SEARCH · Engineering Papers

Results for “Deep Neural Network”

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.

SourceResearch connectionFrom yearThrough year

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 1,603 records · Page 89

All sources NASA NTRS DOE OSTI

DOE OSTI2019

PERSIANN-CNN: Precipitation Estimation from Remotely Sensed Information Using Artificial Neural Networks–Convolutional Neural Networks

Abstract Accurate and timely precipitation estimates are critical for monitoring and forecasting natural disasters such as floods. Despite having high-resolution satellite information, precipitation estimation from remotely sensed data still suffers from methodological limitations. State-of-the-art deep learning algorithms, renowned for their skill in learning accurate patterns within large and complex datasets, appear well suited to the task of precipitation estimation, given the ample amount of high-resolution satellite data. In this study, the effectiveness of applying convolutional neural networks (CNNs) together with the infrared (IR) and water vapor (WV) channels from geostationary satellites for estimating precipitation rate is explored. The proposed model performances are evaluated during summer 2012 and 2013 over central CONUS at the spatial resolution of 0.08° and at an hourly time scale. Precipitation Estimation from Remotely Sensed Information Using Artificial Neural Networks (PERSIANN)–Cloud Classification System (CCS), which is an operational satellite-based product, and PERSIANN–Stacked Denoising Autoencoder (PERSIANN-SDAE) are employed as baseline models. Results demonstrate that the proposed model (PERSIANN-CNN) provides more accurate rainfall estimates compared to the baseline models at various temporal and spatial scales. Specifically, PERSIANN-CNN outperforms PERSIANN-CCS (and PERSIANN-SDAE) by 54% (and 23%) in the critical success index (CSI), demonstrating the detection skills of the model. Furthermore, the root-mean-square error (RMSE) of the rainfall estimates with respect to the National Centers for Environmental Prediction (NCEP) Stage IV gauge–radar data, for PERSIANN-CNN was lower than that of PERSIANN-CCS (PERSIANN-SDAE) by 37% (14%), showing the estimation accuracy of the proposed model.

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2021

A deep learning interatomic potential developed for atomistic simulation of carbon materials

Interatomic potentials based on neural-network machine learning method have attracted considerable attention in recent years owing to their outstanding ability to balance the accuracy and efficiency in atomistic simulations. In this work, a neural-network potential (NNP) for carbon is generated to simulate the structural properties of various carbon structures. The potential is trained using a database consisting of crystalline and liquid structures obtained by the first-principles density functional theory (DFT) calculations. The developed potential accurately predicts the energies and forces in crystalline and liquid carbon structures, the energetic stability of defected graphene, and the structures of amorphous carbon as the function of density. As a result, the excellent accuracy and transferability of the NNP provide a promising tool for accurate atomistic simulations of various carbon materials with faster speed and much lower cost.

36 MATERIALS SCIENCE↗

DOE OSTI2024

Deep learning insights into cosmological structure formation

The evolution of linear initial conditions present in the early Universe into extended halos of dark matter at late times can be computed using cosmological simulations. However, a theoretical understanding of this complex process remains elusive; in particular, the role of anisotropic information in the initial conditions in establishing the final mass of dark matter halos remains a long-standing puzzle. Here, we build a deep learning framework to investigate this question. We train a three-dimensional convolutional neural network to predict the mass of dark matter halos from the initial conditions, and quantify in full generality the amounts of information in the isotropic and anisotropic aspects of the initial density field about final halo masses. We find that anisotropies add a small, albeit statistically significant amount of information over that contained within spherical averages of the density field about final halo mass. However, the overall scatter in the final mass predictions does not change qualitatively with this additional information, only decreasing from 0.9 dex to 0.7 dex. Given such a small improvement, our results demonstrate that isotropic aspects of the initial density field essentially saturate the relevant information about final halo mass. Therefore, instead of searching for information directly encoded in initial conditions anisotropies, a more promising route to accurate, fast halo mass predictions is to add approximate dynamical information based e.g. on perturbation theory. More broadly, our results indicate that deep learning frameworks can provide a powerful tool for extracting physical insight into cosmological structure formation. Published by the American Physical Society 2024

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

DOE OSTI2022

Frame invariant neural network closures for Kraichnan turbulence

Numerical simulations of geophysical and atmospheric flows have to rely on parameterizations of subgrid scale processes due to their limited spatial resolution. Despite substantial progress in developing parameterization (or closure) models for subgrid scale (SGS) processes using physical insights and mathematical approximations, they remain imperfect and can lead to inaccurate predictions. In recent years, machine learning has been successful in extracting complex patterns from high-resolution spatio-temporal data, leading to improved parameterization models, and ultimately better coarse grid prediction. However, the inability to satisfy known physics and poor generalization hinders the application of these models for real-world problems. In this work, we put forth a frame invariant closure approach to improve the accuracy and generalizability of deep learning-based subgrid scale closure models by embedding physical symmetries directly into the structure of the neural network. Specifically, we utilized specialized layers within the convolutional neural network in such a way that desired constraints are theoretically guaranteed without the need for any regularization terms. We demonstrate our framework for a two-dimensional decaying turbulence test case mostly characterized by the forward enstrophy cascade. We show that our frame invariant SGS model (i) accurately predicts the subgrid scale source term, (ii) respects the physical symmetries such as translation, Galilean, and rotation invariance, and (iii) is numerically stable when implemented in coarse-grid simulation with generalization to different initial conditions and Reynolds number. This work opens up a possibility of connecting physics-based theories and data-driven modeling paradigms, and thus represents a promising step towards the development of physically consistent data-driven turbulence closure models.

