Search NASA⌕ Search

SEARCH · Search NASA

Results for “data augmentation”

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 73 records · Page 4
All sourcesNASA NTRSDOE OSTI
DOE OSTI2025

Quantum adversarial learning for kernel methods

We show that hybrid quantum classifiers based on quantum kernel methods and support vector machines are vulnerable against adversarial attacks, namely small engineered perturbations of the input data can deceive the classifier into predicting the wrong result. Nonetheless, we also show that simple defense strategies based on data augmentation with a few crafted perturbations can make the classifier robust against new attacks. Our results find applications in security-critical learning problems and in mitigating the effect of some forms of quantum noise, since the attacker can also be understood as part of the surrounding environment.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2025

Leveraging explainable AI to characterize floating-point exceptions in linear solvers

Linear solver packages are central to many scientific, engineering, and machine learning applications. When floating-point exceptions occur in these solvers, e.g., division by zero or overflow, numerical results are compromised and become unreliable. Existing static and dynamic analysis tools can detect such exceptions, but they do not explain why the exceptions occur in terms of the solver inputs. Here, we present a study to characterize the inputs that cause numerical exceptions in linear solver packages. Our approach uses explainable AI (XAI) to find the most relevant characteristics of input matrices that explain the occurrence of exceptions in the solvers. Since training data in this domain is scarce, we perform extensive data gathering and data augmentation to obtain exception-inducing inputs. Our approach uses a repair strategy on the features blamed by XAI to validate that such features indeed explain the exceptions. We compare the LIME and SHAP XAI techniques using a dozen matrix features with three classifiers. We evaluate the approach on three widely used linear solver packages and find that some input characteristics can explain the occurrence of exceptions 100% of the time, in specific solvers and preconditioners.

Explainable AI↗
DOE OSTI2025

An end-to-end deep learning solution for automated LiDAR tree detection in the urban environment

Cataloging and classifying trees in the urban environment is a crucial step in urban and environmental planning; however, manual collection and maintenance of this data is expensive and time-consuming. Although algorithmic approaches that rely on remote sensing data have been developed for tree detection in forests, they generally struggle in the more varied urban environment. This work proposes a novel end-to-end deep learning method for the detection of trees in the urban environment from remote sensing data. Specifically, we develop and train a novel PointNet-based neural network architecture to predict tree locations directly from LiDAR data augmented with multi-spectral imagery. We compare this model to a number of high-performing baselines on a large and varied dataset in the Southern California region, and find that our method outperforms all baselines in terms of tree detection ability (75.5% F-score) and positional accuracy (2.28 meter root mean squared error), while being highly efficient. We then analyze and compare the sources of errors, and how these reveal the strengths and weaknesses of each approach. Our results highlight the importance of fusing spectral and structural information for remote sensing tasks in complex urban environments.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2025

Resimulation-based self-supervised learning for pretraining physics foundation models

Self-supervised learning (SSL) is at the core of training modern large machine learning models, providing a scheme for learning powerful representations that can be used in a variety of downstream tasks. However, SSL strategies must be adapted to the type of training data and downstream tasks required. We propose resimulation-based self-supervised representation learning (RS3L), a novel simulation-based SSL strategy that employs a method of resimulation to drive data augmentation for contrastive learning in the physical sciences, particularly, in fields that rely on stochastic simulators. By intervening in the middle of the simulation process and rerunning simulation components downstream of the intervention, we generate multiple realizations of an event, thus producing a set of augmentations covering all physics-driven variations available in the simulator. Using experiments from high-energy physics, we explore how this strategy may enable the development of a foundation model; we show how RS3L pretraining enables powerful performance in downstream tasks such as discrimination of a variety of objects and uncertainty mitigation. In addition to our results, we make the RS3L dataset publicly available for further studies on how to improve SSL strategies.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2025

