Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “data representations”

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 307 records · Page 17

Explaining and predicting human behavior and social dynamics in simulated virtual worlds: reproducibility, generalizability, and robustness of causal discovery methods

Ground Truth program was designed to evaluate social science modeling approaches using simulation test beds with ground truth intentionally and systematically embedded to understand and model complex Human Domain systems and their dynamics Lazer et al. (Science 369:1060–1062, 2020). Our multidisciplinary team of data scientists, statisticians, experts in Artificial Intelligence (AI) and visual analytics had a unique role on the program to investigate accuracy, reproducibility, generalizability, and robustness of the state-of-the-art (SOTA) causal structure learning approaches applied to fully observed and sampled simulated data across virtual worlds. In addition, we analyzed the feasibility of using machine learning models to predict future social behavior with and without causal knowledge explicitly embedded. In this paper, we first present our causal modeling approach to discover the causal structure of four virtual worlds produced by the simulation teams—Urban Life, Financial Governance, Disaster and Geopolitical Conflict. Our approach adapts the state-of-the-art causal discovery (including ensemble models), machine learning, data analytics, and visualization techniques to allow a human-machine team to reverse-engineer the true causal relations from sampled and fully observed data. We next present our reproducibility analysis of two research methods team’s performance using a range of causal discovery models applied to both sampled and fully observed data, and analyze their effectiveness and limitations. We further investigate the generalizability and robustness to sampling of the SOTA causal discovery approaches on additional simulated datasets with known ground truth. Our results reveal the limitations of existing causal modeling approaches when applied to large-scale, noisy, high-dimensional data with unobserved variables and unknown relationships between them. We show that the SOTA causal models explored in our experiments are not designed to take advantage from vasts amounts of data and have difficulty recovering ground truth when latent confounders are present; they do not generalize well across simulation scenarios and are not robust to sampling; they are vulnerable to data and modeling assumptions, and therefore, the results are hard to reproduce. Finally, when we outline lessons learned and provide recommendations to improve models for causal discovery and prediction of human social behavior from observational data, we highlight the importance of learning data to knowledge representations or transformations to improve causal discovery and describe the benefit of causal feature selection for predictive and prescriptive modeling.

97 MATHEMATICS AND COMPUTING↗

Do Neural Networks Trained with Topological Features Learn Different Internal Representations?

There is a growing body of work that leverages features extracted via topological data analysis to train machine learning models. While this field, sometimes known as topological machine learning (TML), has seen some notable successes, an understanding of how the process of learning from topological features differs from the process of learning from raw data is still limited. In this work, we begin to address one component of this larger issue by asking whether a model trained with topological features learns internal representations of data that are fundamentally different than those learned by a model trained with the original raw data. To quantify ``different'', we exploit two popular metrics that can be used to measure the similarity of the hidden representations of data within neural networks, neural stitching and centered kernel alignment. From these we draw a range of conclusions about how training with topological features does and does not change the representations that a model learns. Perhaps unsurprisingly, we find that structurally, the hidden representations of models trained and evaluated on topological features differ substantially compared to those trained and evaluated on the corresponding raw data. On the other hand, our experiments show that in some cases, these representations can be reconciled (at least to the degree required to solve the corresponding task) using a simple affine transformation. We conjecture that this means that neural networks trained on raw data may extract some limited topological features in the process of making predictions.

McGuire, Sarah L.↗

Abstract for CRADA between VariGrid Explorations Inc. and NETL (AGMT-0863)

As the energy industry continues to invest in data, big data, and advanced analytics to improve production, reduce operation and maintenance costs, improve safety, and reduce risks, uncertainty will serve as a limiting factor in the success of these operations. Intuitive communication of uncertainty can provide more transparency in the decision-making process; offering the ability to more accurately evaluate the risk and impact of decisions, identify data and knowledge gaps, and evaluate trade-offs. The issue of uncertainty will thus play a large role in the success of investments that rely on big data and data analytics. However, no robust method for representation of uncertainty in spatial data visuals currently exists. As such, uncertainty is, at best, communicated using multiple independent visuals and aggregated in final displays, or frequently omitted altogether. Failure to accurately and effectively quantify and communicate uncertainty poses a barrier, limiting critical information that would better allow businesses to make more accurate and intelligent decisions.

