Interpretable Data-driven Plasticity Models with Symbolic Regression.
Abstract not provided.
Stewart, Kristen↗
SEARCH · Search NASA
Results for “Model interpretation”
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.
A Review on the State of the Art of Machine Learning and Satellite Imaging: Detecting Scene Changes in Selected Nuclear Fuel Cycle Datasets
The timely detection of clandestine nuclear facilities is one of the greatest challenges faced by the International Atomic Energy Agency’s (IAEA). Idaho National Laboratory is currently applying machine learning (ML) to existing satellite imagery (SI) datasets to find facilities within the nuclear fuel life cycle, with primary focus placed on identifying critical predecessor (i.e., fuel fabrication and fuel enrichment) and successor (i.e., nuclear power plants) facilities. This could provide a satellite image methodology that the IAEA could leverage to discover clandestine facilities. The work presented in this paper describes the evolution of a workflow developed by this team for object detection related to critical infrastructure by expanding that workflow for the purpose of identifying nuclear fuel cycle components and automating dependency assessments. This will be done using two methods housed within a single pipeline. The first method involves implementing a DenseNet161 convolutional neural network to classify the images and explain the results using Local Interpretable Model-Agnostic Explanations (LIME). The second method implements You Only Look Once version 5 (YoloV5), to detect objects within images, provide a probability for the detection, and provide a bounding box that corresponds to the object of interest. The results of this work are anticipated to provide a clear picture of this portion of the nuclear fuel cycle and perform as a stand-alone tool for image assessment that can be expanded to additional fuel cycle components and implemented in international safeguards and national security domains. This capability addresses the IAEA’s need to detect undeclared nuclear materials and activities within a state while encompassing the entire nuclear fuel cycle.
97 MATHEMATICS AND COMPUTING↗
Leak Detection and Sensor Importance Within a Solvent Extraction Process Abstract
In anticipation of the Special Nuclear Material test bed (Beartooth), Idaho National Laboratory has developed a smaller, multi-sensor system for analyzing the solvent extraction process. These systems will allow for research into nuclear fuel processing operations. The multi-sensor system consists of a row of centrifugal contactors and allows for measurement sources that are not traditionally used in the solvent extraction process to be explored including temperature, vibration, acoustics, pH, color, flow, and motor current. Currently, the solvent extraction process is very labor intensive and requires vigilant operators to identify the occurrence of leaks, which can be common during startup or after any change to the system. This study aims to locate leaks using non-traditional measurement sources then to identify which signals were of greatest importance in making this classification using Local Interpretable Model-agnostic Explanations. These results can be used to help solvent extraction process operators detect leaks and to inform future test beds designers to which sensors contain relevant, actionable information in this scenario.
46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
Modelling asteroid brightness variations. II - The interpretability of light curves and phase curves
Light curves and phase curves have been computed for various asteroid models using the Lumme-Bowell (1981) scattering law. The effects of the scattering parameters on light curves were found to be almost negligible for homogeneous surfaces. The effects on phase curves were more distinct, but changing any of the scattering parameters affects the phase curves in a very similar way, making it impossible to find a unique set of parameter values corresponding to a given phase curve. Light curve amplitudes, on the other hand, depend very strongly on body shape. At least in the case of a triaxial ellipsoid it is possible to determine the axial ratios. Some observed irregularities of light curves can also be modelled easily, but the uniqueness of such models is far from obvious.
Karttunen, H.↗
Interpretable and flexible non-intrusive reduced-order models using reproducing kernel Hilbert spaces
This paper develops an interpretable, non-intrusive reduced-order modeling technique using regularized kernel interpolation. Existing non-intrusive approaches approximate the dynamics of a reduced-order model (ROM) by solving a data-driven least-squares regression problem for low-dimensional matrix operators. Our approach instead leverages regularized kernel interpolation, which yields an optimal approximation of the ROM dynamics from a user-defined reproducing kernel Hilbert space. We show that our kernel-based approach can produce interpretable ROMs whose structure mirrors full-order model structure by embedding judiciously chosen feature maps into the kernel. The approach is flexible and allows a combination of informed structure through feature maps and closure terms via more general nonlinear terms in the kernel. We also derive a computable a posteriori error bound that combines standard error estimates for intrusive projection-based ROMs and kernel interpolants. In conclusion, the approach is demonstrated in several numerical experiments that include comparisons to operator inference using both proper orthogonal decomposition and quadratic manifold dimension reduction.