Learning broken symmetries with approximate invariance

Recognizing symmetries in data allows for significant boosts in neural network training, which is especially important where training data are limited. In many cases, however, the exact underlying symmetry is present only in an idealized dataset, and is broken in actual data, due to asymmetries in the detector, or varying response resolution as a function of particle momentum. Standard approaches, such as data augmentation or equivariant networks fail to represent the nature of the full, broken symmetry, effectively overconstraining the response of the neural network. We propose a learning model which balances the generality and asymptotic performance of unconstrained networks with the rapid learning of constrained networks. This is achieved through a dual-subnet structure, where one network is constrained by the symmetry and the other is not, along with a learned symmetry factor. In a simplified toy example that demonstrates violation of Lorentz invariance, our model learns as rapidly as symmetry constrained networks but escapes its performance limitations.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2025

Leveraging Pre-Built Catalogs and Object-Level Scheduling to Eliminate I/O Bottlenecks in HPC Environments

Modern High-Performance Computing (HPC) environments face mounting challenges due to the shift from large to small file datasets, along with an increasing number of users and parallelized applications. As HPC systems rely on Parallel File Systems (PFS), such as Lustre for data processing, performance bottlenecks stemming from Object Storage Target (OST) contention have become a significant concern. Existing solutions, such as LADS with its object-level scheduling approach, fall short in large-scale HPC environments due to their inability to effectively address metadata I/O bottlenecks and the growing number of I/O processes. This study highlights the pressing need for a comprehensive solution that tackles both OST contention and metadata I/O challenges in diverse HPC workloads. To address these challenges, we propose SwiftLoad, an object-level I/O scheduling framework that leverages a metadata catalog to enhance the performance and efficiency of parallel HPC utilities. The adoption of the metadata catalog mitigates the metadata I/O bottlenecks that commonly occur in HPC utilities, a challenge that is particularly pronounced in object-level I/O scheduling. SwiftLoad addresses OST contention and the uneven distribution of I/O processes across different OSTs through mathematical modeling and incorporates a Loader Configuration Module to regulate the number of I/O processes. Evaluated with two representative utilities—data deduplication profiling and data augmentation—SwiftLoad achieved performance improvements of up to 5.63x and 11.0x, respectively, on a production supercomputer.

HPC↗
DOE OSTI2025

INTEGRATE – Inverse Network Transformations for Efficient Generation of Robust Airfoil and Turbine Enhancements

The INTEGRATE (Inverse Network Transformations for Efficient Generation of Robust Airfoil and Turbine Enhancements) project developed a new inverse-design capability for the aerodynamic design of wind turbine rotors using invertible neural networks. Training data was obtained from improved turbulence and transition models for RANS and hybrid RANS/LES solvers with machine-learned physics-based data-augmented corrections and then using the resulting neural-network(s) augmented RANS model to run thousands of 2-D and 3-D CFD simulations.

17 WIND ENERGY↗
NASA NTRS1993

Inverse sequential procedures for the monitoring of time series

Climate changes traditionally have been detected from long series of observations and long after they happened. The 'inverse sequential' monitoring procedure is designed to detect changes as soon as they occur. Frequency distribution parameters are estimated both from the most recent existing set of observations and from the same set augmented by 1,2,...j new observations. Individual-value probability products ('likelihoods') are then calculated which yield probabilities for erroneously accepting the existing parameter(s) as valid for the augmented data set and vice versa. A parameter change is signaled when these probabilities (or a more convenient and robust compound 'no change' probability) show a progressive decrease. New parameters are then estimated from the new observations alone to restart the procedure. The detailed algebra is developed and tested for Gaussian means and variances, Poisson and chi-square means, and linear or exponential trends; a comprehensive and interactive Fortran program is provided in the appendix.

Radok, Uwe↗
NASA NTRS1994

Viscosity estimates for the crust and upper mantle from patterns of lacustrine shoreline deformation in the Eastern Great Basin

