Engineering Papers⌕ Search

SEARCH · Engineering Papers

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.

At least 595 records · Page 33

Evidential Deep Learning: Enhancing Predictive Uncertainty Estimation for Earth System Science Applications

Abstract Robust quantification of predictive uncertainty is a critical addition needed for machine learning applied to weather and climate problems to improve the understanding of what is driving prediction sensitivity. Ensembles of machine learning models provide predictive uncertainty estimates in a conceptually simple way but require multiple models for training and prediction, increasing computational cost and latency. Parametric deep learning can estimate uncertainty with one model by predicting the parameters of a probability distribution but does not account for epistemic uncertainty. Evidential deep learning, a technique that extends parametric deep learning to higher-order distributions, can account for both aleatoric and epistemic uncertainties with one model. This study compares the uncertainty derived from evidential neural networks to that obtained from ensembles. Through applications of the classification of winter precipitation type and regression of surface-layer fluxes, we show evidential deep learning models attaining predictive accuracy rivaling standard methods while robustly quantifying both sources of uncertainty. We evaluate the uncertainty in terms of how well the predictions are calibrated and how well the uncertainty correlates with prediction error. Analyses of uncertainty in the context of the inputs reveal sensitivities to underlying meteorological processes, facilitating interpretation of the models. The conceptual simplicity, interpretability, and computational efficiency of evidential neural networks make them highly extensible, offering a promising approach for reliable and practical uncertainty quantification in Earth system science modeling. To encourage broader adoption of evidential deep learning, we have developed a new Python package, Machine Integration and Learning for Earth Systems (MILES) group Generalized Uncertainty for Earth System Science (GUESS) (MILES-GUESS) ( https://github.com/ai2es/miles-guess ), that enables users to train and evaluate both evidential and ensemble deep learning. Significance Statement This study demonstrates a new technique, evidential deep learning, for robust and computationally efficient uncertainty quantification in modeling the Earth system. The method integrates probabilistic principles into deep neural networks, enabling the estimation of both aleatoric uncertainty from noisy data and epistemic uncertainty from model limitations using a single model. Our analyses reveal how decomposing these uncertainties provides valuable insights into reliability, accuracy, and model shortcomings. We show that the approach can rival standard methods in classification and regression tasks within atmospheric science while offering practical advantages such as computational efficiency. With further advances, evidential networks have the potential to enhance risk assessment and decision-making across meteorology by improving uncertainty quantification, a longstanding challenge. This work establishes a strong foundation and motivation for the broader adoption of evidential learning, where properly quantifying uncertainties is critical yet lacking.

Schreck, John S.↗

Search for higgsinos in compressed mass spectra using low-momentum tracks in pp collisions at s=13 TeV with the ATLAS detector

This paper presents two searches for the electroweak production of higgsinos with compressed mass spectra using 140 fb−1 of s=13$$ \sqrt{s}=13 $$ TeV proton-proton collision data collected by the ATLAS experiment at the Large Hadron Collider. Events are required to feature an energetic jet, large missing transverse momentum, and at least one low-momentum charged particle that serves as a candidate higgsino decay product. In the first search, targeting higgsino mass splittings in the range of 0.3–1 GeV, the higgsinos are expected to predominantly decay into pions that are identified as low-momentum charged particles with large transverse impact parameters due to the long higgsino lifetime (cτ ≈ ?(0.1–10 mm)), and neural networks are used to discriminate between signal and background processes. The second search targets larger mass splittings in the range of 1–3 GeV, where the higgsinos are expected to decay promptly into low-momentum leptons, one of which is identified by dedicated low-momentum electron or muon taggers based on neural networks utilising tracking and calorimeter information. No significant excess above the Standard Model prediction is observed in either search and the results are interpreted within simplified models, to set lower limits on the masses of the higgsino-like charginos and neutralinos. Together, these searches exclude chargino masses below 126 GeV at 95% confidence level for mass splittings between the chargino and lightest neutralino in the range of 0.3–2 GeV. This represents the first ATLAS constraints in a portion of this parameter space and surpasses the limits previously set by other experiments.

