Search NASA⌕ Search

SEARCH · Search NASA

Results for “distribution shift”

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 19 records
All sourcesNASA NTRSDOE OSTI
DOE OSTI2024

Assessing Membership Inference Attacks under Distribution Shifts

Membership inference attacks (MIAs) exploit machine learning models to infer whether a data point was in the training set, posing significant privacy risks even with limited black-box access. These attacks rely on the attacker approximating the target model’s training distribution, yet the impact of distribution shifts between target and shadow models on MIA success remains underexplored. We systematically evaluate five types of distribution shifts —-cutout, jitter, Gaussian noise, label shift, and attribute shift —- at varying intensities. Our results reveal that these shifts affect MIA effectiveness in nuanced ways, with some reducing attack success while others exacerbate vulnerabilities, and the same shift can have opposite effects depending on the type of MIA. This highlights the complex interplay between distributional differences and attack performance, offering critical insights for improving model defenses against MIAs.

Shi, Yichuan [Massachusetts Institute of Technolog↗
DOE OSTI2025

Data-Conforming Data-Driven Control: Avoiding Premature Generalizations Beyond Data

Data-driven and adaptive control approaches face the problem of introducing sudden distributional shifts beyond the distribution of data encountered during learning. Therefore, they are prone to invalidating the very assumptions used in their own construction. This is due to the linearity of the underlying system, inherently assumed and formulated in most data-driven control approaches, which may falsely generalize the behavior of the system beyond the behavior experienced in the data. This article seeks to mitigate these problems by enforcing consistency of the newly designed closed-loop systems with data and slowing down any distributional shifts in the joint state-input space. This is achieved through incorporating affine regularization terms and linear matrix inequality constraints to data-driven approaches, resulting in convex semi-definite programs that can be efficiently solved by standard software packages. We discuss the optimality conditions of these programs and then conclude this article with a numerical example that further highlights the problem of premature generalization beyond data and shows the effectiveness of our proposed approaches in enhancing the safety of data-driven control methods.

97 MATHEMATICS AND COMPUTING↗
NASA NTRS1969

Distribution of redshifts of quasars.

Quasars red shifts distribution interpreted as due to cosmological and gravitational red shifts distribution, noting analysis error

Patterson, T. N. L.↗
NASA NTRS1982

Combustion-related pollutants of polydisperse single-composition aerosols and advection fog formation

The most noticeable effect of air pollution on the properties of the atmosphere is the reduction in visibility, with and without the occurrence of condensation, which frequently accompanies polluted air. The present study concerns the formation of advection fog associated with aerosols, due to combustion-related pollutants, with a polydisperse population distribution and a single composition model. The results show that an aerosol population with high particle concentration-shifted distribution provides a more favorable condition for the formation of dense fog than an aerosol population with a low particle concentration-shifted distribution if the value of the mass concentration of the aerosols is kept constant.

Hung, R. J.↗
NASA NTRS1999

The Effects of LOX Post Biasing on SSME Injector Wall Compatibility

An experimental investigation has been carried out to examine the effects of LOX post biasing of a shear coaxial injector on the behavior of the spray near a chamber wall. The experimental work was performed with inert propellant simulants in a high-pressure chamber. Injector flow rates and chamber pressure were designed to match the Space Shuttle Main Engine (SSME) injector gas-to-liquid density and velocity ratio at the point of propellant injection. Measurements of liquid mass flux, gas phase velocity and droplet size were made using mechanical patternation and phase Doppler interferometry techniques. The measurements revealed that the liquid mass flux distribution shifts away from the wall with increasing LOX post bias away from the wall. The shift in the liquid flux distribution was much greater than that caused by the angling of the LOX post alone. Gas velocity near the wall simultaneously increased with increasing LOX post bias away from the wall. The increase in wall side gas velocity was due to the higher fraction of gas injected on the wall side of the injector as a result of the eccentricity at the injector exit. The net result is a decrease in mixture ratio near the wall. Estimates of heat transfer and engine performance relative to the unbiased case are presented.