96 KNOWLEDGE MANAGEMENT AND PRESERVATION↗

Data for Monitoring Fracture Saturation with Internal Transportable Seismic Sources and Twin Neural Networks

Seismic coda-wave analysis is a well-developed method for detecting subtle physical changes in complex media by measuring arrival times in the late-arriving energy from multiply-scattered or reflected waves. However, a challenge arises when multiply-scattered waves are not sufficiently separated in time from the direct arrivals to provide a clear coda wave train. Additional complications for monitoring changes in fracture systems arise when the signals originate from unsynchronized internal sources, such as natural induced seismicity, from acoustic emission, or from transportable intra-fracture sources (chattering dust), that generate uncontrolled signals that vary in arrival time, amplitude and frequency content. Here, we use a twin neural network (also known as a Siamese neural network) for dimensionality reduction to analyze signals from chattering dust to classify the fluid saturation state of a synthetic fracture system. The twin neural network with shared weights generates a low-dimensional representation of the data input by minimizing contrastive loss, serving as the input to a multiclass classifier that accurately classifies whether multiple fractures in a fracture system are fully saturated or partially saturated, or whether a change in saturation has occurred in different fractures in the system. These results show that information buried in unresolved codas from uncontrolled sources can be extracted using machine learning to monitor the evolution of fracture systems caused by physical and chemical processes even when the scattered and direct wave fields overlap.

58 GEOSCIENCES↗

Depleted Uranium Discrimination Using Spectral Ratios from Data-Starved Spectra - 18619

The Institute for Clean Energy Technology (ICET) has developed mobile surveying platforms for the retrieval of fired depleted uranium (DU) munitions. ICET surveying platforms are fitted with differential global positioning system (GPS), scintillation detectors, and central processing units (CPUs). The surveying technology has historically been used to assist facilities where DU contamination is a concern to locate, retrieve, and remediate DU-contaminated areas. Residence times over DU sources for mobile systems normally do not produce enough data to use gamma ray spectroscopy to verify the source type. Discussed is a potential method used to discriminate DU signatures in data-starved gamma ray spectral data collected during field surveys. Laboratory tests have shown that data-starved gamma ray spectra for zero-valence, oxide, and uranium ores can be distinguished from each other and from background by spectral ratioing. Zero-valence DU and DU oxide gamma ray signatures differ due to the attenuation of low-energy gammas emitted from U-238 daughters. DU signatures can be differentiated from naturally occurring uranium spectra due to the presence of daughters below U-234, primarily bismuth (Bi)-214, in the uranium decay series. Three primary regions of interest (ROI) within the analyzed test spectra are used to roughly characterize the shape of each test spectrum and compare them to background. The three ROIs are focused around the gamma-emitting daughters of U-238: thorium (Th)-234, protactinium (Pa)-234 metastable (m), and Bi-214. One of the three regions is considered the base region. This region is used to estimate what the counts in the other two regions should be by multiplying the counts in the ROIs by a ratio of background counts from an area previously found to be impacted. Using this spectral ratioing technique, a non-zero difference in spectral ratio values indicates a shift in the test spectra away from background. Calculating ratios between the base region and the other two regions allows for a graphical representation of the data in a two-dimensional spectral ratioing space, which is used to differentiate between the types of uranium sources previously mentioned. Surveying processes that are based on gross counts from scintillation counters can result in a large number of false positives as a result of shifting background signals or the presence of naturally occurring uranium. Application of the spectral ratioing technique can further assist in site characterization by analyzing all of the data-starved survey spectra to identify false positives or the need for additional data collection. Accurate surveys of DU testing and training ranges reduces the costs associated with range clean-up by discriminating clean areas from those requiring remediation. (authors)

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

Autoencoder node saliency: Selecting relevant latent representations

The autoencoder is an artificial neural network that performs nonlinear dimension reduction and learns hidden representations of unlabeled data. With a linear transfer function it is similar to the principal component analysis (PCA). While both methods use weight vectors for linear transformations, the autoencoder does not come with any indication similar to the eigenvalues in PCA that are paired with eigenvectors. In this work, we propose a novel autoencoder node saliency method that examines whether the features constructed by autoencoders exhibit properties related to known class labels. The supervised node saliency ranks the nodes based on their capability of performing a learning task. It is coupled with the normalized entropy difference (NED). We establish a property for NED values to verify classifying behaviors among the top ranked nodes. Lastly, by applying our methods to real datasets, we demonstrate their ability to provide indications on the performing nodes and explain the learned tasks in autoencoders.