Data-driven model reduction↗
The dynamics of simian immunodeficiency virus after depletion of CD8+ cells
Human immunodeficiency virus infection is still one of the most important causes of morbidity and mortality in the world, with a disproportionate human and economic burden especially in poorer countries. Despite many years of intense research, an aspect that still is not well understood is what (immune) mechanisms control the viral load during the prolonged asymptomatic stage of infection. Because CD8+ T cells have been implicated in this control by multiple lines of evidence, there has been a focus on understanding the potential mechanisms of action of this immune effector population. One type of experiment used to this end has been depleting these cells with monoclonal antibodies in the simian immunodeficiency virus-macaque model and then studying the effect of that depletion on the viral dynamics. In this paper, we review what these experiments have told us. Finally, we emphasize modeling studies to interpret the changes in viral load observed in these experiments, including discussion of alternative models, assumptions and interpretations, as well as potential future experiments.
59 BASIC BIOLOGICAL SCIENCES↗
On the Mechanism of Bilayer Separation by Extrusion, or Why Your LUVs Are Not Really Unilamellar
Extrusion through porous filters is a widely used method for preparing biomimetic model membranes. Of primary importance in this approach is the efficient production of single bilayer (unilamellar) vesicles that eliminate the influence of interlamellar interactions and strictly define the bilayer surface area available to external reagents such as proteins. Sub-microscopic vesicles produced using extrusion are widely assumed to be unilamellar, and large deviations from this assumption would dramatically impact interpretations from many model membrane experiments. In this work, using three probe-free methods—small-angle X-ray and neutron scattering (SAXS and SANS) and cryogenic electron microscopy (cryoEM)—we report unambiguous evidence of extensive multilamellarity in extruded vesicles composed of neutral phosphatidylcholine lipids, including for the common case of neutral lipids dispersed in physiological buffer and extruded through 100 nm diameter pores. In such preparations, only ~35% of lipids are externally accessible, and this fraction is highly dependent on preparation conditions. Charged lipids promote unilamellarity, as does decreasing solvent ionic strength, indicating the importance of electrostatic interactions in determining the lamellarity of extruded vesicles. Smaller extrusion pore sizes also robustly increase the fraction of unilamellar vesicles, suggesting a role for membrane bending. Taken together, these observations suggest a mechanistic model for extrusion, wherein formation of unilamellar vesicles involves competition between bilayer bending and adhesion energies. The findings presented here have wide-ranging implications for the design and interpretation of model membrane studies, especially ensemble-averaged observations relying on the assumption of unilamellarity.
59 BASIC BIOLOGICAL SCIENCES↗
Interpretable Convolutional Learning Classifier System (C-LCS) for Higher Dimensional Datasets
The purpose of this paper is to devise an interpretable hybrid classification model for Convolutional Neural Networks (CNN) and a Learning Classifier System (LCS). The presented hybrid system integrates the fundamental attributes from both types of these classifiers. In the proposed hybrid model CNN works as an automatic feature extractor, and LCS works to provide interpretable rule-based classification results. Although LCS has limitations working on higher dimensional datasets, we resolve this limitation by using CNN as a feature extractor. The other concept of the non-interpretability of CNN is addressed by using the LCS rule. Furthermore, our experiment with higher dimensional datasets like CIFAR-10 and Fashion-MNIST shows that extended LCS provides comparable performance to the standard neural network model while also providing interpretable results. We named this extended LCS method Convolutional Learning Classifier Cystem (C-LCS).