Aad, G↗

Modeling sequential context effects in diagnostic interpretation of screening mammograms

Prior research has shown that physicians’ medical decisions can be influenced by sequential context, particularly in cases where successive stimuli exhibit similar characteristics when analyzing medical images. This type of systematic error is known to psychophysicists as sequential context effect as it indicates that judgments are influenced by features of and decisions about the preceding case in the sequence of examined cases, rather than being based solely on the peculiarities unique to the present case. We determine if radiologists experience some form of context bias, using screening mammography as the use case. To this end, we explore correlations between previous perceptual behavior and diagnostic decisions and current decisions. We hypothesize that a radiologist’s visual search pattern and diagnostic decisions in previous cases are predictive of the radiologist’s current diagnostic decisions. To test our hypothesis, we tasked 10 radiologists of varied experience to conduct blind reviews of 100 four-view screening mammograms. Eye-tracking data and diagnostic decisions were collected from each radiologist under conditions mimicking clinical practice. Perceptual behavior was quantified using the fractal dimension of gaze scanpath, which was computed using the Minkowski–Bouligand box-counting method. To test the effect of previous behavior and decisions, we conducted a multifactor fixed-effects ANOVA. Further, to examine the predictive value of previous perceptual behavior and decisions, we trained and evaluated a predictive model for radiologists’ current diagnostic decisions. ANOVA tests showed that previous visual behavior, characterized by fractal analysis, previous diagnostic decisions, and image characteristics of previous cases are significant predictors of current diagnostic decisions. Additionally, predictive modeling of diagnostic decisions showed an overall improvement in prediction error when the model is trained on additional information about previous perceptual behavior and diagnostic decisions.

60 APPLIED LIFE SCIENCES↗

Interpretable machine learning for knowledge generation in heterogeneous catalysis

Most applications of machine learning in heterogeneous catalysis thus far have used black-box models to predict computable physical properties (descriptors), such as adsorption or formation energies, that can be related to catalytic performance (that is, activity or stability). Here, extracting meaningful physical insights from these black-box models has proved challenging, as the internal logic of these black-box models is not readily interpretable due to their high degree of complexity. Interpretable machine learning methods that merge the predictive capacity of black-box models with the physical interpretability of physics-based models offer an alternative to black-box models. In this Perspective, we discuss the various interpretable machine learning methods available to catalysis researchers, highlight the potential of interpretable machine learning to accelerate hypothesis formation and knowledge generation, and outline critical challenges and opportunities for interpretable machine learning in heterogeneous catalysis.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

The Structural Basis for Partitioning of the XRCC1/DNA Ligase III-alpha BRCT-mediated Dimer Complexes

The ultimate step common to almost all DNA repair pathways is the ligation of the nicked intermediate to form contiguous double-stranded DNA. In the mammalian nucleotide and base excision repair pathways, the ligation step is carried out by ligase III-{alpha}. For efficient ligation, ligase III-{alpha} is constitutively bound to the scaffolding protein XRCC1 through interactions between the C-terminal BRCT domains of each protein. Although structural data for the individual domains has been available, no structure of the complex has been determined and several alternative proposals for this interaction have been advanced. Interpretation of the models is complicated by the formation of homodimers that, depending on the model, may either contribute to, or compete with heterodimer formation. We report here the structures of both homodimer complexes as well as the heterodimer complex. Structural characterization of the heterodimer formed from a longer XRCC1 BRCT domain construct, including residues comprising the interdomain linker region, revealed an expanded heterodimer interface with the ligase III-{alpha} BRCT domain. This enhanced linker-mediated binding interface plays a significant role in the determination of heterodimer/homodimer selectivity. These data provide fundamental insights into the structural basis of BRCT-mediated dimerization, and resolve questions related to the organization of this important repair complex.

