Search NASA⌕ Search

SEARCH · Search NASA

Results for “inference methods”

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 1,351 records · Page 75
All sourcesNASA NTRSDOE OSTI
NASA NTRS1977

Nonplanar wing load-line and slender wing theory

Nonplanar load line, slender wing, elliptic wing, and infinite aspect ratio limit loading theories are developed. These are quasi two dimensional theories but satisfy wing boundary conditions at all points along the nonplanar spanwise extent of the wing. These methods are applicable for generalized configurations such as the laterally nonplanar wing, multiple nonplanar wings, or wing with multiple winglets of arbitrary shape. Two dimensional theory infers simplicity which is practical when analyzing complicated configurations. The lateral spanwise distribution of angle of attack can be that due to winglet or control surface deflection, wing twist, or induced angles due to multiwings, multiwinglets, ground, walls, jet or fuselage. In quasi two dimensional theory the induced angles due to these extra conditions are likewise determined for two dimensional flow. Equations are developed for the normal to surface induced velocity due to a nonplanar trailing vorticity distribution. Application examples are made using these methods.

Deyoung, J.↗
DOE OSTI2025

Machine Learning–Guided Boolean Matrix Inference for Real-Time O-RAN Conflict Detection

Open Radio Access Networks (O-RAN) are emerging, software-driven cellular architectures that promote flexibility by enabling components from different vendors to interoperate. Multiple control applications called xApps can independently adjust network parameters in near real time, often without awareness of each other's actions. This creates a system highly prone to unintended conflicts and performance degradation due to the inherent complexity of such openness. To model such systems and ultimately prevent or mitigate xApp conflicts, it is essential to understand the dynamic relationships between xApps (A), the control parameters they adjust (P), and the resulting KPI responses (K). While the mappings from A to P and from K to A can often be derived from xApp specifications, the relationship from P to K is typically hidden within the system’s dynamics and must be inferred from observed data. We propose a novel data-driven Boolean inference framework that uncovers the hidden P?K dependencies using machine learning and interpretable rule induction. Continuous parameters and KPIs are first binarized using decision tree classifiers, and a binary influence matrix L is then inferred by solving Boolean matrix equations over time. This compact representation improves interpretability and enables real-time tracking of dynamically evolving parameter-KPI dependencies. We demonstrate the effectiveness of our method in a realistic mobile handover scenario, where it accurately recovers the underlying logic and enables proactive conflict detection.

42 - ENGINEERING↗
DOE OSTI2022

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↗
DOE OSTI2025

Simulation-Based Inference for Neutrino Interaction Model Parameter Tuning

High-energy physics experiments studying neutrinos rely heavily on simulations of their interactions with atomic nuclei. Limitations in the theoretical understanding of these interactions typically necessitate ad hoc tuning of simulation model parameters to data. Traditional tuning methods for neutrino experiments have largely relied on simple algorithms for numerical optimization. While adequate for the modest goals of initial efforts, the complexity of future neutrino tuning campaigns is expected to increase substantially, and new approaches will be needed to make progress. In this paper, we examine the application of simulation-based inference (SBI) to the neutrino interaction model tuning for the first time. Using a previous tuning study performed by the MicroBooNE experiment as a test case, we find that our SBI algorithm can correctly infer the tuned parameter values when confronted with a mock data set generated according to the MicroBooNE procedure. This initial proof-of-principle illustrates a promising new technique for next-generation simulation tuning campaigns for the neutrino experimental community.

Tame-Narvaez, Karla Maria [Fermilab]↗
DOE OSTI2025

Simulation-based inference for neutrino interaction model parameter tuning

High-energy physics experiments studying neutrinos rely heavily on simulations of their interactions with atomic nuclei. Limitations in the theoretical understanding of these interactions typically necessitate ad hoc tuning of simulation model parameters to data. Traditional tuning methods for neutrino experiments have largely relied on simple algorithms for numerical optimization. While adequate for the modest goals of initial efforts, the complexity of future neutrino tuning campaigns is expected to increase substantially, and new approaches will be needed to make progress. In this paper, we examine the application of simulation-based inference (SBI) to the neutrino interaction model tuning for the first time. Using a previous tuning study performed by the MicroBooNE experiment as a test case, we find that our SBI algorithm can correctly infer the tuned parameter values when confronted with a mock data set generated according to the MicroBooNE procedure. This initial proof-of-principle illustrates a promising new technique for next-generation simulation tuning campaigns for the neutrino experimental community.