Strakey, P. A.↗
DOE OSTI2025

Evaluating Entrainment–Mixing Characteristics through Direct Comparisons of Drop Size Distributions Using In Situ Observations from ACE-ENA

Abstract Constraining the impacts of entrainment and associated mixing (i.e., entrainment–mixing) on cloud properties continues to be difficult, partly due to observational uncertainties as well as a lacking consensus of which methodologies for diagnosing entrainment–mixing are most appropriate. This study introduces a novel method to evaluate the presence and degree of inhomogeneous and homogeneous mixing using ∼100 h of in situ observations from a research aircraft over the northeastern Atlantic. Specifically, drop size distributions are compared between regions containing negligible and significant entrainment for select flight legs, making a direct characterization of the degree of homogeneous and inhomogeneous mixing possible. A measure of drop concentration variance is used as a proxy variable to diagnose entrainment–mixing. Results correspond well with entrainment–mixing metrics, showing lower Damköhler numbers where drop size distributions shift toward smaller drop sizes (i.e., inhomogeneous mixing) and greater transition length scales where drop size distributions do not (i.e., homogeneous mixing). Inhomogeneous mixing occurs in most samples from Aerosol and Cloud Experiment in the Eastern North Atlantic (ACE-ENA) (regardless of homogeneous mixing frequencies increasing with increasing spatial resolution from ∼100 to ∼10 m) and is associated with decreased drop size relative dispersion and both greater aerosol and drop concentrations compared with homogeneous mixing. Precipitating clouds have a greater frequency of homogeneous mixing compared with nonprecipitating clouds. The proposed methodology is similarly applied to in situ observations of southeast Pacific stratocumulus, shallow convective clouds over central Oklahoma and low-level clouds over the Southern Ocean. All four locations are primarily dominated by inhomogeneous mixing with minimal variability among each region.

Clouds↗
DOE OSTI2026

SimLBR: Learning to Detect Fake Images by Learning to Detect Real Images

The rapid advancement of generative models has made the detection of AI-generated images a critical challenge for both research and society. Recent works have shown that most state-of-the-art fake image detection methods overfit to their training data and catastrophically fail when evaluated on curated hard test sets with strong distribution shifts. In this work, we argue that it is more principled to learn a tight decision boundary around the real image distribution and treat the fake category as a sink class. To this end, we propose SimLBR, a simple and efficient framework for fake image detection with Latent Blending Regularization (LBR). Our method significantly improves cross-generator generalization, achieving up to +24.85% accuracy and +69.62% recall on the challenging Chameleon benchmark. SimLBR is also highly efficient, training orders of magnitude faster than existing approaches. Furthermore, we emphasize the need for reliability-oriented evaluation in fake image detection, introducing risk-adjusted metrics and worst-case estimates to better assess model robustness. All the code and models are availabe at: https://github.com/mvrl/SimLBR

Dhakal, Aayush [Washington University, St. Louis]↗
NASA NTRS2018

Climate Shifts Within Major Agricultural Seasons for +1.5 and +2.0 °C Worlds: HAPPI Projections and AgMIP Modeling Scenarios

This study compares climate changes in major agricultural regions and current agricultural seasons associated with global warming of +1.5 or +2.0 °C above pre-industrial conditions. It describes the generation of climate scenarios for agricultural modeling applications conducted as part of the Agricultural Model Intercomparison and Improvement Project (AgMIP) Coordinated Global and Regional Assessments. Climate scenarios from the Half a degree Additional warming, Projections, Prognosis and Impacts project (HAPPI) are largely consistent with transient scenarios extracted from RCP4.5 simulations of the Coupled Model Intercomparison Project phase 5 (CMIP5). Focusing on food and agricultural systems and top-producing breadbaskets in particular, we distinguish maize, rice, wheat, and soy season changes from global annual mean climate changes. Many agricultural regions warm at a rate that is faster than the global mean surface temperature (including oceans) but slower than the mean land surface temperature, leading to regional warming that exceeds 0.5 °C between the +1.5 and +2.0 °C Worlds. Agricultural growing seasons warm at a pace slightly behind the annual temperature trends in most regions, while precipitation increases slightly ahead of the annual rate. Rice cultivation regions show reduced warming as they are concentrated where monsoon rainfall is projected to intensify, although projections are influenced by Asian aerosol loading in climate mitigation scenarios. Compared to CMIP5, HAPPI slightly underestimates the CO2 concentration that corresponds to the +1.5 °C World but overestimates the CO2 concentration for the +2.0 °C World, which means that HAPPI scenarios may also lead to an overestimate in the beneficial effects of CO2 on crops in the +2.0 °C World. HAPPI enables detailed analysis of the shifting distribution of extreme growing season temperatures and precipitation, highlighting widespread increases in extreme heat seasons and heightened skewness toward hot seasons in the tropics. Shifts in the probability of extreme drought seasons generally tracked median precipitation changes; however, some regions skewed toward drought conditions even where median precipitation changes were small. Together, these findings highlight unique seasonal and agricultural region changes in the +1.5 °C and +2.0 °C worlds for adaptation planning in these climate stabilization targets.