42 ENGINEERING↗

DOE OSTI2026

Peak2Patch: High-Fidelity Functional Group Identification through Attention-Based Fusion of Infrared and Mass Spectra

Identifying molecular structure based on spectroscopic readings is a key task in a variety of chemical and biological applications. Common spectroscopy techniques, such as Infrared (IR) Spectroscopy and Mass Spectrometry (MS), provide detailed information on the structure of molecular compounds but nonetheless require expert-level knowledge to decode. Machine learning has emerged as a potential solution for automating structure prediction from chemical spectra; however, current approaches generally focus on single sensor modalities, neglecting to leverage the complementary information contained within differing spectra. In this paper, we introduce Peak2Patch, a novel approach to fusion-enhanced prediction of functional groups from IR and mass spectra. First, we perform a detailed comparison of backbone networks for encoding both sparse mass spectra and dense IR spectra and demonstrate the superior performance of transformer neural networks over current state-of-the-art convolutional neural networks. Second, we evaluate three broad categories of fusion: early (raw feature), middle (deep feature), and late (decision) fusion, demonstrating the potential of a deep feature fusion-based approach. Lastly, we present Peak2Patch, our attention-based fusion scheme, which leverages cross-attention to mix features between encoded tokens of the two modalities. We validate our approach on a publicly available multimodal spectroscopic data set of 790k simulated molecules, demonstrating a large improvement in functional group prediction over both the previous state-of-the-art and our own strong single-modal baselines.

Jacobson, Philip [Sandia National Laboratories (SN↗

DOE OSTI2022

Decoding the protein–ligand interactions using parallel graph neural networks

Abstract Protein–ligand interactions (PLIs) are essential for biochemical functionality and their identification is crucial for estimating biophysical properties for rational therapeutic design. Currently, experimental characterization of these properties is the most accurate method, however, this is very time-consuming and labor-intensive. A number of computational methods have been developed in this context but most of the existing PLI prediction heavily depends on 2D protein sequence data. Here, we present a novel parallel graph neural network (GNN) to integrate knowledge representation and reasoning for PLI prediction to perform deep learning guided by expert knowledge and informed by 3D structural data. We develop two distinct GNN architectures: $$\hbox {GNN}_{\mathrm{F}}$$ GNN F is the base implementation that employs distinct featurization to enhance domain-awareness, while $$\hbox {GNN}_{\mathrm{P}}$$ GNN P is a novel implementation that can predict with no prior knowledge of the intermolecular interactions. The comprehensive evaluation demonstrated that GNN can successfully capture the binary interactions between ligand and protein’s 3D structure with 0.979 test accuracy for $$\hbox {GNN}_{\mathrm{F}}$$ GNN F and 0.958 for $$\hbox {GNN}_{\mathrm{P}}$$ GNN P for predicting activity of a protein–ligand complex. These models are further adapted for regression tasks to predict experimental binding affinities and $$\hbox {pIC}_{\mathrm{50}}$$ pIC 50 crucial for compound’s potency and efficacy. We achieve a Pearson correlation coefficient of 0.66 and 0.65 on experimental affinity and 0.50 and 0.51 on $$\hbox {pIC}_{\mathrm{50}}$$ pIC 50 with $$\hbox {GNN}_{\mathrm{F}}$$ GNN F and $$\hbox {GNN}_{\mathrm{P}}$$ GNN P , respectively, outperforming similar 2D sequence based models. Our method can serve as an interpretable and explainable artificial intelligence (AI) tool for predicted activity, potency, and biophysical properties of lead candidates. To this end, we show the utility of $$\hbox {GNN}_{\mathrm{P}}$$ GNN P on SARS-Cov-2 protein targets by screening a large compound library and comparing the prediction with the experimentally measured data.

59 BASIC BIOLOGICAL SCIENCES↗

DOE OSTI2020

DiHydrogen

DiHydrogen is the second version of the Hydrogen fork of the well-known distributed linear algebra library, Elemental. DiHydrogen is a GPU-accelerated distributed multilinear algebra interface with a particular emphasis on the needs of the scalable distributed deep learning training and inference. DiHydrogen is part of the Livermore Big Artificial Neural Network (LBANN) software stack.