Jelani Owens↗
An Investigation of LES Wall Modeling for Rayleigh–Bénard Convection via Interpretable and Physics-Aware Feedforward Neural Networks with DNS
Abstract The traditional approach of using the Monin–Obukhov similarity theory (MOST) to model near-surface processes in large-eddy simulations (LESs) can lead to significant errors in natural convection. In this study, we propose an alternative approach based on feedforward neural networks (FNNs) trained on output from direct numerical simulation (DNS). To evaluate the performance, we conduct both a priori and a posteriori tests. In the a priori (offline) tests, we compare the statistics of the surface shear stress and heat flux, computed from filtered DNS input variables, to the stress and flux obtained from the filtered DNS. Additionally, we investigate the importance of various input features using the Shapley additive explanations value and the conditional average of the filter grid cells. In the a posteriori (online) tests, we implement the trained models in the System for Atmospheric Modeling (SAM) LES and compare the LES-generated surface shear stress and heat flux with those in the DNS. Our findings reveal that vertical velocity, a traditionally overlooked flow quantity, is one of the most important input features for determining the wall fluxes. Increasing the number of input features improves the a priori test results but does not always improve the model performance in the a posteriori tests because of the differences in input variables between the LES and DNS. Last, we show that physics-aware FNN models trained with logarithmic and scaled parameters can well extrapolate to more intense convection scenarios than in the training dataset, whereas those trained with primitive flow quantities cannot. Significance Statement The traditional near-surface turbulence model, based on a shear-dominated boundary layer flow, does not represent near-surface turbulence in natural convection. Using a feedforward neural network (FNN), we can construct a more accurate model that better represents the near-surface turbulence in various flows and reveals previously overlooked controlling factors and process interactions. Our study shows that the FNN-generated models outperform the traditional model and highlight the importance of the near-surface vertical velocity. Furthermore, the physics-aware FNN models exhibit the potential to extrapolate to convective flows of various intensities beyond the range of the training dataset, suggesting their broader applicability for more accurate modeling of near-surface turbulence.
54 ENVIRONMENTAL SCIENCES↗
Experimental Observations of the Topology of Convolutional Neural Network Activations
Topological data analysis (TDA) is a branch of computational mathematics, bridging algebraic topology and data science, that provides compact, noise-robust representations of complex structures. Deep neural networks (DNNs) learn millions of parameters associated with a series of transformations defined by the model architecture resulting in high-dimensional, difficult to interpret internal representations of input data. As DNNs become more ubiquitous across multiple sectors of our society, there is increasing recognition that mathematical methods are needed to aid analysts, researchers, and practitioners in understanding and interpreting how these models' internal representations relate to the final classification. In this paper we apply cutting edge techniques from TDA with the goal of gaining insight towards interpretability of convolutional neural networks used for image classification. We use two common TDA approaches to explore several methods for modeling hidden layer activations as high-dimensional point clouds, and provide experimental evidence that these point clouds capture valuable structural information about the model's process. First, we demonstrate that a distance metric based on persistent homology can be used to quantify meaningful differences between layers and discuss these distances in the broader context of existing representational similarity metrics for neural network interpretability. Second, we show that a mapper graph can provide semantic insight as to how these models organize hierarchical class knowledge at each layer. These observations demonstrate that TDA is a useful tool to help deep learning practitioners unlock the hidden structures of their models.
topological data analysis, deep learning↗
Interpretation of TOMS Observations of Tropical Tropospheric Ozone with a Global Model and In Situ Observations
We interpret the distribution of tropical tropospheric ozone columns (TTOCs) from the Total Ozone Mapping Spectrometer (TOMS) by using a global three-dimensional model of tropospheric chemistry (GEOS-CHEM) and additional information from in situ observations. The GEOS-CHEM TTOCs capture 44% of the variance of monthly mean TOMS TTOCs from the convective cloud differential method (CCD) with no global bias. Major discrepancies are found over northern Africa and south Asia where the TOMS TTOCs do not capture the seasonal enhancements from biomass burning found in the model and in aircraft observations. A characteristic feature of these northern topical enhancements, in contrast to southern tropical enhancements, is that they are driven by the lower troposphere where the sensitivity of TOMS is poor due to Rayleigh scattering. We develop an efficiency correction to the TOMS retrieval algorithm that accounts for the variability of ozone in the lower troposphere. This efficiency correction increases TTOC's over biomass burning regions by 3-5 Dobson units (DU) and decreases them by 2-5 DU over oceanic regions, improving the agreement between CCD TTOCs and in situ observations. Applying the correction to CCD TTOCs reduces by approximately DU the magnitude of the "tropical Atlantic paradox" [Thompson et al, 2000], i.e. the presence of a TTOC enhancement over the southern tropical Atlantic during the northern African biomass burning season in December-February. We reproduce the remainder of the paradox in the model and explain it by the combination of upper tropospheric ozone production from lightning NOx, peristent subsidence over the southern tropical Atlantic as part of the Walker circulation, and cross-equatorial transport of upper tropospheric ozone from northern midlatitudes in the African "westerly duct." These processes in the model can also account for the observed 13-17 DU persistent wave-1 pattern in TTOCs with a maximum above the tropical Atlantic and a minimum over the tropical Pacific during all seasons. The photochemical effects of mineral dust have only a minor role on the modeled distribution of TTOCs, including over northern Africa, due to multiple competing effects. The photochemical effects of mineral dust globally decease annual mean OH concentrations by 9%. A global lightning NOx source of 6 Tg N yr(sup -1) in the model produces a simulation that is most consistent with TOMS and in situ observations.