59 BASIC BIOLOGICAL SCIENCES↗

Higgs Portal Vector Dark Matter Interpretation: Review of Effective Field Theory Approach and Ultraviolet Complete Models

In this report a review of the Higgs portal-vector dark matter interpretation of the spin-independent dark matter nu-cleon elastic scattering cross section is presented, where the invisible Higgs decay width measured at the LHC is used. Effective Field Theory and ultraviolet complete models are discussed. LHC interpretations show only the scalar and Majorana dark matter scenarios; we propose including interpretation for vec-tor dark matter in the EFT and UV completion theoretical framework. In addition, our studies suggest an extension of the LHC dark matter interpretations to the sub-GeV regime.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Multiscale graph neural network autoencoders for interpretable scientific machine learning

The goal of this work is to address two limitations in autoencoder-based models: latent space interpretability and compatibility with unstructured meshes. This is accomplished here with the development of a novel graph neural network (GNN) autoencoding architecture with demonstrations on complex fluid flow applications. To address the first goal of interpretability, the GNN autoencoder achieves reduction in the number nodes in the encoding stage through an adaptive graph reduction procedure. Further, this reduction procedure essentially amounts to flowfieldconditioned node sampling and sensor identification, and produces interpretable latent graph representations tailored to the flowfield reconstruction task in the form of so-called masked fields. These masked fields allow the user to (a) visualize where in physical space a given latent graph is active, and (b) interpret the time-evolution of the latent graph connectivity in accordance with the time-evolution of unsteady flow features (e.g. recirculation zones, shear layers) in the domain. To address the goal of unstructured mesh compatibility, the autoencoding architecture utilizes a series of multi-scale message passing (MMP) layers, each of which models information exchange among node neighborhoods at various lengthscales. The MMP layer, which augments standard single-scale message passing with learnable coarsening operations, allows the decoder to more efficiently reconstruct the flowfield from the identified regions in the masked fields. Analysis of latent graphs produced by the autoencoder for various model settings are conducted using unstructured snapshot data sourced from large-eddy simulations in a backward-facing step (BFS) flow configuration with an OpenFOAM-based flow solver at high Reynolds numbers.

97 MATHEMATICS AND COMPUTING↗

Integrating multi-modal remote sensing, deep learning, and attention mechanisms for yield prediction in plant breeding experiments

In both plant breeding and crop management, interpretability plays a crucial role in instilling trust in AI-driven approaches and enabling the provision of actionable insights. The primary objective of this research is to explore and evaluate the potential contributions of deep learning network architectures that employ stacked LSTM for end-of-season maize grain yield prediction. A secondary aim is to expand the capabilities of these networks by adapting them to better accommodate and leverage the multi-modality properties of remote sensing data. In this study, a multi-modal deep learning architecture that assimilates inputs from heterogeneous data streams, including high-resolution hyperspectral imagery, LiDAR point clouds, and environmental data, is proposed to forecast maize crop yields. The architecture includes attention mechanisms that assign varying levels of importance to different modalities and temporal features that, reflect the dynamics of plant growth and environmental interactions. The interpretability of the attention weights is investigated in multi-modal networks that seek to both improve predictions and attribute crop yield outcomes to genetic and environmental variables. This approach also contributes to increased interpretability of the model's predictions. The temporal attention weight distributions highlighted relevant factors and critical growth stages that contribute to the predictions. The results of this study affirm that the attention weights are consistent with recognized biological growth stages, thereby substantiating the network's capability to learn biologically interpretable features. Accuracies of the model's predictions of yield ranged from 0.82-0.93 R 2 ref in this genetics-focused study, further highlighting the potential of attention-based models. Further, this research facilitates understanding of how multi-modality remote sensing aligns with the physiological stages of maize. The proposed architecture shows promise in improving predictions and offering interpretable insights into the factors affecting maize crop yields, while demonstrating the impact of data collection by different modalities through the growing season. By identifying relevant factors and critical growth stages, the model's attention weights provide valuable information that can be used in both plant breeding and crop management. The consistency of attention weights with biological growth stages reinforces the potential of deep learning networks in agricultural applications, particularly in leveraging remote sensing data for yield prediction. To the best of our knowledge, this is the first study that investigates the use of hyperspectral and LiDAR UAV time series data for explaining/interpreting plant growth stages within deep learning networks and forecasting plot-level maize grain yield using late fusion modalities with attention mechanisms.