97 MATHEMATICS AND COMPUTING↗

Physical discovery in representation learning via conditioning on prior knowledge

Recent advances in electron, scanning probe, optical, and chemical imaging and spectroscopy yield bespoke data sets containing the information of structure and functionality of complex systems. In many cases, the resulting data sets are underpinned by low-dimensional simple representations encoding the factors of variability within the data. The representation learning methods seek to discover these factors of variability, ideally further connecting them with relevant physical mechanisms. However, generally, the task of identifying the latent variables corresponding to actual physical mechanisms is extremely complex. Here, we present an empirical study of an approach based on conditioning the data on the known (continuous) physical parameters and systematically compare it with the previously introduced approach based on the invariant variational autoencoders. The conditional variational autoencoder (cVAE) approach does not rely on the existence of the invariant transforms and hence allows for much greater flexibility and applicability. Interestingly, cVAE allows for limited extrapolation outside of the original domain of the conditional variable. However, this extrapolation is limited compared to the cases when true physical mechanisms are known, and the physical factor of variability can be disentangled in full. We further show that introducing the known conditioning results in the simplification of the latent distribution if the conditioning vector is correlated with the factor of variability in the data, thus allowing us to separate relevant physical factors. We initially demonstrate this approach using 1D and 2D examples on a synthetic data set and then extend it to the analysis of experimental data on ferroelectric domain dynamics visualized via piezoresponse force microscopy.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Novel deep learning methods for track reconstruction

For the past year, the HEP.TrkX project has been investigating machine learning solutions to LHC particle track reconstruction problems. A variety of models were studied that drew inspiration from computer vision applications and operated on an image-like representation of tracking detector data. While these approaches have shown some promise, image-based methods face challenges in scaling up to realistic HL-LHC data due to high dimensionality and sparsity. In contrast, models that can operate on the spacepoint representation of track measurements ("hits") can exploit the structure of the data to solve tasks efficiently. In this paper we will show two sets of new deep learning models for reconstructing tracks using space-point data arranged as sequences or connected graphs. In the first set of models, Recurrent Neural Networks (RNNs) are used to extrapolate, build, and evaluate track candidates akin to Kalman Filter algorithms. Such models can express their own uncertainty when trained with an appropriate likelihood loss function. The second set of models use Graph Neural Networks (GNNs) for the tasks of hit classification and segment classification. These models read a graph of connected hits and compute features on the nodes and edges. They adaptively learn which hit connections are important and which are spurious. The models are scaleable with simple architecture and relatively few parameters. Results for all models will be presented on ACTS generic detector simulated data.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Elastic functional principal component regression

We study regression using functional predictors in situations where these functions contains both phase and amplitude variability. In other words, the functions are misaligned due to errors in time measurements, and these errors can significantly degrade both model estimation and prediction performance. The current techniques either ignore the phase variability, or handle it via preprocessing, that is, use an off–the–shelf technique for functional alignment and phase removal. We develop a functional principal component regression model which has a comprehensive approach in handling phase and amplitude variability. The model utilizes a mathematical representation of the data known as the square–root slope function. These functions preserve the L 2 norm under warping and are ideally suited for simultaneous estimation of regression and warping parameters. Furthermore, using both simulated and real–world data sets, we demonstrate our approach and evaluate its prediction performance relative to current models. In addition, we propose an extension to functional logistic and multinomial logistic regression.

97 MATHEMATICS AND COMPUTING↗

MFA-DVR: direct volume rendering of MFA models

3D volume rendering is widely used to reveal insightful intrinsic patterns of volumetric datasets across many domains. However, the complex structures and varying scales of volumetric data can make efficiently generating high-quality volume rendering results a challenging task. Multivariate functional approximation (MFA) is a new data model that addresses some of the critical challenges: high-order evaluation of both value and derivative anywhere in the spatial domain, compact representation for large-scale volumetric data, and uniform representation of both structured and unstructured data. Here, in this paper, we present MFA-DVR, the first direct volume rendering pipeline utilizing the MFA model, for both structured and unstructured volumetric datasets. We demonstrate improved rendering quality using MFA-DVR on both synthetic and real datasets through a comparative study. We show that MFA-DVR not only generates more faithful volume rendering than using local filters but also performs faster on high-order interpolations on structured and unstructured datasets. MFA-DVR is implemented in the existing volume rendering pipeline of the Visualization Toolkit (VTK) to be accessible by the scientific visualization community.

97 MATHEMATICS AND COMPUTING↗

A comparison of Redlich-Kister polynomial and cubic spline representations of the chemical potential in phase field computations

Free energies play a central role in many descriptions of equilibrium and non-equilibrium properties of solids. Continuum partial differential equations (PDEs) of atomic transport, phase transformations and mechanics often rely on first and second derivatives of a free energy function. The stability, accuracy and robustness of numerical methods to solve these PDEs are sensitive to the particular functional representations of the free energy. In this communication we investigate the influence of different representations of thermodynamic data on phase field computations of diffusion and two-phase reactions in the solid state. First-principles statistical mechanics methods were used to generate realistic free energy data for HCP titanium with interstitially dissolved oxygen. While Redlich-Kister polynomials have formed the mainstay of thermodynamic descriptions of multi-component solids, they require high order terms to fit oscillations in chemical potentials around phase transitions. Here, we demonstrate that high fidelity fits to rapidly fluctuating free energy functions are obtained with spline functions. As a result, spline functions that are many degrees lower than Redlich-Kister polynomials provide equal or superior fits to chemical potential data and, when used in phase field computations, result in solution times approaching an order of magnitude speed up relative to the use of Redlich-Kister polynomials.