Martin, Randall V.↗
Search for supersymmetry in events with at least one photon, missing transverse momentum, and large transverse event activity in proton-proton collisions at $ \sqrt{s}=13 $ TeV
A search for physics beyond the standard model in final states with at least one photon, large transverse momentum imbalance, and large total transverse event activity is presented. Such topologies can be produced in gauge-mediated supersymmetry models in which pair-produced gluinos or squarks decay to photons and gravitinos via short-lived neutralinos. The data sample corresponds to an integrated luminosity of 35.9 fb$^{−1}$ of proton-proton collisions at $ \sqrt{s}=13 $ TeV recorded by the CMS experiment at the LHC in 2016. No significant excess of events above the expected standard model background is observed. The data are interpreted in simplified models of gluino and squark pair production, in which gluinos or squarks decay via neutralinos to photons. Gluino masses of up to 1.50-2.00 TeV and squark masses up to 1.30-1.65 TeV are excluded at 95% confidence level, depending on the neutralino mass and branching fraction.
72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
Catalytic Resonance Theory: Forecasting the Flow of Programmable Catalytic Loops
Chemical transformations on catalyst surfaces occur through series and parallel reaction pathways. These complex networks and their behavior can be most simply evaluated through a three-species surface reaction loop (A* to B* to C* to A*) that is internal to the overall chemical reaction. Application of an oscillating dynamic catalyst to this reactive loop has been shown to exhibit one of three types of behavior: (1) a positive net flux of molecules about the loop in the clockwise direction, (2) a negative net flux of molecules about the loop in the counterclockwise direction, or (3) negligible flux of molecules about the loop at the limit cycle of reaction. Three-species surface loops were simulated with microkinetic modeling to assess the reaction loop behavior resulting from a catalytic surface oscillating between two or more catalyst surface energy states. Selected input parameters for the simulations spanned an 11-dimensional parameter space using 127 688 different parameter combinations. Their converged limit cycle solutions were analyzed for their loop turnover frequencies, the majority of which were found to be approximately zero. Classification and regression machine learning models were trained to predict the sign and magnitude of the loop turnover frequency and successfully performed above accessible baselines. Notably, the classification models exhibited a baseline weighted F1 score of 0.49, whereas trained models achieved weighted F1 scores of 0.94 and 0.96 when trained on the parameters used to define the simulations and derived rate constants, respectively. The trained models successfully predicted catalytic loop behavior, and interpretation of these models revealed all input parameters to be important for the prediction and performance of each model.