59 BASIC BIOLOGICAL SCIENCES↗

Fracture propagation in Indiana Limestone interpreted via linear softening cohesive fracture model

In this study, we examine the use of a linear softening cohesive fracture model (LCFM) to predict single-trace fracture growth in short-rod (SR) and notched 3-point-bend (N3PB) test configurations in Indiana Limestone. The broad goal of this work is to (a) understand the underlying assumptions of LCFM and (b) use experimental similarities and deviations from the LCFM to understand the role of loading paths of tensile fracture propagation. Cohesive fracture models are being applied in prediction of structural and subsurface fracture propagation in geomaterials. They lump the inelastic processes occurring during fracture propagation into a thin zone between elastic subdomains. LCFM assumes that the cohesive zone initially deforms elastically to a maximum tensile stress ( σ max ) and then softens linearly from the crack opening width at σ max to zero stress at a critical crack opening width w 1 . Using commercial finite element software, we developed LCFMs for the SR and N3PB configurations. After fixing σ max with results from cylinder splitting tests and finding an initial Young's modulus ( E ) with unconfined compressive strength tests, we manually calibrate E and w 1 in the SR model against an envelope of experimental data. We apply the calibrated LCFM parameters in the N3PB geometry and compare the model against an envelope of N3PB experiments. For accurate simulation of fracture propagation, simulated off-crack stresses are high enough to require inclusion of damage. Different elastic moduli are needed in tension and compression. We hypothesize that the timing and location of shear versus extensional micromechanical failures control the qualitative macroscopic force-versus-displacement response in different tests. For accurate prediction, the LCFM requires a constant style of failure, which the SR configuration maintains until very late in deformation. The N3PB configuration does not maintain this constancy. To be broadly applicable between geometries and failure styles, the LCFM would require additional physics, possibly including elastoplastic damage in the bulk material and more complicated cohesive softening models.

58 GEOSCIENCES↗

Search for new phenomena with large jet multiplicities and missing transverse momentum using large-radius jets and flavour-tagging at ATLAS in 13 TeV pp collisions

A search is presented for particles that decay producing a large jet multiplicity and invisible particles. The event selection then applies a veto on the presence of isolated electrons or muons and additional requirements on the number of b-tagged jets and the scalar sum of masses of large-radius jets. In having explored the full ATLAS 2015-2016 dataset of LHC proton-proton collisions at √s=13 TeV, which corresponds to 36.1 fb -1 of integrated luminosity, no evidence is found for physics beyond the Standard Model. The results are interpreted in the context of simplified models inspired by R-parity-conserving and R-parity-violating supersymmetry, where gluinos are pair-produced. More generic models within the phenomenological minimal supersymmetric Standard Model are also considered.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Integrating GLL-Weibull Distribution Within a Bayesian Framework for Life Prediction of Shape Memory Alloy Spring Undergoing Thermo-mechanical Fatigue

The present paper tackles an important but unmapped problem of the reliability estimations of smart materials. First, an experimental setup is developed for accelerated life testing of the shape memory alloy (SMA) springs. Generalized log-linear Weibull (GLL-Weibull) distribution-based novel approach is then developed for SMA spring life estimation. Applied stimulus (voltage), elongation and cycles of operation are used as inputs for the life prediction model. The values of the parameter coefficients of the model provide better interpretability compared to artificial intelligence based life prediction approaches. In addition, the model also considers the effect of operating conditions, making it generic for a range of the operating conditions. Moreover, a Bayesian framework is used to continuously update the prediction with the actual degradation value of the springs, thereby reducing the uncertainty in the data and improving the prediction accuracy. In addition, the deterioration of material with number of cycles is also investigated using thermogravimetric analysis and scanning electron microscopy.