36 MATERIALS SCIENCE↗

A subzone reconstruction algorithm for efficient staggered compatible remapping

Staggered-grid Lagrangian hydrodynamics algorithms frequently make use of subzonal discretization of state variables for the purposes of improved numerical accuracy, generality to unstructured meshes, and exact conservation of mass, momentum, and energy. For Arbitrary Lagrangian–Eulerian (ALE) methods using a geometric overlay, it is difficult to remap subzonal variables in an accurate and efficient manner due to the number of subzone–subzone intersections that must be computed. This becomes prohibitive in the case of 3D, unstructured, polyhedral meshes. A new procedure is outlined in this paper to avoid direct subzonal remapping. The new algorithm reconstructs the spatial profile of a subzonal variable using remapped zonal and nodal representations of the data. The reconstruction procedure is cast as an under-constrained optimization problem. Enforcing conservation at each zone and node on the remapped mesh provides the set of equality constraints; the objective function corresponds to a quadratic variation per subzone between the values to be reconstructed and a set of target reference values. Numerical results for various pure-remapping and hydrodynamics tests are provided. Ideas for extending the algorithm to staggered-grid radiation-hydrodynamics are discussed as well as ideas for generalizing the algorithm to include inequality constraints.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Evolving radon diffusion through earthen barriers at uranium waste disposal sites

Field measurements of Rn-222 fluxes from the tops and bottoms of compacted clay radon barriers were used to calculate effective Rn diffusion coefficients (D Rn ) at four uranium waste disposal sites in the western United States to assess cover performance after more than 20 years of service. Values of D Rn ranged from 7.4 × 10 -7 to 6.0 × 10 -9 m 2 /s, averaging 1.42 × 10 -7 . Water saturation (S W ) from soil cores indicated that there was relatively little control of D Rn by S W , especially at higher moisture levels, in contrast to estimates from most steady-state diffusion models. Further, this is attributed to preferential pathways intrinsic to construction of the barriers or to natural process that have developed over time including desiccation cracks, root channels, and insect burrows in the engineered earthen barriers. A modification to some models in which fast and slow pathway D Rn values are partitioned appears to give a good representation of the data; 4% of the fast pathway was needed to fit the data regression. For locations with high S w and highest D Rn (and fluxes) at each site, the proportion of fast pathway ranged from 1.7% to 34%, but for many locations with lower fluxes, little if any fast pathway was needed.

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

Dielectric Properties of Organic Solvents in an Electric Field

The effect of an external electric field ( E ) on the dielectric constant (ε) of several pure solvents [propylene carbonate (PC), ethylene carbonate (EC), acetonitrile (MeCN), and dimethyl carbonate (DMC)] and for several EC/DMC mixtures is explored using classical molecular dynamics simulation. Force fields were chosen that accurately predict the density and zero-field dielectric constant with respect to experiment. The simulation results for ε( E ) for field strengths up to 0.4 V/Å are calculated and fit to the Booth model, which is a standard functional form for the dependence of the dielectric constant on electric field. For PC and DMC, the Booth model gives an excellent representation of the data at all electric fields studied. For EC and MeCN, the Booth model works well at lower field strengths (up to 0.15 and 0.2 V/Å, respectively), but at the higher electric fields, these systems are observed to crystallize, a phenomenon referred to as electrofreezing . In conclusion, this work will provide useful input data for the continuum modeling of devices, such as electric double-layer capacitors, that utilize organic electrolyte solvents.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Monitoring Fracture Saturation With Internal Seismic Sources and Twin Neural Networks