The deformed shorelines of Lake Bonneville constitute a classic source of information on lithospheric elastic thickness and upper mantle viscosity. We describe and apply a new model to a recently augmented data set. New data better constrain both the complex spatio-temporal pattern of the lake load and the crustal deformation response to that load. The history of lake level fluctuations has been significantly refined and somewhat modified. This is due to both more radiocarbon dates from within the Bonneville basin and to an improved calibration of the radiocarbon timescale itself. The data which constrain the crustal deformation pattern consist of ages and shoreline elevations from several hundred points which sample three major levels of Lake Bonneville and corresponding elevations from the high stands of three smaller lakes situated to the west of Lake Bonneville. The geometry of the Earth model incorporates an arbitrary number of layers overlying a half-space, and the rheology of each level can accommodate an arbitrary number of Maxwell viscoelastic elements in parallel. The inverse modeling comprises three complementary approaches: for the simplest configurations, we performed a direct search of the parameter space and delineated the irregular boundary of the subspace of acceptable models. For more complex configurations, we constrained the elastic parameters to their seismically determined values and then solved for viscosity versus depth profiles by either expressing the log(viscosity) versus log(depth) profile as a series of specially constructed orhtogonal polynomials, or by allowing each of 8-10 layers (plus the half-space) to have an independently determined viscosity. We found that the data do not strongly support (nor can they conclusively exclude) a more complex rheology than simple Maxwell viscoelasticity. The orthogonal polynomial solution exhibits an essentially monotonic decrease in viscosity with depth.

Bills, Bruce G.↗
NASA NTRS2000

Recent Results From The Nasa Earth Science Terra Mission and Future Possibilities

The NASA Earth Sciences Enterprise has made some remarkable strides in recent times in using developing, implementing, and utilizing spaceborne observations to better understand how the Earth works as a coupled, interactive system of the land, ocean, and atmosphere. Notable examples include the Upper Atmosphere Research (UARS) Satellite, the Topology Ocean Experiment (TOPEX) mission, Landsat-7, SeaWiFS, the Tropical Rainfall Monitoring Mission (TRMM), Quickscatt, the Shuttle Radar Topography Mission (SRTM), and, quite recently, the Terra'/Earth Observing System-1 mission. The Terra mission, for example, represents a major step forward in providing sensors that offer considerable advantages and progress over heritage instruments. The Moderate Resolution Imaging Spectrometer (MODIS), the Multi-angle Imaging SpectroRadiometer (MISR), the Measurements of Pollution in the Troposphere (MOPITT), the Advanced Spaceborne Thermal Emissions and Reflections (ASTER) radiometer, and the Clouds and Earth's Radiant Energy System (CERES) radiometer are the instruments involved. Early indications in March indicate that each of these instruments are working well and will be augmenting data bases from heritage instruments as well as producing new, unprecedented observations of land, ocean, and atmosphere features. Several missions will follow the Terra mission as the Earth Observing mission systems complete development and go into operation. These missions include EOS PM-1/'Aqua', Icesat, Vegetation Canopy Lidar (VCL), Jason/TOPEX Follow-on, the Chemistry mission, etc. As the Earth Observing systems completes its first phase in about 2004 a wealth of data enabling better understanding of the Earth and the management of its resources will have been provided. Considerable thought is beginning to be placed on what advances in technology can be implemented that will enable further advances in the early part of the 21st century; e.g., in the time from of 2020. Concepts such as 'constellation' missions or 'formation flying' with 'sensorcraft', 'sensor webs', autonomous operation of satellites, more on-board processing and delivery to individual users, data synthesis and analysis in real-time, etc. are being considered. With the data now having been and soon to be received plus the very real possibilities of further advances in use and applicability of data the potential for very significant gains in knowledge for Earth studies and applications looks quite high in the next decade or two.

Salomonson, Vincent V.↗
NASA NTRS2019

Citizen Science Twitter Data Management for Earth Science Applications

Social media data can provide useful real-time and historical information relating to the natural world, but managing this data poses challenges. Scientists at GES DISC are exploring the potential of Twitter data to augment precipitation data from the Global Precipitation Measurement (GPM) mission. However, the format of Twitter data is unconventional in the context of NASA data centers, resulting in frustration for scientists who need to work with the data. This study investigated procedures and standards needed to properly manage Twitter data to make them compatible with these data centers. After comparing databases, the study found that the MongoDB database was best suited for the storage of raw Twitter data due to its flexibility, ability to be accessed by multiple users, and querying functionality. The study used the Python package Zarr to transform processed Twitter data into a gridded format similar to that of satellite data. Each Tweet was mapped onto a time-space grid; each grid location contained information about Tweet attributes and precipitation. The study developed a pipeline for downloading, storing, and gridding Twitter data and transformed Twitter data into an understandable format for users of NASA satellite data.