Tame-Narvaez, Karla [Fermilab] (ORCID:000000022249↗
DOE OSTI2022

Gantryless Associated-Particle Neutron Radiography

The present work reports on the development of techniques for in-field fast-neutron radiography measurements using the associated-particle imaging (API) method. The API method employs alpha-neutron coincidences from the d+t→α+n reaction to enable fast neutron transmission imaging with excellent contrast using a wide cone beam. However, for field radiography applications, the API method is burdened by the need for the relative positions of the source and detector to be known. Fortunately, these relative positions can be inferred from transmission data. The inferred positions also enable accurate stitching of multiple images into a composite image even when using a low-resolution detector panel and acquiring images having few overlapping pixels. The developed techniques address analysis of measurements where (1) the source and detector panel are separately hand positioned rather than held in registration by a gantry, (2) multiple detector panel positions within the “coincident cone” of tagged neutrons are required to piece together an image of an item of interest, and (3) normalization measurements that have identical source-detector positioning but without the inspected object are not possible. The present work will describe the system calibrations (including timing calibrations and neutron direction calibrations) necessary for subsequent analyses, the method of locating the detector in the coincident cone of neutrons with millimeter precision using the timing and directions of coincident neutrons, the method of calculating a normalization image for a given detector panel position, and the method used to project multiple images into a common image.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2018

The Fifth International Workshop on Ice Nucleation phase 2 (FIN-02): laboratory intercomparison of ice nucleation measurements