Seismic coda-wave analysis is a well-developed method for detecting subtle physical changes in complex media by measuring arrival times in the late-arriving energy from multiply scattered or reflected waves. However, a challenge arises when multiply scattered waves are not sufficiently separated in time from the direct arrivals to provide a clear coda wave train. Additional complications for monitoring changes in fracture systems arise when the signals originate from unsynchronized internal sources, such as natural or induced seismicity, from acoustic emission, or from transportable intra-fracture sources (chattering dust), that generate uncontrolled signals that vary in arrival time, amplitude and frequency content. Here, we use a twin neural network (TNN also known as a Siamese neural network) for dimensionality reduction to analyze signals from chattering dust to classify the fluid saturation state of a synthetic fracture system. The TNN with shared weights generates a low-dimensional representation of the data input by minimizing contrastive loss, serving as the input to a multiclass classifier that accurately classifies whether multiple fractures in a fracture system are fully saturated or partially saturated, or whether a change in saturation has occurred in different fractures in the system. Furthermore, these results show that information buried in unresolved codas from uncontrolled sources can be extracted using machine learning to monitor the evolution of fracture systems caused by physical and chemical processes even when the scattered and direct wave fields overlap.

58 GEOSCIENCES↗

Discovering equations that govern experimental materials stability under environmental stress using scientific machine learning

Abstract While machine learning (ML) in experimental research has demonstrated impressive predictive capabilities, extracting fungible knowledge representations from experimental data remains an elusive task. In this manuscript, we use ML to infer the underlying differential equation (DE) from experimental data of degrading organic-inorganic methylammonium lead iodide (MAPI) perovskite thin films under environmental stressors (elevated temperature, humidity, and light). Using a sparse regression algorithm, we find that the underlying DE governing MAPI degradation across a broad temperature range of 35 to 85 °C is described minimally by a second-order polynomial. This DE corresponds to the Verhulst logistic function, which describes reaction kinetics analogous to self-propagating reactions. We examine the robustness of our conclusions to experimental variance and Gaussian noise and describe the experimental limits within which this methodology can be applied. Our study highlights the promise and challenges associated with ML-aided scientific discovery by demonstrating its application in experimental chemical and materials systems.

36 MATERIALS SCIENCE↗

Augmenting machine learning of energy landscapes with local structural information

We present a machine learning approach for accurately predicting formation energies of binary compounds in the context of crystal structure predictions. The success of any machine learning model depends significantly on the choice of representation used to encode the relevant physical information into machine-learnable data. We test different representation schemes based on partial radial and angular distribution functions (RDF+ADF) on Al–Ni and Cd–Te structures generated using our genetic algorithm for structure prediction. We observe a remarkable improvement in predictive accuracy upon transitioning from global to atom-centered representations, resulting in a threefold decrease in prediction errors. We show that a support vector regression model using a combination of atomic radial and angular distribution functions performs best at the formation energy prediction task, providing small root mean squared errors of 3.9 meV/atom and 10.9 meV/atom for Al–Ni and Cd–Te, respectively. We test the performance of our models against common traditional descriptors and find that RDF- and ADF-based representations significantly outperform many of those in the prediction of formation energies. The high accuracy of predictions makes our machine learning models great candidates for the exploration of energy landscapes.

Honrao, Shreyas J. (ORCID:0000000166155257)↗

MEASURING THE ULTIMATE HALO MASS OF GALAXY CLUSTERS: REDSHIFTS AND MASS PROFILES FROM THE HECTOSPEC CLUSTER SURVEY (HeCS)

The infall regions of galaxy clusters represent the largest gravitationally bound structures in a {Lambda}CDM universe. Measuring cluster mass profiles into the infall regions provides an estimate of the ultimate mass of these halos. We use the caustic technique to measure cluster mass profiles from galaxy redshifts obtained with the Hectospec Cluster Survey (HeCS), an extensive spectroscopic survey of galaxy clusters with MMT/Hectospec. We survey 58 clusters selected by X-ray flux at 0.1 < z < 0.3. The survey includes 22,680 unique MMT/Hectospec redshifts for individual galaxies; 10,145 of these galaxies are cluster members. For each cluster, we acquired high signal-to-noise spectra for {approx}200 cluster members and a comparable number of foreground/background galaxies. The cluster members trace out infall patterns around the clusters. The members define a very narrow red sequence. We demonstrate that the determination of velocity dispersion is insensitive to the inclusion of bluer members (a small fraction of the cluster population). We apply the caustic technique to define membership and estimate the mass profiles to large radii. The ultimate halo mass of clusters (the mass that remains bound in the far future of a {Lambda}CDM universe) is on average (1.99 {+-} 0.11)M{sub 200}, a new observational cosmological test in essential agreement with simulations. Summed profiles binned in M{sub 200} and in L{sub X} demonstrate that the predicted Navarro-Frenk-White form of the density profile is a remarkably good representation of the data in agreement with weak lensing results extending to large radius. The concentration of these summed profiles is also consistent with theoretical predictions.

79 ASTRONOMY AND ASTROPHYSICS↗