Search NASA⌕ Search

SEARCH · Search NASA

Results for “Distributed training”

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 685 records · Page 38

A semi-supervised learning method to produce explainable radioisotope proportion estimates for NaI-based synthetic and measured gamma spectra

Quantifying the radioactive sources present in gamma spectra is an ever-present and growing national security mission and a time-consuming process for human analysts. While machine learning models exist that are trained to estimate radioisotope proportions in gamma spectra, few address the eventual need to provide explanatory outputs beyond the estimation task. In this work, we develop two machine learning models for a NaI detector measurements: one to perform the estimation task, and the other to characterize the first model’s ability to provide reasonable estimates. To ensure the first model exhibits a behavior that can be characterized by the second model, the first model is trained using a custom, semi-supervised loss function which constrains proportion estimates to be explainable in terms of a spectral reconstruction. The second auxiliary model is an out-of-distribution detection function (a type of meta-model) leveraging the proportion estimates of the first model to identify when a spectrum is sufficiently unique from the training domain and thus is out-of-scope for the model. In demonstrating the efficacy of this approach, we encourage the use of meta-models to better explain ML outputs used in radiation detection and increase trust.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Data-Driven Robust RVFLNs Modeling of a Blast Furnace Iron-Making Process Using Cauchy Distribution Weighted M-Estimation

Optimal operation of a practical blast furnace (BF) ironmaking process depends largely on a good measurement of molten iron quality (MIQ) indices. However, measuring the MIQ online is not feasible using the available techniques. In this paper, a novel data-driven robust modeling is proposed for online estimation of MIQ using improved random vector functional-link networks (RVFLNs). Since the output weights of traditional RVFLNs are obtained by the least squares approach, a robustness problem may occur when the training dataset is contaminated with outliers. This affects the modeling accuracy of RVFLNs. To solve this problem, a Cauchy distribution weighted M-estimation based robust RFVLNs is proposed. Since the weights of different outlier data are properly determined by the Cauchy distribution, their corresponding contribution on modeling can be properly distinguished. Thus robust and better modeling results can be achieved. Moreover, given that the BF is a complex nonlinear system with numerous coupling variables, the data-driven canonical correlation analysis is employed to identify the most influential components from multitudinous factors that affect the MIQ indices to reduce the model dimension. Finally, experiments using industrial data and comparative studies have demonstrated that the obtained model produces a better modeling and estimating accuracy and stronger robustness than other modeling methods.

42 ENGINEERING↗

Asymptotic-state prediction for fast flavor transformation in neutron star mergers

Neutrino flavor instabilities appear to be omnipresent in dense astrophysical environments, thus presenting a challenge to large-scale simulations of core-collapse supernovae and neutron star mergers (NSMs). Subgrid models offer a path forward, but require an accurate determination of the local outcome of such conversion phenomena. Focusing on “fast” instabilities, related to the existence of a crossing between neutrino and antineutrino angular distributions, we consider a range of analytical mixing schemes, including a new, fully three-dimensional one, and also introduce a new machine learning (ML) model. We compare the accuracy of these models with the results of several thousands of local dynamical calculations of neutrino evolution from the conditions extracted from classical NSM simulations. Our ML model shows good overall performance, but struggles to generalize to conditions from a NSM simulation not used for training. The multidimensional analytic model performs and generalizes even better, while other analytic models (which assume axisymmetric neutrino distributions) do not have reliably high performances, as they notably fail as expected to account for effects resulting from strong anisotropies. As a result, the ML and analytic subgrid models extensively tested here are both promising, with different computational requirements and sources of systematic errors.