36 MATERIALS SCIENCE↗

inner-sphere complexation of cations at the rutile-water interface: A concise surface structural interpretation with the CD and MUSIC model

Acid base reactivity and ion-interaction between mineral surfaces and aqueous solutions is most frequently investigated at the macroscopic scale as a function of pH. Experimental data are then rationalized by a variety of surface complexation models. These models are thermodynamically based which in principle does not require a molecular picture. The models are typically calibrated to relatively simple solid-electrolyte solution pairs and may provide poor descriptions of complex multicomponent mineral aqueous solutions, including those found in natural environments. Surface complexation models may be improved by incorporating molecular-scale surface structural information to constrain the modeling efforts. Here, we apply a concise, molecularly-constrained surface complexation model to a diverse suite of surface titration data for rutile and thereby begin to address the complexity of multi-component systems. Primary surface charging curves in NaCl, KCl, and RbCl electrolyte media were fit simultaneously using a charge distribution (CD) and multisite complexation (MUSIC) model [Hiemstra T. and Van Riemsdijk W. H. (1996) A surface structural approach to ion adsorption: the charge distribution (CD) model. J. Colloid Interf. Sci. 179, 488 508], coupled with a Basic Stern layer description of the electric double layer. In addition, data for the specific interaction of Ca2+ and Sr2+ with rutile, in NaCl and RbCl media, were modeled. In recent developments, spectroscopy, quantum calculations, and molecular simulations have shown that electrolyte and divalent cations are principally adsorbed in various inner-sphere configurations on the rutile 110 surface [Zhang Z., Fenter P., Cheng L., Sturchio N. C., Bedzyk M. J., Pr edota M., Bandura A., Kubicki J., Lvov S. N., Cummings P. T., Chialvo A. A., Ridley M. K., Be ne zeth P., Anovitz L., Palmer D. A., Machesky M. L. and Wesolowski D. J. (2004) Ion adsorption at the rutile water interface: linking molecular and macroscopic properties. Langmuir 20, 4954 4969]. Our CD modeling results are consistent with these adsorbed configurations provided adsorbed cation charge is allowed to be distributed between the surface (0-plane) and Stern plane (1-plane). Additionally, a complete description of our titration data required inclusion of outer-sphere binding, principally for Cl which was common to all solutions, but also for Rb+ and K+. These outer-sphere species were treated as point charges positioned at the Stern layer, and hence determined the Stern layer capacitance value. The modeling results demonstrate that a multi-component suite of experimental data can be successfully rationalized within a CD and MUSIC model using a Stern-based description of the EDL. Furthermore, the fitted CD values of the various inner-sphere complexes of the mono- and divalent ions can be linked to the microscopic structure of the surface complexes and other data found by spectroscopy as well as molecular dynamics (MD). For the Na+ ion, the fitted CD value points to the presence of bidenate inner-sphere complexation as suggested by a recent MD study. Moreover, its MD dominance quantitatively agrees with the CD model prediction. For Rb+, the presence of a tetradentate complex, as found by spectroscopy, agreed well with the fitted CD and its predicted presence was quantitatively in very good agreement with the amount found by spectroscopy.