Alexander C. Ruane↗
DOE OSTI2024

Elucidating the Impact of Cis – Trans Organic Structure Directing Agent Isomer Ratios on the Aluminum Distribution Within SSZ-39

Despite their widespread use, the mechanisms governing the synthesis of zeolite catalysts are still poorly understood. A notable example of this problem is the uncertainty surrounding the influence of synthesis conditions on the placement of Al atoms in the zeolite framework which determines the active sites available for catalytic species. In this work, the role of the cis to trans isomer ratio of the OSDA N,N-dimethyl-3-5-dimethylpiperidinium on the energetics of 26 distinct Al pair distributions in SSZ-39 is examined both in the presence and absence of Na using density functional theory calculations. The initial orientation of the OSDA was found to have a significant impact on the final energies present, necessitating the screening of a large number of initial orientations with force field calculations and single point DFT calculations. Ground state energies were found to vary significantly with the ratio of cis to trans OSDAs with a Boltzmann distribution revealing the most likely Al pair distributions shift from sharing the same 8 membered rings to sharing the same double six membered rings to having no shared subunits as one increases the amount of cis OSDA present within the framework. The presence of Na was found to favor Al pair distributions where both Als occupied the same 6-membered ring. When an implicit solvent model was used to evaluate ground state energies the ideal Na sites shifted from 6-membered rings to empty SSZ-39 cages while OSDA positions and orientations remained largely the same. To provide insight on how kinetic factors may influence Al distributions, formation energies we calculated for connected double six membered rings. Further, these formation energies revealed a preference for Al pairs to occupy the same 4-membered ring which indicates kinetic and thermodynamic control may lead to different Al distributions in SSZ-39.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2025

Lens Modeling of STRIDES Strongly Lensed Quasars Using Neural Posterior Estimation

Strongly lensed quasars can be used to constrain cosmological parameters through time-delay cosmography. Models of the lens masses are a necessary component of this analysis. To enable time-delay cosmography from a sample of $\mathcal{O}(10^3)$ lenses, which will soon become available from surveys like the Rubin Observatory’s Legacy Survey of Space and Time and the Euclid Wide Survey, we require fast and standardizable modeling techniques. To address this need, we apply neural posterior estimation (NPE) for modeling galaxy-scale strongly lensed quasars from the Strong Lensing Insights into the Dark Energy Survey (STRIDES) sample. NPE brings two advantages: speed and the ability to implicitly marginalize over nuisance parameters. We extend this method by employing sequential NPE to increase precision of mass model posteriors. We then fold individual lens models into a hierarchical Bayesian inference to recover the population distribution of lens mass parameters, accounting for out-of-distribution shift. After verifying our method using simulated analogs of the STRIDES lens sample, we apply our method to 14 Hubble Space Telescope single-filter observations. We find the population mean of the power-law elliptical mass distribution slope, γ lens , to be $\mathcal{M}_γ$ lens = 2.13 ± 0.06. Our result represents the first population-level constraint for these systems. This population-level inference from fully automated modeling is an important stepping stone toward cosmological inference with large samples of strongly lensed quasars.