79 ASTRONOMY AND ASTROPHYSICS↗

Uncertainty quantification of machine learning models to improve streamflow prediction under changing climate and environmental conditions

Machine learning (ML) models, and Long Short-Term Memory (LSTM) networks in particular, have demonstrated remarkable performance in streamflow prediction and are increasingly being used by the hydrological research community. However, most of these applications do not include uncertainty quantification (UQ). ML models are data driven and can suffer from large extrapolation errors when applied to changing climate/environmental conditions. UQ is required to quantify the influence of data noises on model predictions and avoid overconfident projections in extrapolation. In this work, we integrate a novel UQ method, called PI3NN, with LSTM networks for streamflow prediction. PI3NN calculates Prediction Intervals by training 3 Neural Networks. It can precisely quantify the predictive uncertainty caused by the data noise and identify out-of-distribution (OOD) data in a non-stationary condition to avoid overconfident predictions. We apply the PI3NN-LSTM method in the snow-dominant East River Watershed in the western US and in the rain-driven Walker Branch Watershed in the southeastern US. Results indicate that for the prediction data which have similar features as the training data, PI3NN precisely quantifies the predictive uncertainty with the desired confidence level; and for the OOD data where the LSTM network fails to make accurate predictions, PI3NN produces a reasonably large uncertainty indicating that the results are not trustworthy and should avoid overconfidence. PI3NN is computationally efficient, robust in performance, and generalizable to various network structures and data with no distributional assumptions. It can be broadly applied in ML-based hydrological simulations for credible prediction.

54 ENVIRONMENTAL SCIENCES↗

Image feature extraction and galaxy classification: a novel and efficient approach with automated machine learning

ABSTRACT In this work, we explore the possibility of applying machine learning methods designed for 1D problems to the task of galaxy image classification. The algorithms used for image classification typically rely on multiple costly steps, such as the point spread function deconvolution and the training and application of complex Convolutional Neural Networks of thousands or even millions of parameters. In our approach, we extract features from the galaxy images by analysing the elliptical isophotes in their light distribution and collect the information in a sequence. The sequences obtained with this method present definite features allowing a direct distinction between galaxy types. Then, we train and classify the sequences with machine learning algorithms, designed through the platform Modulos AutoML. As a demonstration of this method, we use the second public release of the Dark Energy Survey (DES DR2). We show that we are able to successfully distinguish between early-type and late-type galaxies, for images with signal-to-noise ratio greater than 300. This yields an accuracy of $86{{\ \rm per\ cent}}$ for the early-type galaxies and $93{{\ \rm per\ cent}}$ for the late-type galaxies, which is on par with most contemporary automated image classification approaches. The data dimensionality reduction of our novel method implies a significant lowering in computational cost of classification. In the perspective of future data sets obtained with e.g. Euclid and the Vera Rubin Observatory, this work represents a path towards using a well-tested and widely used platform from industry in efficiently tackling galaxy classification problems at the peta-byte scale.

79 ASTRONOMY AND ASTROPHYSICS↗

Accelerating geostatistical modeling using geostatistics-informed machine Learning

Ordinary Kriging (OK) is a popular geostatistical algorithm for spatial interpolation and estimation. The computational complexity of OK changes quadratically and cubically for memory and speed, respectively, given the number of data. Therefore, it is computationally intensive and also challenging to process a large set of data, especially in three-dimensional (3D) cases. This paper develops a geostatistics-informed machine learning (GIML) model to improve the efficiency of OK by reducing the number of points required to be estimated using OK. Specifically, only a very few of the unknown points are estimated by OK to get the weights and estimations, which are used as the training dataset. Moreover, the governing equations of OK are used to guide our proposed machine learning to better reproduce the spatial distributions. Our results show that the proposed GIML can reduce the computational time of OK by at least one order of magnitude. The effectiveness of the GIML is evaluated and compared using a 2D case. Furthermore, we demonstrate its efficiency and robustness by considering a different number of training samples on various 3D simulation grids.

58 GEOSCIENCES↗

TH-A-9A-01: Active Optical Flow Model: Predicting Voxel-Level Dose Prediction in Spine SBRT

