Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “feature exploration”

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 OSTI2020

eFESTA: Ensemble Feature Exploration with Surface Density Estimates

We propose surface density estimate (SDE) to model the spatial distribution of surface features-isosurfaces, ridge surfaces, and streamsurfaces-in 3D ensemble simulation data. The inputs of SDE computation are surface features represented as polygon meshes, and no field datasets are required (e.g., scalar fields or vector fields). The SDE is defined as the kernel density estimate of the infinite set of points on the input surfaces and is approximated by accumulating the surface densities of triangular patches. We also propose an algorithm to guide the selection of a proper kernel bandwidth for SDE computation. An ensemble Feature Exploration method based on Surface densiTy EstimAtes (eFESTA) is then proposed to extract and visualize the major trends of ensemble surface features. For an ensemble of surface features, each surface is first transformed into a density field based on its contribution to the SDE, and the resulting density fields are organized into a hierarchical representation based on the pairwise distances between them. The hierarchical representation is then used to guide visual exploration of the density fields as well as the underlying surface features. We demonstrate the application of our method using isosurface in ensemble scalar fields, Lagrangian coherent structures in uncertain unsteady flows, and streamsurfaces in ensemble fluid flows.

Computational modeling↗
DOE OSTI2024

Exploring Robust Features for Improving Adversarial Robustness

While deep neural networks (DNNs) have revolutionized many fields, their fragility to carefully designed adversarial attacks impedes the usage of DNNs in safety-critical applications. In this article, we strive to explore the robust features that are not affected by the adversarial perturbations, that is, invariant to the clean image and its adversarial examples (AEs), to improve the model’s adversarial robustness. Specifically, we propose a feature disentanglement model to segregate the robust features from nonrobust features and domain-specific features. Here, the extensive experiments on five widely used datasets with different attacks demonstrate that robust features obtained from our model improve the model’s adversarial robustness compared to the state-of-the-art approaches. Moreover, the trained domain discriminator is able to identify the domain-specific features from the clean images and AEs almost perfectly. This enables AE detection without incurring additional computational costs. With that, we can also specify different classifiers for clean images and AEs, thereby avoiding any drop in clean image accuracy.

97 MATHEMATICS AND COMPUTING↗
NASA NTRSDate not supplied

Photonic Lanterns Applied to Deep Space Optical Communication Receivers

This presentation will begin with an introduction to the design aspects of photon counting optical communications light coupling to detectors. The discussion will continue an introduction to the trade space of light coupling and detector architectures; including solutions that use a photonic lantern. The explored features of the alternative designs will be presented as well as a discussion of remaining aspects that need to be explored.

fiber↗
NASA NTRSDate not supplied

Transient Events in the Chromosphere and Corona with IBIS, ALMA, IRIS, and XRT

The connection between the photosphere, chromosphere, and corona is essential for understanding energy flow throughout the solar atmosphere. Recent work shows strong correlations between observations of the chromospheric mm-continuum (e.g., ALMA) and the low chromospheric H alpha (e.g., IBIS), and also between mm-continuum and the upper chromosphere and corona, such as SDO and XRT. However, the features which correlate between ALMA and IBIS are not those that correlate between ALMA and observations higher in the atmosphere. In this work, we examine transient features observed from the photosphere to the corona, and determine when and where the atmospheric features agree, and when they diverge. In particular, we show that the established IBIS line-width and ALMA Band 3 relationships break down during transient events (such as blue-wing excursions), though the ALMA and the H-alpha blue wing transients appear related. We also explore features of these transients that appear to foreshadow coronal signatures and find consistent connections and time-lags between transients observed in ALMA and features seen in the upper atmosphere.

A.R. Kobelski↗
NASA NTRS2003

MHD Instability and Turbulence in the Tachocline

The focus of this project was to study the physical processes that govern tachocline dynamics and structure. Specific features explored included stratification, shear, waves, and toroidal and poloidal background fields. In order to address recent theoretical work on anisotropic mixing and dynamics in the tachocline, we were particularly interested in such anisotropic mixing for the specific tachocline processes studied. Transition to turbulence often shapes the largest-scale features that appear spontaneously in a flow during the development of turbulence. The resulting large-scale straining field can control the subsequent dynamics; therefore, anticipation of the large-scale straining field that results for individual realizations of the transition to turbulence can be important for subsequent dynamics, flow morphology, and transport characteristics. As a result, we paid particular attention to the development of turbulence in the stratified and sheared environment of the tachocline. This is complicated by the fact that the linearly stability of sheared MHD flows is non-self-adjoint, implying that normal asymptotic linear stability theory may not be relevant.