58 GEOSCIENCES↗

In-situ neutron diffraction of a quasicrystal-containing Mg alloy interpreted using a new polycrystal plasticity model of hardening due to {10.2} tensile twinning

Due to the excellent balance of strength and ductility exhibited by some Mg-Zn-RE (Y subgroup rare earth element) alloys, which contain icosahedral quasicrystalline precipitates, it is of interest to examine their deformation mechanisms. Furthermore, the internal strain evolution Mg-3at%Zn-0.5 at%Y with 4 vol% i-phase was measured using in-situ neutron diffraction. The extruded samples exhibit an initially weak <10.0> || extrusion direction “rod texture,” distinct from the normally strong texture of extruded Mg alloys, but the grain size is unexceptional (16.7 ± 2.1 μm). The initially weak texture contributes to a nearly symmetric yielding response between tension and compression. The hardening responses are asymmetric, however, since {10.2} extension twinning is significantly more active during compressive straining, despite the initially weak texture. In-situ neutron diffraction tension and compression experiments parallel to the extrusion direction, together with elasto-plastic self-consistent (EPSC) crystal plasticity modeling, reveal the strength and hardening behavior of individual slip and twinning modes. A model that was previously published about twinning-detwinning (TDT) is implemented within the EPSC framework, and it is proven effective for describing the observed, mild tension-compression asymmetry. This is not possible with previous EPSC-based models of twinning. Finally, the description of hardening within the TDT model is modified, in order to accurately describe the evolution of internal strains within the twins.

36 MATERIALS SCIENCE↗

Divertor heat flux mitigation in the National Spherical Torus Experiment

Steady-state handling of divertor heat flux is a critical issue for both ITER and spherical torus-based devices with compact high power density divertors. Significant reduction of heat flux to the divertor plate has been achieved simultaneously with favorable core and pedestal confinement and stability properties in a highly shaped lower single null configuration in the National Spherical Torus Experiment (NSTX) [M. Ono , Nucl. Fusion 40, 557 2000] using high magnetic flux expansion at the divertor strike point and the radiative divertor technique. A partial detachment of the outer strike point was achieved with divertor deuterium injection leading to peak flux reduction from 4-6 MW m(-2) to 0.5-2 MW m(-2) in small-ELM 0.8-1.0 MA, 4-6 MW neutral beam injection-heated H-mode discharges. A self-consistent picture of the outer strike point partial detachment was evident from divertor heat flux profiles and recombination, particle flux and neutral pressure measurements. Analytic scrape-off layer parallel transport models were used for interpretation of NSTX detachment experiments. The modeling showed that the observed peak heat flux reduction and detachment are possible with high radiated power and momentum loss fractions, achievable with divertor gas injection, and nearly impossible to achieve with main electron density, divertor neutral density or recombination increases alone.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Expecting a child: pregnancy in light of an ontological health model

Aim To make visible the existential character of pregnancy by searching for its health dimensions as it is described in the ontological health model. Method Eight women were interviewed two or three times during their pregnancy and one time shortly after birth. The women have taken part in the routine programme that constitutes maternity care in S weden. A hermeneutic approach inspired by G adamer was used to analyse the data. An ontological health model was used to interpret the findings. Findings Nine themes on a rational and a contextual level were clustered to three themes on an existential level: A new life stage, The new life takes shape and Health is jeopardised. Contents of these themes were interpreted together with an ontological health model. Expecting a child means to do, be and become in expectation of the new life. Suffering and health are two different dimensions in a woman's life during pregnancy; they are integrated with one another and ever‐present. It is in the meeting with the inevitable, life‐changing gravidity and vulnerability that the integrative movement creates the new. Longing is the desire that provides motivation to continue.

Larsson, Åsa↗

Nonradiative recombination in GaN quantum dots formed in the AlN matrix

The mechanisms of temperature quenching of steady-state photoluminescence are studied for structures with hexagonal GaN quantum dots embedded in the AlN matrix. The structures are grown by molecular beam epitaxy. The study is conducted for structures with differently sized quantum dots, for which the peak of the photoluminescence band is in the range from 2.5 to 4.0 eV. It is found that the activation energy of thermal quenching of photoluminescence varies from 27 to 110 meV, as the quantum-dot height is decreased from 5 to 2 nm. A model is suggested to interpret the results. According to the model, the photo-luminescence signal is quenched because of the transfer of charge carriers from energy levels in the quantum dots to defect levels in the matrix.

39↗