Li, Rachel↗
NASA NTRS2019

Small target detection for search and rescue operations using distributed deep learning and synthetic data generation

It is important to find the target as soon as possible for search and rescue operations. Surveillance camera systems and unmanned aerial vehicles (UAVs) are used to support search and rescue. Automatic object detection is important because a person cannot monitor multiple surveillance screens simultaneously for 24 hours. Also, the object is often too small to be recognized by the human eye on the surveillance screen. This study used UAVs around the Port of Houston and fixed surveillance cameras to build an automatic target detection system that supports the US Coast Guard (USCG) to help find targets (e.g., person overboard). We combined image segmentation, enhancement, and convolution neural networks to reduce detection time to detect small targets. We compared the performance between the auto-detection system and the human eye. Our system detected the target within 8 seconds, but the human eye detected the target within 25 seconds. Our systems also used synthetic data generation and data augmentation techniques to improve target detection accuracy. This solution may help the search and rescue operations of the first responders in a timely manner.

Chow, Edward↗
NASA NTRS2022

A Comprehensive Machine Learning Study to Classify Precipitation Type over Land from Global Precipitation Measurement Microwave Imager (GPM-GMI) Measurements

Precipitation type is a key parameter used for better retrieval of precipitation characteristics as well as to understand the cloud–convection–precipitation coupling processes. Ice crystals and water droplets inherently exhibit different characteristics in different precipitation regimes (e.g., convection, stratiform), which reflect on satellite remote sensing measurements that help us distinguish them. The Global Precipitation Measurement (GPM) Core Observatory’s microwave imager (GMI) and dual-frequency precipitation radar (DPR) together provide ample information on global precipitation characteristics. As an active sensor, the DPR provides an accurate precipitation type assignment, while passive sensors such as the GMI are traditionally only used for empirical understanding of precipitation regimes. Using collocated precipitation type flags from the DPR as the “truth”, this paper employs machine learning (ML) models to train and test the predictability and accuracy of using passive GMI-only observations together with ancillary information from a reanalysis and GMI surface emissivity retrieval products. Out of six ML models, four simple ones (support vector machine, neural network, random forest, and gradient boosting) and the 1-D convolutional neural network (CNN) model are identified to produce 90–94% prediction accuracy globally for five types of precipitation (convective, stratiform, mixture, no precipitation, and other precipitation), which is much more robust than previous similar effort. One novelty of this work is to introduce data augmentation (subsampling and bootstrapping) to handle extremely unbalanced samples in each category. A careful evaluation of the impact matrices demonstrates that the polarization difference (PD), brightness temperature (Tc) and surface emissivity at high-frequency channels dominate the decision process, which is consistent with the physical understanding of polarized microwave radiative transfer over different surface types, as well as in snow and liquid clouds with different microphysical properties. Furthermore, the view-angle dependency artifact that the DPR’s precipitation flag bears with does not propagate into the conical-viewing GMI retrievals. This work provides a new and promising way for future physics-based ML retrieval algorithm development.

machine learning/artificial intelligence↗
NASA NTRSDate not supplied

Contextualizing Air Traffic Management Conversations using Natural Language Understanding

Efficient management of air traffic and mitigation of delays depend on extracting actionable information from unstructured data, such as dialogues from the Federal Aviation Administration’s (FAA’s) Air Traffic Control System Command Center (ATCSCC) telecons. This study presents a pipeline utilizing Natural Language Processing (NLP) methods for Intent Classification (IC) and Slot Filling (SF) to identify and extract Traffic Management Initiatives (TMIs) from aviation-specific dialogues. We leveraged DeBERTa, a pre-trained transformer model, and fine-tuned it to the nuances of the aviation domain. Despite challenges posed by annotation complexities, the IC model achieved promising results with a weighted average F1-score of 0.81. Our results are close to those of human annotators, which demonstrates the model’s strong alignment with human-level performance. The SF model also showed strong performance, achieving a weighted F1-score of 0.97, which demonstrates its effectiveness in accurately predicting key slots. Our analysis revealed limitations in handling less frequent intents and slot labels due to data sparsity, motivating future efforts to adopt joint IC-SF modeling and data augmentation strategies. This research highlights the potential of domain-specific NLP to streamline decision-making in the aviation industry and improve the management of TMIs.