Werne, Joe↗
DOE OSTI2020

Predicting Elastic Properties of Materials from Electronic Charge Density Using 3D Deep Convolutional Neural Networks

Materials representation plays a key role in machine learning-based prediction of materials properties and new materials discovery. Currently both graph and three-dimensional (3D) voxel representation methods are based on the heterogeneous elements of the crystal structures. Here, we propose to use electronic charge density (ECD) as a generic unified 3D descriptor for materials property prediction with the advantage of possessing close relation with the physical and chemical properties of materials. We developed an ECD-based 3D convolutional neural networks (CNNs) for predicting the elastic properties of materials, in which CNNs can learn effective hierarchical features with multiple convolving and pooling operations. Extensive benchmark experiments over 2170 $Fm\bar3m$ face-centered-cubic materials show that our ECD-based CNNs can achieve good performance for elasticity prediction. Especially, our CNN models based on the fusion of elemental Materials-Agnostic Platform for Informatics and Exploration features and ECD descriptors achieved the best fivefold cross-validation performance. More importantly, we showed that our ECD-based CNN models can achieve significantly better extrapolation performance when evaluated over nonredundant data sets, where there are few neighbor-training samples around test samples. As an additional validation, we evaluated the predictive performance of our models on 329 materials of space group $Fm\bar3m$ by comparing to density functional theory calculated values, which shows a better prediction power of our model for bulk modulus than shear modulus. Because of the unified representation power of ECD, it is expected that our ECD-based CNN approach can also be applied to predict other physical and chemical properties of crystalline materials.

36 MATERIALS SCIENCE↗
DOE OSTI2022

Feature engineering for machine learning enabled early prediction of battery lifetime

Accurate battery lifetime estimates enable accelerated design of novel battery materials and determination of optimal use protocols for longevity in deployments. Unfortunately, traditional battery testing may take years to reach thousands of cycles. Recent studies have shown that machine learning (ML) tools can predict lithium-ion battery lifetimes from 100 or fewer preliminary cycles, representing only a few weeks of cycling. Until now, conclusions about the efficacy and broad applicability of these predictions across a variety of cathode chemistries have been limited by available experimental information. In this work, we leverage a battery cycling dataset representing six cathode chemistries (NMC111, NMC532, NMC622, NMC811, HE5050, and 5Vspinel), multiple electrolyte/anode compositions, and 300 total carefully prepared pouch batteries to explore feature selection and battery chemistry's role in ML battery lifetime predictions. Here, a mean absolute error (MAE) of 78 cycles in prediction was seen for a chemistry-spanning test set from 100 preliminary cycles. Furthermore, an MAE of 103 cycles was seen when using only the first cycle. This study represents an in-depth investigation of strategies for feature selection for battery lifetime prediction, ML models' generalization across multiple battery chemistries, and predictions beyond the training set in the chemical space.

25 ENERGY STORAGE↗
NASA NTRS2020

Ancient Martian aeolian sand dune deposits recorded in the stratigraphy of Valles Marineris and implications for past climates

Aeolian sediment transport, deposition, and erosion have been ongoing throughout Mars’ history. This record of widespread aeolian processes is preserved in landforms and geologic units that retain important clues about past environmental conditions including wind patterns. In this study we describe landforms within Melas Chasma, Valles Marineris, that occur in distinct groups with linear to crescentic shapes, arranged with a characteristic wavelength; some possess slope profiles analogous to modern sand dunes, yet show evidence for lithification. Based on the features’ dimensions, asymmetry, and spatial patterns relative to modern equivalents, we interpret these landforms to be two classes of aeolian bedforms: decameter-scale mega-ripples and sand dunes. The presence of superposed erosional features and depositional units indicates these landforms were cemented and likely ancient. Melas paleo-dunes are found atop Hesperian-aged layered deposits, but we estimate them to be younger, likely lithified in the Amazonian period. Although a range of degradation was observed, some paleo-dunes are >10 m tall and maintain steep lee sides (>25°), an uncommon scenario for terrestrial examples as other geologic processes lead to dune obliteration. The preserved paleo-bedform geometries are largely consistent with those of modern aeolian indicators, suggesting no major shifts in wind regime or contributing boundary conditions. Finally, we propose their appearance and context require sequential periods of dune migration, stabilization following catastrophic burial, cementation, differential erosion, exposure, and burial. The presence of wholly preserved duneforms appears to be more common on Mars compared to the Earth and may signal something important about Martian landscape evolution. Plain Language Summary Wind-driven sand dunes are common on modern Mars and the presence of certain sedimentary rock layers indicates these landforms occurred there in the past. Here, we explore features in the canyons of Valles Marineris which show characteristics frequently attributed to dunes, yet evidence for their lithification and burial is clear. Their horizontal extent, height, shape, slopes, and collective spatial patterns led to our interpretation of them as lithified dune fields. The heavily eroded appearance along with superposed elements such as craters and boulders led to the conclusion that these are relatively ancient landforms. Despite this erosion This level of preservation is rare for terrestrial sand dunes due to erosion and tectonics, thus providing an opportunity to reconstruct the various factors contributing to their history. Understanding the circumstances that led to the preservation of these ancient duneforms will yield crucial information regarding planetary sedimentary processes and the geologic history of the region.