Abstract. The second phase of the Fifth International Ice Nucleation Workshop (FIN-02)involved the gathering of a large number of researchers at the KarlsruheInstitute of Technology's Aerosol Interactions and Dynamics of the Atmosphere(AIDA) facility to promote characterization and understanding of icenucleation measurements made by a variety of methods used worldwide.Compared to the previous workshop in 2007, participation was doubled,reflecting a vibrant research area. Experimental methods involved sampling ofaerosol particles by direct processing ice nucleation measuring systems fromthe same volume of air in separate experiments using different ice nucleatingparticle (INP) types, and collections of aerosol particle samples ontofilters or into liquid for sharing amongst measurement techniques thatpost-process these samples. In this manner, any errors introduced bydifferences in generation methods when samples are shared across laboratorieswere mitigated. Furthermore, as much as possible, aerosol particle sizedistribution was controlled so that the size limitations of different methodswere minimized. The results presented here use data from the workshop toassess the comparability of immersion freezing measurement methods activatingINPs in bulk suspensions, methods that activate INPs in condensation and/orimmersion freezing modes as single particles on a substrate, continuous flowdiffusion chambers (CFDCs) directly sampling and processing particles wellabove water saturation to maximize immersion and subsequent freezing ofaerosol particles, and expansion cloud chamber simulations in which liquidcloud droplets were first activated on aerosol particles prior to freezing.The AIDA expansion chamber measurements are expected to be the closestrepresentation to INP activation in atmospheric cloud parcels in thesecomparisons, due to exposing particles freely to adiabatic cooling. The different particle types used as INPs included the minerals illite NX andpotassium feldspar (K-feldspar), two natural soil dusts representative of arable sandy loam(Argentina) and highly erodible sandy dryland (Tunisia) soils, respectively,and a bacterial INP (Snomax ® ). Consideredtogether, the agreement among post-processed immersion freezing measurementsof the numbers and fractions of particles active at different temperaturesfollowing bulk collection of particles into liquid was excellent, withpossible temperature uncertainties inferred to be a key factor in determiningINP uncertainties. Collection onto filters for rinsing versus directly intoliquid in impingers made little difference. For methods that activatedcollected single particles on a substrate at a controlled humidity at orabove water saturation, agreement with immersion freezing methods was good inmost cases, but was biased low in a few others for reasons that have not beenresolved, but could relate to water vapor competition effects. AmongstCFDC-style instruments, various factors requiring (variable) highersupersaturations to achieve equivalent immersion freezing activation dominatethe uncertainty between these measurements, and for comparison with bulkimmersion freezing methods. When operated above water saturation to includeassessment of immersion freezing, CFDC measurements often measured at orabove the upper bound of immersion freezing device measurements, but oftenunderestimated INP concentration in comparison to an immersion freezingmethod that first activates all particles into liquid droplets prior tocooling (the PIMCA-PINC device, or Portable Immersion Mode Cooling chAmber–Portable Ice Nucleation Chamber), and typically slightly underestimated INPnumber concentrations in comparison to cloud parcel expansions in the AIDAchamber; this can be largely mitigated when it is possible to raise therelative humidity to sufficiently high values in the CFDCs, although this isnot always possible operationally. Correspondence of measurements of INPs among direct sampling andpost-processing systems varied depending on the INP type. Agreement was bestfor Snomax ® particles in the temperature regimecolder than −10 ∘ C, where their ice nucleation activity is nearlymaximized and changes very little with temperature. At temperatures warmer than −10 ∘ C, Snomax ® INP measurements (allvia freezing of suspensions) demonstrated discrepancies consistent withprevious reports of the instability of its protein aggregates that appear tomake it less suitable as a calibration INP at these temperatures. ForArgentinian soil dust particles, there was excellent agreement across allmeasurement methods; measures ranged within 1 order of magnitude for INPnumber concentrations, active fractions and calculated active site densitiesover a 25 to 30 ∘ C range and 5 to 8 orders of correspondingmagnitude change in number concentrations. This was also the case for alltemperatures warmer than −25 ∘ C in Tunisian dust experiments. Incontrast, discrepancies in measurements of INP concentrations or active sitedensities that exceeded 2 orders of magnitude across a broad range of temperaturemeasurements found at temperatures warmer than −25 ∘ C in a previous study werereplicated for illite NX. Discrepancies also exceeded 2 orders of magnitude attemperatures of −20 to −25 ∘ C for potassium feldspar (K-feldspar), but these coincidedwith the range of temperatures at which INP concentrations increase rapidly atapproximately an order of magnitude per 2 ∘ C cooling forK-feldspar. These few discrepancies did not outweigh the overall positive outcomes of theworkshop activity, nor the future utility of this data set or future similarefforts for resolving remaining measurement issues. Measurements of the samematerials were repeatable over the time of the workshop and demonstratedstrong consistency with prior studies, as reflected by agreement of databroadly with parameterizations of different specific or general (e.g., soildust) aerosol types. The divergent measurements of the INP activity of illiteNX by direct versus post-processing methods were not repeated for otherparticle types, and the Snomax ® datademonstrated that, at least for a biological INP type, there is no expectedmeasurement bias between bulk collection and direct immediately processedfreezing methods to as warm as −10 ∘ C. Since particle size rangeswere limited for this workshop, it can be expected that for atmosphericpopulations of INPs, measurement discrepancies will appear due to thedifferent capabilities of methods for sampling the full aerosol sizedistribution, or due to limitations on achieving sufficient watersupersaturations to fully capture immersion freezing in direct processinginstruments. Overall, this workshop presents an improved picture of presentcapabilities for measuring INPs than in past workshops, and providesdirection toward addressing remaining measurement issues.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2017

On seasonal and semi-annual approach for flood frequency analysis

As a supplementary method to the conventional flood frequency analysis based on annual peak flows, we propose an approach in this paper to infer the flood frequency distribution on quarterly and semi-annual time scale, which are then converted to annual time scale to obtain the floods corresponding to return periods in unit of year. Two criteria for test of data independence, namely, minimum 7 and 15-day interval between two consecutive peak flows, are tested. The proposed approach was applied to Des Moines River at Fort Dodge, Iowa, USA using its 62 years of observation daily flows. The results show that the estimated floods for given return periods from quarter-annual data series are in general higher than the corresponding estimated floods from semi-annual data series, which is further larger than estimated floods from annual peak flows. In conclusion, the floods estimated from semi-annual data series agree well with the results of previous US Geological Survey study.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2025

Jensen–Shannon divergence based novel loss functions for Bayesian neural networks