79 ASTRONOMY AND ASTROPHYSICS↗
NASA NTRS1973

The luminosity function of quasars and its evolution: A comparison of optically selected quasars and quasars found in radio catalogs

The luminosity function of quasars and its evolution are discussed, based on comparison of available data on optically selected quasars and quasars found in radio catalogs. It is assumed that the red shift of quasars is cosmological and the results are expressed in the framework of the Lambda = 0, Q sub Q = 1 cosmological model. The predictions of various density evolution laws are compared with observations of an optically selected sample of quasars and quasar samples from radio catalogs. The differences between the optical luminosity functions, the red shift distributions and the radio to optical luminosity ratios of optically selected quasars and radio quasars rule out luminosity functions where there is complete absence of correlation between radio and optical luminosities. These differences also imply that Schmidt's (1970) luminosity function, where there exists a statistical correlation between radio and optical luminosities, although may be correct for high red shift objects, disagrees with observation at low red shifts. These differences can be accounted for by postulating existence of two classes (1 and 2) of objects.

Petrosian, V.↗
DOE OSTI2025

Object detection with deep learning for rare event search in the GADGET II TPC

In the pursuit of identifying rare two-particle events within the GADGET II Time Projection Chamber (TPC), this paper presents a comprehensive approach for leveraging Convolutional Neural Networks (CNNs) and various data processing methods. To address the inherent complexities of 3D TPC track reconstructions, the data is expressed in 2D projections and 1D quantities. This approach capitalizes on the diverse data modalities of the TPC, allowing for the efficient representation of the distinct features of the 3D events, with no loss in topology uniqueness. Additionally, it leverages the computational efficiency of 2D CNNs and benefits from the extensive availability of pre-trained models. Given the scarcity of real training data for the rare events of interest, simulated events are used to train the models to detect real events. To account for potential distribution shifts when predominantly depending on simulations, significant perturbations are embedded within the simulations. This produces a broad parameter space that works to account for potential physics parameter and detector response variations and uncertainties. These parameter-varied simulations are used to train sensitive 2D CNN object detectors. When combined with 1D histogram peak detection algorithms, this multi-modal detection framework is highly adept at identifying rare, two-particle events in data taken during experiment 21072 at the Facility for Rare Isotope Beams (FRIB), demonstrating a 100% recall for events of interest. Here, we present the methods and outcomes of our investigation and discuss the potential future applications of these techniques.

Convolutional neural network↗
DOE OSTI2024

Enabling dynamic 3D coherent diffraction imaging via adaptive latent space tuning of generative autoencoders

Abstract Coherent diffraction imaging (CDI) is an advanced non-destructive 3D X-ray imaging technique for measuring a sample’s electron density. The main challenge of CDI is loss of phase information in diffraction intensity measurements, resulting in lengthy iterative reconstruction processes that can return non-unique solutions, which pose challenges for experiments attempting to track dynamic sample evolution through multiple states. As the increased brightness of fourth-generation light sources enables faster sample measurements and drives operando experiments with Bragg CDI, there is a growing need for faster reconstruction techniques that can keep pace. We have developed an adaptive generative autoencoder approach for uniquely tracking a sample’s electron density as it dynamically evolves. Our approach adaptively tunes the low-dimensional latent embedding of a generative autoencoder, enabling a computationally efficient manner to account for time-varying shifting distributions in real-time. Analytic proof of convergence is provided as well as numerical demonstration of sample tracking with noisy measurements.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

Nonstationarity in the global terrestrial water cycle and its interlinkages in the Anthropocene

Climate change and human activities alter the global freshwater cycle, causing nonstationary processes as its distribution shifting over time, yet a comprehensive understanding of these changes remains elusive. Here, we develop a remote sensing–informed terrestrial reanalysis and assess the nonstationarity of and interconnections among global water cycle components from 2003 to 2020. We highlight 20 hotspot regions where terrestrial water storage exhibits strong nonstationarity, impacting 35% of the global population and 45% of the area covered by irrigated agriculture. Emerging long-term trends dominate the most often (48.2%), followed by seasonal shifts (32.8%) and changes in extremes (19%). Notably, in mid-latitudes, this encompasses 34% of Asia and 27% of North America. The patterns of nonstationarity and their dominant types differ across other water cycle components, including precipitation, evapotranspiration, runoff, and gross primary production. These differences also manifest uniquely across hotspot regions, illustrating the intricate ways in which each component responds to climate change and human water management. Our findings emphasize the importance of considering nonstationarity when assessing water cycle information toward the development of strategies for sustainable water resource usage, enhancing resilience to extreme events, and effectively addressing other challenges associated with climate change.