Matthew Chojnacki↗
DOE OSTI2022

Downwind control of oceanic air by land: the land wake and its sensitivity to CO 2

Oceans are well-known to be directly altered by global climate forcings such as greenhouse gas changes, but how oceans are indirectly influenced by land and its response to such forcings remains less explored. Here, we assess the present-day and projected future state of a little-explored feature of the climate system—a 'land wake' in relative humidity downwind of the east coast of North America, consisting of low-humidity continental air extending roughly 1000 km over the Atlantic ocean. The wake exists throughout the year, but is supported by high continental temperatures in summer and low continental moisture in winter. The wake is well represented in an ensemble of global climate models (GCMs), qualitatively matching reanalysis data. Under increasing atmospheric CO 2 , the land wake intensifies in GCM simulations through two pathways: the radiative effects of CO 2 on surface temperatures, and the biogeochemical effect of CO 2 on terrestrial vegetation. Vegetation responses to increased CO 2 alter the summer wake from Florida to Newfoundland, and both the radiative and biogeochemical effects of CO 2 drive reductions in coastal cloud cover. These changes illustrate the potential of rapidly changing terrestrial climate to influence coastal regions and the ocean environment downwind of continents through both light conditions and the energy balance of the surface ocean.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2023

Characterizing Quantum Classifier Utility in Natural Language Processing Workflows

Quantum Natural Language Processing (QNLP) develops natural language processing (NLP) models for deployment on quantum computers. We explore feature and data prototype selection techniques to address challenges posed by encoding high dimensional features. Our study builds quantum circuit classifiers that includes classical feature pre-processing, quantum embedding and quantum model training. The quantum models are built on 4 or 6 qubits and the quantum neural network (QNN) uses the established bricklayer design. We compare the dependence of model performance (in terms of accuracy and F1 scores) on feature length, embedding gates and parameterized unitary design. We compare the performance of quantum machine learning models to classical convolution neural network model (CNN) on binary and multi-class classification tasks using two datasets of synthetic features and labels. The first is the ECP-CANDLE P3B3 dataset a corpus of synthetically generated cancer pathology reports. The second dataset is extracted from well-known benchmark dataset (MADELON) - features are generated with a combination of informative, repeated and uninformative features. Both datasets are used for binary classification and multi-class classification with 3 classes. We observe robust, accurate performance from all models on the binary classification tasks, but multiclass classification is a challenge for the quantum models-there is a notable decrease in accuracy when using 3 classes. Overall the performance is comparable in terms of recall and accuracy between QNNs and CNNs, even with large datasets. These results provide a point of comparison between quantum and classical models on real-world datasets.

Hamilton, Kathleen↗
DOE OSTI2026

Powertrain Autoresearch (TETA Autoresearch: RouteE Powertrain Experiments) [SWR-26-090]

Powertrain Autoresearch is an experiment repository based on the template set in teta-autoresearch containing branch history of LLM workers investigating powertrain ML model improvements. This repository contains an instance of teta-autoresearch applied to ML model exploration for RouteE Powertrain. An AI agent and human-in-the-loop iteratively explored feature space, ML model type, and hyperparameter spaces in an attempt to improve over existing models. The repo catalogs the result for review at USRSE 2026 Conference. Related: https://github.com/NatLabRockies/teta-autoresearch.

Reinicke, Nicholas [National Laboratory of the Roc↗
DOE OSTI2021

TX$^2$: Transformer eXplainability and eXploration

The Transformer eXplainability and eXploration (Martindale & Stewart, 2021), or TX 2 software package, is a library designed for artificial intelligence researchers to better understand the performance of transformer models (Vaswani et al., 2017) used for sequence classification. The tool is capable of integrating with a trained transformer model and a dataset split into training and testing populations to produce an ipywidget (Project Jupyter Contributors, 2021) dashboard with a number of visualizations to understand model performance with an emphasis on explainability and interpretability. The TX 2 package is primarily intended to integrate into a workflow centered around Jupyter Notebooks (Kluyver et al., 2016), and currently assumes the use of PyTorch (Paszke et al., 2019) and Hugging Face transformers library (Wolf et al., 2020). The dashboard includes visualization and data exploration features to aid researchers, including an interactive UMAP embedding graph (McInnes et al., 2018) to understand classification clusters, a word salience map that can be updated as researchers alter textual entries in near real time, a set of tools to understand word frequency and importance based on the clusters in the UMAP embedding graph, and a set of traditional confusion matrix analysis tools.

97 MATHEMATICS AND COMPUTING↗