Purpose: To predict voxel-level dose distribution and enable effective evaluation of cord dose sparing in spine SBRT. Methods: We present an active optical flow model (AOFM) to statistically describe cord dose variations and train a predictive model to represent correlations between AOFM and PTV contours. Thirty clinically accepted spine SBRT plans are evenly divided into training and testing datasets. The development of predictive model consists of 1) collecting a sequence of dose maps including PTV and OAR (spinal cord) as well as a set of associated PTV contours adjacent to OAR from the training dataset, 2) classifying data into five groups based on PTV's locations relative to OAR, two “Top”s, “Left”, “Right”, and “Bottom”, 3) randomly selecting a dose map as the reference in each group and applying rigid registration and optical flow deformation to match all other maps to the reference, 4) building AOFM by importing optical flow vectors and dose values into the principal component analysis (PCA), 5) applying another PCA to features of PTV and OAR contours to generate an active shape model (ASM), and 6) computing a linear regression model of correlations between AOFM and ASM.When predicting dose distribution of a new case in the testing dataset, the PTV is first assigned to a group based on its contour characteristics. Contour features are then transformed into ASM's principal coordinates of the selected group. Finally, voxel-level dose distribution is determined by mapping from the ASM space to the AOFM space using the predictive model. Results: The DVHs predicted by the AOFM-based model and those in clinical plans are comparable in training and testing datasets. At 2% volume the dose difference between predicted and clinical plans is 4.2±4.4% and 3.3±3.5% in the training and testing datasets, respectively. Conclusion: The AOFM is effective in predicting voxel-level dose distribution for spine SBRT. Partially supported by NIH/NCI under grant #R21CA161389 and a master research grant by Varian Medical System.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Conditional diffusion machine-learning framework for mapping valence electron distribution from convergent beam electron diffraction

Quantitative convergent beam electron diffraction (CBED) enables determination of aspherical valence electron distributions through refinement of low-order structure factors, which are highly sensitive to chemical bonding and charge density variations. However, conventional quantitative CBED (QCBED) requires solving a highly nonlinear inverse problem with many coupled parameters, and computationally intensive dynamical diffraction calculations, making it time-consuming and difficult to apply to complex systems. More broadly, reconstructing charge density and orbital electron distribution from diffraction data has long been a central challenge in both x-ray and electron crystallography. Here, in this study, we introduce an artificial-intelligence (AI)-based framework that replaces traditional refinement with a data-driven inverse solver. Using a large synthetic CBED dataset generated by Bloch-wave simulations, we train a conditional diffusion model to directly infer crystal structural parameters and multipole density formalism parameters, and hence valence electron distributions, from CBED patterns alone. By learning from forward simulations across realistic parameter space, the model effectively solves the inverse problem. Compared with direct regression approaches, the diffusion-based framework provides posterior parameter distributions for rigorous uncertainty quantification while preserving quantitative fidelity and reducing analysis time by orders of magnitude. By eliminating the need for external single-crystal x-ray diffraction data and complex nonlinear refinement, this approach enables practical, high-throughput, and in situ quantitative CBED, enabling real-time mapping of valence electron distributions and their correlation with functional responses in quantum and energy materials.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

Machine learning method for enforcing variable independence in background estimation with LHC data: ABCDisCoTEC

A novel solution is presented for the problem of estimating the backgrounds of a signal search using observed data while simultaneously maximizing the sensitivity of the search to the signal. The 'ABCD method' provides a reliable framework for background estimation by partitioning events into one signal-enhanced region (A) and three background-enhanced control regions (B, C, and D) via two smoothly varying, statistically independent variables. In practice, even slight correlations between the two variables can significantly undermine the method's performance. Thus, choosing appropriate variables by hand can present a formidable challenge, especially when background and signal differ only subtly. To address this issue, the ABCD with distance correlation (ABCDisCo) method was developed to construct two learned variables via a neural network trained to provide strong signal-background discrimination with small values of the distance correlation (DisCo) measure between the two learned variables. However, relying solely on minimizing the DisCo can result in learned variables that may not have distributions of background events that are smoothly varying and localized at extreme values, as necessary for the validity of the background estimation. The ABCDisCo training enhanced with closure (ABCDisCoTEC) method is introduced to solve this issue by directly minimizing the nonclosure, expressed as a dedicated differentiable loss term. This extended method is applied to a data set of proton-proton collisions at a center-of-mass energy of 13 TeV recorded by the CMS detector at the CERN Large Hadron Collider. Additionally, given the complexity of the minimization problem with constraints on multiple loss terms, the modified differential method of multipliers is applied and shown to greatly improve the stability and robustness of the ABCDisCoTEC method, compared to grid search hyperparameter optimization procedures.