37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
Evaluation of historical precipitation interannual variability in CMIP6 over the United States
Interannual precipitation variability profoundly influences society via its effects on agriculture, water resources, infrastructure, and disaster risks. In this study, we use daily in situ precipitation observations from the global historical climatology network-daily (GHCN-D) to assess the ability of 21 Coupled Model Intercomparison Project Phase 6 (CMIP6) models, including the 50-member fifth-generation Canadian Earth System Model single model initial-condition large ensemble (CanESM5_SMILE), to realistically simulate historical interannual precipitation variability trends within 17 regions of the contiguous United States (CONUS). We assess how accurately the CMIP6 simulations align with observational data across annual, summer, and winter periods, focusing on four key hydrometeorological metrics, including interannual precipitation variability, relative interannual precipitation variability (coefficient of variation), annual mean precipitation, and annual wet day frequency. Our findings reveal that CMIP6 ensemble members generally reproduce the spatial patterns of observed trends in annual mean precipitation. In most regions, models agree well with the signs of observed changes in annual mean precipitation, though discrepancies in trend magnitude are evident. Further, observed trends in winter mean precipitation broadly exhibit a spatial pattern similar to that of the observed annual mean. However, analysis of the CanESM5_SMILE shows that trends in precipitation variability may primarily be the result of model-simulated internal variability, suggesting caution in interpreting multi-model single-realization ensemble results. Challenges in accurately simulating interannual precipitation variability underscore the need for ongoing model refinement and validation to enhance climate projections, especially in regions vulnerable to extreme precipitation events.
54 ENVIRONMENTAL SCIENCES↗
Search for Dark Matter Particles Produced in Association with a Top Quark Pair at s = 13 TeV
A search is performed for dark matter particles produced in association with a top quark pair in proton-proton collisions at $\sqrt{s} =$ 13 TeV. The data correspond to an integrated luminosity of 35.9 fb$^{-1}$ recorded by the CMS detector at the LHC. No significant excess over the standard model expectation is observed. The results are interpreted using simplified models of dark matter production via spin-0 mediators that couple to dark matter particles and to standard model quarks, providing constraints on the coupling strength between the mediator and the quarks. In conclusion, these are the most stringent collider limits to date for scalar mediators, and the most stringent for pseudoscalar mediators at low masses.
72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
Intercomparison and interpretation of climate feedback processes in 19 atmospheric general circulation models
The present study provides an intercomparison and interpretation of climate feedback processes in 19 atmospheric general circulation models. This intercomparison uses sea surface temperature change as a surrogate for climate change. The interpretation of cloud-climate interactions is given special attention. A roughly threefold variation in one measure of global climate sensitivity is found among the 19 models. The important conclusion is that most of this variation is attributable to differences in the models' depiction of cloud feedback, a result that emphasizes the need for improvements in the treatment of clouds in these models if they are ultimately to be used as reliable climate predictors. It is further emphazied that cloud feedback is the consequence of all interacting physical and dynamical processes in a general circulation model. The result of these processes is to produce changes in temperature, moisture distribution, and clouds which are integrated into the radiative response termed cloud feedback.
Cess, R. D.↗
Spatial reasoning in remotely sensed data
Photointerpreters employ a variety of implicit spatial models to provide interpretations from remotely sensed aerial or satellite imagery. In this paper one application is illustrated: how ridges and valleys can be automatically interpreted from Landsat imagery of a mountainous area, and how a relative elevation terrain model can be constructed from this interpretation. How to examine valleys for the possible presence of streams or rivers is shown, and how a spatial relational model can be set up to make a final interpretation of the river drainage network is explored.
Campbell, J.↗
Physical Interpretation of Early Battery Life Prediction Models
Early battery life prediction models are most useful for R&D if they help us understand the early changes in battery electrochemical response that correspond with long-term degradation and failure. Linear regression models such as Fused lasso and Partial Least Squares can fit coefficients directly to high-dimensional electrochemical data like capacity-voltage and ΔV–state-of-charge, i.e., Q(V) and ΔV(SOC) curves, learning coefficients that can be physically interpreted. We leverage the ISU-ILCC battery aging data set to learn high-dimensional coefficients for early battery life prediction from traditional slow-rate capacity check data, demonstrating learning on Q(V), d Q· d V −1 , and ΔV(SOC) curves. A thorough study on the dependence of coefficient values on train/test size and data preprocessing methods is made, demonstrating the reliability of high-dimensional regression approaches unless very small amounts of data are used for model training. For this data set, coefficients from Q(V) and d Q· d V −1 models highlight changes in electrode stoichiometry due to lithium loss, while ΔV(SOC) coefficients highlight changes in positive electrode diffusivity due to particle cracking as well as electrode stoichiometry shifts. By directly interpreting the coefficients of a regression model, we make physical insights into battery degradation mechanisms without requiring the assumptions of traditional battery data analysis methods.
25 ENERGY STORAGE↗