Science & Technology - Other Topics↗
DOE OSTI2025

Physics-constrained superresolution diffusion for six-dimensional phase space diagnostics

Adaptive physics-constrained superresolution diffusion is developed for noninvasive virtual diagnostics of the six-dimensional (6D) phase space density of charged particle beams. An adaptive variational autoencoder embeds initial beam condition images and scalar measurements to a low-dimensional latent space from which a 32 6 pixel 6D tensor representation of the beam's 6D phase space density is generated. Projecting from a 6D tensor generates physically consistent two-dimensional projections. Physics-guided superresolution diffusion transforms low-resolution images of the 6D density to high resolution 256 × 256 pixel images. Unsupervised adaptive latent space tuning enables tracking of time-varying beams without knowledge of time-varying initial conditions. The method is demonstrated with experimental data and multiparticle simulations at the HiRES UED. The general approach is applicable to a wide range of complex dynamic systems evolving in high-dimensional phase space. The method is shown to be robust to distribution shift without retraining. Published by the American Physical Society 2025

43 PARTICLE ACCELERATORS↗
DOE OSTI2025

Enhancing synchrotron radiation micro-CT images using deep learning: an application of Noise2Inverse on bone imaging

In bone-imaging research, in situ synchrotron radiation micro-computed tomography (SRµCT) mechanical tests are used to investigate the mechanical properties of bone in relation to its microstructure. Low-dose computed tomography (CT) is used to preserve bone's mechanical properties from radiation damage, though it increases noise. To reduce this noise, the self-supervised deep learning method Noise2Inverse was used on low-dose SRµCT images where segmentation using traditional thresholding techniques was not possible. Simulated-dose datasets were created by sampling projection data at full, one-half, one-third, one-fourth and one-sixth frequencies of an in situ SRµCT mechanical test. After convolutional neural networks were trained, Noise2Inverse performance on all dose simulations was assessed visually and by analyzing bone microstructural features. Visually, high image quality was recovered for each simulated dose. Lacunae volume, lacunae aspect ratio and mineralization distributions shifted slightly in full, one-half and one-third dose network results, but were distorted in one-fourth and one-sixth dose network results. Following this, new models were trained using a larger dataset to determine differences between full dose and one-third dose simulations. Significant changes were found for all parameters of bone microstructure, indicating that a separate validation scan may be necessary to apply this technique for microstructure quantification. Noise present during data acquisition from the testing setup was determined to be the primary source of concern for Noise2Inverse viability. While these limitations exist, incorporating dose calculations and optimal imaging parameters enables self-supervised deep learning methods such as Noise2Inverse to be integrated into existing experiments to decrease radiation dose.

Obata, Yoshihiro (ORCID:0000000303659129)↗
DOE OSTI2024

Counter Data Paucity through Adversarial Invariance Encoding: A Case Study on Modeling Battery Thermal Runaway

Lithium-ion batteries, widely used for their durability and high energy storage, face the risk of internal short circuits leading to catastrophic thermal runaway events. These events, triggered by external stimuli like mechanical loads, pose safety concerns in applications such as electric vehicles. Detecting and understanding thermal runaway events is crucial, but physics-driven models struggle to explain the non-linear evolution of battery temperature during these events, considering factors like material composition and state-of-charge. Due to the rarity of these events and the cost of data collection, we propose a deep learning (DL) model to predict battery temperature responses during thermal runaway. The challenge lies in the scarcity of data, making traditional DL models prone to overfitting and learning low-quality representations of the complex process.Our approach introduces a novel few-shot architecture that incorporates an adversarially governed invariant encoding process. This architecture aims to distill "invariant" relationships by addressing distributional shifts in data across various battery properties, facilitating the detection of thermal runaway events. Specifically, our results demonstrate that deep learning models conditioned on these "invariant" representations outperform state-of-the-art baselines, achieving a remarkable 96.8% performance improvement in terms of the popular metric MAPE. This framework presents a promising direction for enhancing battery safety modeling, particularly in the context of rare and complex events like thermal runaway. Our code and code and dataset used for the paper are public1.

Tabassum, Anika [ORNL] (ORCID:0000000254600955)↗