here to be an approximation of the COT map, is computationally challenging even in moderate dimensions. To improve scalability, our numerical algorithms use neural networks to parameterize candidate maps and further exploit the structure of the COT problem. Our static approach approximates the map as the gradient of a partially input convex neural network. It uses a novel numerical implementation to increase computational efficiency compared to state-of-the-art alternatives. Our dynamic approach approximates the conditional optimal transport via the flow map of a regularized neural ODE; compared to the static approach, it is slower to train but offers more modeling choices and can lead to faster sampling. We demonstrate both algorithms numerically, comparing them with competing state-of-the-art approaches, using benchmark datasets and simulation-based Bayesian inverse problems.

97 MATHEMATICS AND COMPUTING↗

Target tracking via recursive Bayesian state estimation in cognitive radar networks

To cope with complicated environments and stealthier targets, incorporating intelligence and cognition cycles into target tracking is of great importance in modern sensor network management. With remarkable advances in sensor techniques and deployable platforms, a sensing system has freedom to select a subset of available radars, plan their trajectories, and transmit designed waveforms. In this paper, we propose a general framework for single target tracking in cognitive networks of radars, including consideration of waveform design, path planning, and radar selection, which are separately but not jointly taken into account in existing work. The tracking procedure, built on the theories of dynamic graphical models (DGM) and recursive Bayesian state estimation (RBSE), is formulated as two iterative steps: (i) solving a combinatorial optimization problem to select the optimal subset of radars, waveforms, and locations for the next tracking instant, and (ii) acquiring the recursive Bayesian state estimation to accurately track the target. Further, an illustrative example introduces a specific scenario in 2-D space. Simulation results based on the scenario demonstrate that the proposed framework can accurately track the target under the management of the network of radars.

47 OTHER INSTRUMENTATION↗

Search for the standard model Higgs boson produced in association with a W ± boson with 7.5 fb⁻¹ integrated luminosity at CDF

We present a search for the standard model Higgs boson produced in association with a W ± boson. This search uses data corresponding to an integrated luminosity of 7.5 fb⁻¹ collected by the CDF detector at the Tevatron. We select WH→lνbb¯ candidate events with two jets, large missing transverse energy, and exactly one charged lepton. We further require that at least one jet be identified to originate from a bottom quark. Discrimination between the signal and the large background is achieved through the use of a Bayesian artificial neural network. The number of tagged events and their distributions are consistent with the standard model expectations. We observe no evidence for a Higgs boson signal and set 95% C.L. upper limits on the WH production cross section times the branching ratio to decay to bb¯ pairs, σ(pp¯→W ± H)×B(H→bb¯), relative to the rate predicted by the standard model. For the Higgs boson mass range of 100 to 150 GeV/c² we set observed (expected) upper limits from 1.34 (1.83) to 38.8 (23.4). For 115 GeV/c² the upper limit is 3.64 (2.78). The combination of the present search with an independent analysis that selects events with three jets yields more stringent limits ranging from 1.12 (1.79) to 34.4 (21.6) in the same mass range. For 115 and 125 GeV/c² the upper limits are 2.65 (2.60) and 4.36 (3.69), respectively.

Aaltonen, T.↗

Towards Compact Neural Networks via End-to-End Training: A Bayesian Tensor Approach with Automatic Rank Determination

Post-training model compression can reduce the inference costs of deep neural networks, but uncompressed training still consumes enormous hardware resources and energy. To enable low-energy training on edge devices, it is highly desirable to directly train a compact neural network from scratch with a low memory cost. Low-rank tensor decomposition is an effective approach to reduce the memory and computing costs of large neural networks. However, directly training low-rank tensorized neural networks is a very challenging task because it is hard to determine a proper tensor rank a priori, and the tensor rank controls both model complexity and accuracy. Here, this paper presents a novel end-to-end framework for low-rank tensorized training. We first develop a Bayesian model that supports various low-rank tensor formats (e.g., CANDECOMP/PARAFAC, Tucker, tensor-train, and tensor-train matrix) and reduces neural network parameters with automatic rank determination during training. Then we develop a customized Bayesian solver to train large-scale tensorized neural networks. Our training methods shows orders-of-magnitude parameter reduction and little accuracy loss (or even better accuracy) in the experiments. On a very large deep learning recommendation system with over 4.2 ×10 9 model parameters, our method can reduce the parameter number to 1.6 ×10 5 automatically in the training process (i.e., by 2.6 ×10 4 times) while achieving almost the same accuracy. Code is available at https://github.com/colehawkins/bayesian-tensor-rank-determination.

compact neural networks↗