Bayesian neural networks (BNNs) are state-of-the-art machine learning methods that can naturally regularize and systematically quantify uncertainties using their stochastic parameters. Kullback–Leibler (KL) divergence-based variational inference used in BNNs suffer from unstable optimization and challenges in approximating light-tailed posteriors due to the unbounded nature of the KL divergence. To resolve these issues, we formulate a novel loss function for BNNs based on a new modification to the generalized Jensen–Shannon (JS) divergence, which is bounded. In addition, we propose a Geometric JS divergence-based loss, which is computationally efficient since it can be evaluated analytically. We found that the JS divergence-based variational inference is intractable, and hence employed a constrained optimization framework to formulate these losses. Our theoretical analysis and empirical experiments on multiple regression and classification data sets suggest that the proposed losses perform better than the KL divergence-based loss, especially when the data sets are noisy or biased. Specifically, there are approximately 5% and 8% improvements in accuracy for a noise-added CIFAR-10 dataset and a regression dataset, respectively. There is about 13% reduction in false negative predictions of a biased histopathology dataset. Additionally, we quantify and compare the uncertainty metrics for the regression and classification tasks.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2025

Emulation With Uncertainty Quantification of Regional Sea‐Level Change Caused by the Antarctic Ice Sheet

Abstract Projecting regional sea‐level change under various climate‐change scenarios typically involves running forward simulations of the Earth's gravitational, rotational and deformational (GRD) response to ice‐mass change, which requires substantial computational cost if applied to probabilistic frameworks requiring thousands to millions of samples. Here we build emulators of regional sea‐level change at 27 coastal locations, due to the GRD effects associated with future Antarctic Ice Sheet mass change over the 21st century. The emulators are evaluated against a numerical sea‐level model applied to an ensemble of ice‐sheet model simulations of the Antarctic Ice Sheet through 2100. We build a physics‐based emulator using a recent sensitivity kernel approach and compare it to machine learning based emulators (neural network and conditional variational autoencoder methods). In order to quantify uncertainty, we derive well‐calibrated prediction intervals for regional sea‐level change via split‐conformal inference and linear regression, and show that Monte Carlo dropout does not yield well‐calibrated uncertainties in this instance. We also demonstrate substantial gains in computational efficiency using both the physics‐based emulator and neural networks in comparison to the numerical model for the complete regional sea‐level solution. Overall, we find the physics‐based emulator modestly outperforms the machine learning emulators for this problem.

58 GEOSCIENCES↗
DOE OSTI2013

Inner heliospheric evolution of a 'STEALTH' CME derived from multi-view imaging and multipoint in situ observations. I. Propagation to 1 AU

Coronal mass ejections (CMEs) are the main driver of space weather. Therefore, a precise forecasting of their likely geo-effectiveness relies on an accurate tracking of their morphological and kinematical evolution throughout the interplanetary medium. However, single viewpoint observations require many assumptions to model the development of the features of CMEs. The most common hypotheses were those of radial propagation and self-similar expansion. The use of different viewpoints shows that, at least for some cases, those assumptions are no longer valid. From radial propagation, typical attributes that can now be confirmed to exist are over-expansion and/or rotation along the propagation axis. Understanding the 3D development and evolution of the CME features will help to establish the connection between remote and in situ observations, and hence help forecast space weather. We present an analysis of the morphological and kinematical evolution of a STEREO-B-directed CME on 2009 August 25-27. By means of a comprehensive analysis of remote imaging observations provided by the SOHO, STEREO, and SDO missions, and in situ measurements recorded by Wind, ACE, and MESSENGER, we prove in this paper that the event exhibits signatures of deflection, which are usually associated with changes in the direction of propagation and/or also with rotation. The interaction with other magnetic obstacles could act as a catalyst of deflection or rotation effects. We also propose a method to investigate the change of the CME tilt from the analysis of height-time direct measurements. If this method is validated in further work, it may have important implications for space weather studies because it will allow for inference of the interplanetary counterpart of the CME's orientation.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2014

Evaluating gyrochronology on the zero-age-main-sequence: rotation periods in the southern open cluster Blanco 1 from the Kelt-South survey