Hayrapetyan, Aram [Yerevan Phys. Inst.]↗

Linear discriminant analysis with misallocation in training samples

Linear discriminant analysis for a two-class case is studied in the presence of misallocation in training samples. A general appraoch to modeling of mislocation is formulated, and the mean vectors and covariance matrices of the mixture distributions are derived. The asymptotic distribution of the discriminant boundary is obtained and the asymptotic first two moments of the two types of error rate given. Certain numerical results for the error rates are presented by considering the random and two non-random misallocation models. It is shown that when the allocation procedure for training samples is objectively formulated, the effect of misallocation on the error rates of the Bayes linear discriminant rule can almost be eliminated. If, however, this is not possible, the use of Fisher rule may be preferred over the Bayes rule.

Chhikara, R.↗

Naturalistic Decision Making for Power System Operators

Motivation – Investigations of large-scale outages in the North American interconnected electric system often attribute the causes to three T’s: Trees, Training and Tools. To document and understand the mental processes used by expert operators when making critical decisions, a naturalistic decision making (NDM) model was developed. Transcripts of conversations were analyzed to reveal and assess NDM-based performance criteria. Findings/Design – An item analysis indicated that the operators’ Situation Awareness Levels, mental models, and mental simulations can be mapped at different points in the training scenario. This may identify improved training methods or analytical/ visualization tools. Originality/Value – This study applies for the first time, the concepts of Recognition Primed Decision Making, Situation Awareness Levels and Cognitive Task Analysis to training of electric power system operators. Take away message – The NDM approach provides a viable framework for systematic training management to accelerate learning in simulator-based training scenarios for power system operators and teams.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Labels as a feature: Network homophily for systematically annotating human GPCR drug-target interactions

Machine learning has revolutionized drug discovery by enabling the exploration of vast, uncharted chemical spaces essential for discovering novel patentable drugs. Despite the critical role of human G protein-coupled receptors in FDA-approved drugs, exhaustive in-distribution drug-target interaction testing across all pairs of human G protein-coupled receptors and known drugs is rare due to significant economic and technical challenges. This often leaves off-target effects unexplored, which poses a considerable risk to drug safety. In contrast to the traditional focus on out-of-distribution exploration (drug discovery), we introduce a neighborhood-to-prediction model termed Chemical Space Neural Networks that leverages network homophily and training-free graph neural networks with labels as features. We show that Chemical Space Neural Networks’ ability to make accurate predictions strongly correlates with network homophily. Thus, labels as features strongly increase a machine learning model’s capacity to enhance in-distribution prediction accuracy, which we show by integrating labeled data during inference. We validate these advancements in a high-throughput yeast biosensing system (3773 drug-target interactions, 539 compounds, 7 human G protein-coupled receptors) to discover novel drug-target interactions for FDA-approved drugs and to expand the general understanding of how to build reliable predictors to guide experimental verification.

Hansson, Frederik G↗

Power Plant Testing of Ultrasonic Measurements of Temperature Distributions and Heat Fluxes to Heat Exchange Surfaces

Extreme environmental conditions are common in energy conversion applications, in which the most hardened insertion sensors often do not perform reliably for long. On the other hand, ultrasonic measurements can be acquired noninvasively, with sensitive components kept away from the damaging environment. We have previously developed the ultrasonic method for measuring the spatial distribution of temperatures in solid materials applicable when large thermal gradients are present. In the developed approach, we use the echogenically segmented ultrasound propagation path, structured to contain engineered or naturally occurring echogenic features, to produce a train of echoes in response to an external pulse of ultrasonic excitation. This paper outlines the testing results of this approach applied to the temperature distribution measurements to a waterwall of a 500 MW power-plant utility boiler. The validation results obtained over the prolonged power plant test show that the estimated temperature profile is correctly captured, and the measurement accuracy can be comparable with traditional insertion sensors, such as thermocouples. Overall, the testing has confirmed that the developed approach has matured to become an attractive alternative to conventional sensing in solving challenging problems of long-term temperature measurements in extreme environments. Heat fluxes and thermal stresses in the structure can then be characterized noninvasively using the measured temperature distribution as the basis.