Air Traffic Control Management↗
NASA NTRS1968

Development of Electronic Data Processing /EDP/ augmented management system

To tailor the existing Unified Flight Analysis System to management data rather than technical data, a pilot model could be produced in breadboard form, using electronic data processing, in a matter of a few months at very moderate cost. Such a system lends itself to continuous refinement.

Scott, J. E.↗
NASA NTRS2018

Using Concurrent Cardiovascular Information to Augment Survival Time Data for Evaluating Orthostatic Tilt Test Performance

Head-up tilt (HUT) tests often are used in research to measure orthostatic intolerance (OI) (inability to appropriately control blood pressure while upright) in clinical populations and otherwise healthy individuals after interventions. Post-space flight orthostatic intolerance is a well-known phenomenon, and countermeasures to its development has been an active area of research at NASA. In the NASA HUT protocol, subjects lie horizontally on an automatic tilt table for baseline measurements before being raised to 80deg head-up tilt for a defined period of time or until signs or symptoms of presyncope ensues (light-headedness, nausea, dizziness, sweating, weakness or fainting). Multiple measures are collected to evaluate the cardiovascular system's ability to respond appropriately to the orthostatic challenge. However if the intended duration of the HUT is short, the ability to detect changes in OI due to an intervention or its prevention by a countermeasure may be limited by a small number of failures to permit comparisons based on survival time alone. Thus, the time-trajectory of the cardiovascular data becomes an important additional source of information. In particular, we will show how various measures of trajectory variability can effectively augment survival analysis for the assessment of OI in a joint model when high censoring rates are present.

Feiveson, Alan H.↗
NASA NTRS1992

Operationally efficient propulsion system study (OEPSS) data book: Air Augmented Rocket Afterburning - Volume 10

A study was directed towards assessing viability and effectiveness of an air augmented ejector/rocket. Successful thrust augmentation could potentially reduce a multi-stage vehicle to a single stage-to-orbit vehicle (SSTO) and, thereby, eliminate the associated ground support facility infrastructure and ground processing required by the eliminated stage. The results of this preliminary study indicate that an air augmented ejector/rocket propulsion system is viable. However, uncertainties resulting from simplified approach and assumptions must be resolved by further investigations.

Farhangi, Shahram↗
NASA NTRS1999

Preliminary Assessment of JERS-1 SAR to Discriminating Boreal Landscape Features for the Boreal Forest Mapping Project

This paper presents an overview of the JERS-1 North American Boreal Forest Mapping Project and a preliminary assessment of JERS-1 SAR imagery for application to discriminating features applicable to boreal landscape processes. The present focus of the JERS-1 North American Boreal Forest Mapping Project is the production of continental scale wintertime and summertime SAR mosaics of the North American boreal forest for distribution to the science community. As part of this effort, JERS-1 imagery has been collected over much of Alaska and Canada during the 1997-98 winter and 1998 summer seasons. To complete the mosaics, these data will be augmented with data collected during previous years. These data will be made available to the scientific community via CD ROM containing these and similar data sets compiled from companion studies of Asia and Europe. Regional landscape classification with SAR is important for the baseline information it will provide about distribution of woodlands, positions of treeline, current forest biomass, distribution of wetlands, and extent of major rivercourses. As well as setting the stage for longer term change detection, comparisons across several years provides additional baseline information about short-term landscape change. Rapid changes, including those driven by fire, permafrost heat balance, flooding, and insect outbreaks can dominate boreal systems. We examine JERS-1 imagery covering selected sites in Alaska and Canada to assess quality and applicability to such relevant ecological and hydrological issues. The data are generally of high quality and illustrate many potential applications. A texture-based classification scheme is applied to selected regions to assess the applicability of these data for distinguishing distribution of such landcover types as wetland, tundra, woodland and forested landscapes.

McDonald, Kyle↗