We report periods for 33 members of Blanco 1 as measured from Kilodegree Extremely Little Telescope-South light curves, the first reported rotation periods for this benchmark zero-age-main-sequence open cluster. The distribution of these stars spans from late-A or early-F dwarfs to mid-K with periods ranging from less than a day to ∼8 days. The rotation period distribution has a morphology similar to the coeval Pleiades cluster, suggesting the universal nature of stellar rotation distributions. Employing two different gyrochronology methods, we find an age of 146{sub −14}{sup +13} Myr for the cluster. Using the same techniques, we infer an age of 134{sub −10}{sup +9} Myr for the Pleiades measured from existing literature rotation periods. These rotation-derived ages agree with independently determined cluster ages based on the lithium depletion boundary technique. Additionally, we evaluate different gyrochronology models and quantify levels of agreement between the models and the Blanco 1/Pleiades rotation period distributions, including incorporating the rotation distributions of clusters at ages up to 1.1 Gyr. We find the Skumanich-like spin-down rate sufficiently describes the rotation evolution of stars hotter than the Sun; however, we find cooler stars rotating faster than predicted by a Skumanich law, suggesting a mass dependence in the efficiency of stellar angular momentum loss rate. Finally, we compare the Blanco 1 and Pleiades rotation period distributions to available nonlinear angular momentum evolution models. We find they require a significant mass dependence on the initial rotation rate of solar-type stars to reproduce the observed range of rotation periods at a given stellar mass and are furthermore unable to predict the observed over-density of stars along the upper envelope of the clusters' rotation distributions.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2015

X-ray perspective of the twisted magnetospheres of magnetars

Anomalous X-ray pulsars (AXPs) and soft gamma-ray repeaters (SGRs) are recognized as the most promising magnetar candidates, as indicated by their energetic bursts and rapid spin-downs. It is expected that the strong magnetic field leaves distinctive imprints on the emergent radiation both by affecting the radiative processes in atmospheres of magnetars and by scattering in the upper magnetospheres. We construct a self-consistent physical model that incorporates emission from the magnetar surface and its reprocessing in the three-dimensional (3D) twisted magnetosphere using a Monte Carlo technique. The synthetic spectra are characterized by four parameters: surface temperature kT, surface magnetic field strength B, magnetospheric twist angle Δϕ, and the normalized electron velocity β. We also create a tabular model (STEMS3D) and apply it to a large sample of XMM-Newton spectra of magnetars. The model successfully fits nearly all spectra, and the obtained magnetic field for 7 out of the 11 sources are consistent with the values inferred from the spin-down rates. We conclude that the continuum-fitting by our model is a robust method to measure the magnetic field strength and magnetospheric configuration of AXPs and SGRs. Investigating the multiple observations of variable sources, we also study the mechanism of their spectral evolution. Our results suggest that the magnetospheres in these sources are highly twisted (Δϕ>1), and the behavior of magnetospheric twisting and untwisting is revealed in the 2002 outburst of 1E 2259+586.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2017

Simultaneous inversion of multiple microseismic data for event locations and velocity model with Bayesian inference

We have applied Bayesian inference for simultaneous inversion of multiple microseismic data to obtain event locations along with the subsurface velocity model. The traditional method of using a predetermined velocity model for event location may be subject to large uncertainties, particularly if the prior velocity model is poor. Our study indicated that microseismic data can help to construct the velocity model, which is usually a major source of uncertainty in microseismic event locations. The simultaneous inversion eliminates the requirement for an accurate predetermined velocity model in microseismic event location estimation.Here, we estimate the posterior probability density of the velocity model and microseismic event locations with the maximum a posteriori estimation, and the posterior covariance approximation under the Gaussian assumption. This provides an efficient and effective way to quantify the uncertainty of the microseismic location estimation and capture the correlation between the velocity model and microseismic event locations. We have developed successful applications on both synthetic examples and real data from the Newberry enhanced geothermal system. Comparisons with location results based on a traditional predetermined velocity model method demonstrated that we can construct a reliable effective velocity model using only microseismic data and determine microseismic event locations without prior knowledge of the velocity model.