Walton, Kenneth↗

Scalable deep text comprehension for Cancer surveillance on high-performance computing

Background: Deep Learning (DL) has advanced the state-of-the-art capabilities in bioinformatics applications which has resulted in trends of increasingly sophisticated and computationally demanding models trained by larger and larger data sets. This vastly increased computational demand challenges the feasibility of conducting cutting-edge research. One solution is to distribute the vast computational workload across multiple computing cluster nodes with data parallelism algorithms. In this study, we used a High-Performance Computing environment and implemented the Downpour Stochastic Gradient Descent algorithm for data parallelism to train a Convolutional Neural Network (CNN) for the natural language processing task of information extraction from a massive dataset of cancer pathology reports. We evaluated the scalability improvements using data parallelism training and the Titan supercomputer at Oak Ridge Leadership Computing Facility. To evaluate scalability, we used different numbers of worker nodes and performed a set of experiments comparing the effects of different training batch sizes and optimizer functions.Results: We found that Adadelta would consistently converge at a lower validation loss, though requiring over twice as many training epochs as the fastest converging optimizer, RMSProp. The Adam optimizer consistently achieved a close 2nd place minimum validation loss significantly faster; using a batch size of 16 and 32 allowed the network to converge in only 4.5 training epochs.Conclusions: We demonstrated that the networked training process is scalable across multiple compute nodes communicating with message passing interface while achieving higher classification accuracy compared to a traditional machine learning algorithm.

Qiu, John X.↗

Modeling Ring Current Proton Fluxes Using Artificial Neural Network and Van Allen Probe Measurements

Abstract Terrestrial ring current dynamics are a critical part of the near‐space environment, in that they directly drive geomagnetic field variations that control particle drifts, and define geomagnetic storms. The present study aims to specify a global and time‐varying distribution of ring current proton using geomagnetic indices and solar wind parameters with their history as input. We train an artificial neural network (ANN) model to reproduce proton fluxes measured by the Radiation Belt Storm Probes Ion Composition Experiment instrument onboard Van Allen Probes. By choosing optimal feature parameters and their history length, the model results show a high correlation and a small error between model specifications and satellite measurements. The modeled results well capture energy‐dependent proton dynamics in association with geomagnetic storms, including inward radial diffusion, acceleration and decay. Our ANN model produces proton fluxes with their corresponding 3D spatiotemporal variations, capturing the latitudinal distribution and local time asymmetry that are consistent with observations and that can further inform theory.

Li, Jinxing↗

A Path Towards Quantum Advantage in Training Deep Generative Models with Quantum Annealing

A class of quantum-classical hybrid machine-learning algorithms can be obtained by integrating classical deep generative models with quantum probability distributions as 'priors' over their latent variables. We introduce a hybrid implementation of variational autoencoders (QVAE) and also present a technique to hybridize flow-based invertible generative models. We demonstrate the use of D-Wave quantum annealers as physical simulators of quantum Boltzmann machines (QBM) to perform quantum-assisted training of QVAE. Latent-space QBM develop slowly mixing modes, opening a path to obtain quantum advantage in generative modeling with available quantum devices.

Vinci, Walter↗

Promoting A-Priori Interoperability of HLA-Based Simulations in the Space Domain: The SISO Space Reference FOM Initiative

Distributed and Real-Time Simulation plays a key-role in the Space domain being exploited for missions and systems analysis and engineering as well as for crew training and operational support. One of the most popular standards is the 1516-2010 IEEE Standard for Modeling and Simulation (M&S) High Level Architecture (HLA). HLA supports the implementation of distributed simulations (called Federations) in which a set of simulation entities (called Federates) can interact using a Run-Time Infrastructure (RTI). In a given Federation, a Federate can publish and/or subscribes objects and interactions on the RTI only in accordance with their structures as defined in a FOM (Federation Object Model). Currently, the Space domain is characterized by a set of incompatible FOMs that, although meet the specific needs of different organizations and projects, increases the long-term cost for interoperability. In this context, the availability of a reference FOM for the Space domain will enable the development of interoperable HLA-based simulators for related joint projects and collaborations among worldwide organizations involved in the Space domain (e.g. NASA, ESA, Roscosmos, and JAXA). The paper presents a first set of results achieved by a SISO standardization effort that aims at providing a Space Reference FOM for international collaboration on Space systems simulations.

Moller, Bjorn↗