58 GEOSCIENCES↗
DOE OSTI2025

Designing an Optimal Sensor Network via Minimizing Information Loss

Optimal experimental design is a classic topic in statistics, with many well-studied problems, applications, and solutions. The design problem we study is the placement of sensors to monitor spatiotemporal processes, explicitly accounting for the temporal dimension in our modeling and optimization. We observe that recent advancements in computational sciences often yield large datasets based on physics-based simulations, which are rarely leveraged in experimental design. We introduce a novel model-based sensor placement criterion, along with a highly-efficient optimization algorithm, which integrates physics-based simulations and Bayesian experimental design principles to identify sensor networks that “minimize information loss” from simulated data. Our technique relies on sparse variational inference and (separable) Gauss-Markov priors, and thus may adapt many techniques from Bayesian experimental design. We validate our method through a case study monitoring air temperature in Phoenix, Arizona, using state-of-the-art physics-based simulations. Our results show our framework to be superior to random or quasi-random sampling, particularly with a limited number of sensors. We conclude by discussing practical considerations and implications of our framework, including more complex modeling tools and real-world deployments.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2019

Density-driven ionization potential depression and its effects on material properties

Changes in ionization potential at high density drive changes in material transport properties and often in crystal structure and equation of state. Current models used to describe and predict these effects have not been tested in regimes much beyond ~200 GPa, because of limitations in experimental techniques. Now, however, high-energy laser facilities with flexible pulse-shaping capabilities have made it possible to reach pressures in the 1000 GPa range, while maintaining temperatures below 1 eV. We have used these new methods to ramp-compress cobalt and magnesium and have measured shifts in the k-shell emission lines to infer changes in ionization stage, and x-ray diffraction to measure changes in crystal structure. In the case of cobalt, commonly-used detailed models for effective ionization based on isolated-atom electronic structure are shown to give the wrong answer, while self-consistent density-functional-theory-based models reproduce the data well. In the case of magnesium, zero-kelvin density functional theory predictions of density-driven electron localization in lattice interstitials (so-called electride phases) at high pressure are found to be too simple to describe material evolution along a laser ramp-compression pathway. Instead, by a combination of experiment, first-principles theory and evolutionary structure-searching algorithms, we have discovered several unexpected lower-symmetry metastable structures. First principles results find that the density of localized interstitial charge in the high-pressure magnesium phases is too low to be measurable with x-ray diffraction, and the body of theory work which has predicted high pressure electride phases for many systems, based on analysis of electron localization function, needs to be re-examined.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗
DOE OSTI2022

Optimal vocabulary selection approaches for privacy-preserving deep NLP model training for information extraction and cancer epidemiology

With the use of artificial intelligence and machine learning techniques for biomedical informatics, security and privacy concerns over the data and subject identities have also become an important issue and essential research topic. Without intentional safeguards, machine learning models may find patterns and features to improve task performance that are associated with private personal information. The privacy vulnerability of deep learning models for information extraction from medical textural contents needs to be quantified since the models are exposed to private health information and personally identifiable information. The objective of the study is to quantify the privacy vulnerability of the deep learning models for natural language processing and explore a proper way of securing patients’ information to mitigate confidentiality breaches. The target model is the multitask convolutional neural network for information extraction from cancer pathology reports, where the data for training the model are from multiple state population-based cancer registries. This study proposes the following schemes to collect vocabularies from the cancer pathology reports; (a) words appearing in multiple registries, and (b) words that have higher mutual information. We performed membership inference attacks on the models in high-performance computing environments. The comparison outcomes suggest that the proposed vocabulary selection methods resulted in lower privacy vulnerability while maintaining the same level of clinical task performance.

59 BASIC BIOLOGICAL SCIENCES↗
NASA NTRS1973

Imaging radars provide terrain texture and roughness parameters in semi-arid environments

Geoscience potentialities and limitations of available imaging radars are examined for alluvial valley-playa terrain environments. In such arid environments, a dual-sensor combination of an imaging radar with aerial photography is shown to provide a practical method for monitoring gross changes in surface textures of alluvial fans and playa surface conditions. It also appears feasible to infer surface materials and relative surface roughness with an improved degree of interpretive reliability.

Macdonald, H. C.↗