Maruyama, Naoya↗

DOE OSTI2024

Identifying Neutrino Final States and Energies in MicroBooNE with New Deep-Learning Based LArTPC Reconstruction Frameworks

MicroBooNE, a Liquid Argon Time Projection Chamber (LArTPC) located in the $\nu_{\mu}$-dominated Booster Neutrino Beam at Fermilab, has been studying $\nu_{e}$ charged-current (CC) interaction rates to shed light on the MiniBooNE low energy excess. The LArTPC technology employed by MicroBooNE provides the capability to image neutrino interactions with mm-scale precision. Computer vision and other machine learning techniques are promising tools for image processing that could boost efficiencies for selecting $\nu_{e}$-CC and other rare signals, reduce cosmic and beam-induced backgrounds, and improve the reconstruction of neutrino energies. The MicroBooNE experiment has been at the forefront of developing and testing such techniques for use in physics analyses. In this poster we overview deep-learning based reconstruction methods. We will showcase the use of a recurrent neural network to estimate neutrino energies and present a new reconstruction framework that uses convolutional neural networks to locate neutrino interaction vertices, tag pixels with track and shower labels, and perform particle identification on reconstructed clusters. We will present studies characterizing the performance of these new tools and demonstrate their effectiveness through their use in an inclusive $\nu_{e}$-CC event selection.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

DOE OSTI2025

Towards AI Based Data Classification for Decision Making During Testing

During the development of high-consequence items, test systems should be capable of differentiating between test failures resulting from narrowly missing requirements versus those indicating potentially catastrophic faults. In many instances, classifying the data corresponds to simply identifying whether measured waveforms have approximately the anticipated shape. Cast in this light, the problem reduces to converting raw data into a form optimal for use with neural network classifiers. This manuscript investigates different means of representing raw data for image classification. Raw data plots and Short Time Fourier Transform (STFT) spectrograms are classified by both custom built, small-scale, Convolution Neural Networks (CNN) and open-source, multi-million parameter, pre-trained deep CNNs. In the case of time varying frequency content, the STFTs provide images with greater detail and can be accurately classified with simpler networks. This requires less memory and runs faster than classifying the raw data using the more sophisticated options—making STFTs optimal for applications with memory constraints. STFTs are not a panacea. In some cases the time-domain signal contains useful information that should not be discarded. Rather than using raw data or STFTs, the images can be constructed from both by using red and green channels of an RGB image to visualize the real and imaginary components of the transform, with the raw data occupying the blue channel.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2024

Identifying Neutrino Final States and Energies in MicroBooNE with New Deep-Learning Based LArTPC Reconstruction Frameworks

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

DOE OSTI2026

Michel Electron Selection with SPINE for DUNE Far Detector Simulation

Michel electrons are a valuable input for particle detector calibration due to their consistent kinetic energy distribution. This report details the evaluation of a Michel electron identification method's application to simulated data from the DUNE (Deep Underground Neutrino Experiment) far detector. This method, which relies on the neural network-based particle classification software SPINE (Scalable Particle Imaging with Neural Embeddings), was developed and calibrated using simulated data for the SBND (Short-Baseline Neutrino Detector) experiment before being applied to simulated DUNE data from a 1x2x6 subset of far detector modules.

Wilson, Dante [Colorado State U.]↗

DOE OSTI2023

Scaling Subspace-Driven Approaches Using Information Fusion

In this work, we seek to exploit the deep structure of multi-modal data to robustly exploit the group subspace distribution of the information using the Convolutional Neural Networks (CNNs) formalism. Upon unfolding the set of subspaces constituting each data modality, and learning their corresponding encoders, an optimized integration of the generated inherent information is carried out to yield a characterization of various classes. Referred to as deep Multimodal Robust Group Subspace Clustering (DRoGSuRe), this approach is compared against the independently developed state-of-the-art approach named Deep Multimodal Subspace Clustering (DMSC). Experiments on different multimodal datasets show that our approach is competitive and more robust in the presence of noise.

Ghanem, Sally↗

DOE OSTI2020

Analog architectures for neural network acceleration based on non-volatile memory

Analog hardware accelerators, which perform computation within a dense memory array, have the potential to overcome the major bottlenecks faced by digital hardware for data-heavy workloads such as deep learning. Exploiting the intrinsic computational advantages of memory arrays, however, has proven to be challenging principally due to the overhead imposed by the peripheral circuitry and due to the non-ideal properties of memory devices that play the role of the synapse. We review the existing implementations of these accelerators for deep supervised learning, organizing our discussion around the different levels of the accelerator design hierarchy, with an emphasis on circuits and architecture. We explore and consolidate the various approaches that have been proposed to address the critical challenges faced by analog accelerators, for both neural network inference and training, and highlight the key design trade-offs underlying these techniques.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2021

Explicit physics-informed neural networks for nonlinear closure: The case of transport in tissues

In upscaling methods, closures for nonlinear problems present a well-known challenge. While a number of theoretical methods have been proposed for handling such closures, nonlinearities still remain a significant obstacle for many problems. In this work, we use a combination of formal upscaling and data-driven machine learning for explicitly closing a nonlinear transport and reaction process in multiscale tissues. The classical effectiveness factor model is used to formulate the macroscale reaction kinetics. We train a multilayer perceptron network using training data generated by direct numerical simulations over microscale examples. Once trained, the network is used in an algorithm for numerically solving the upscaled (coarse-grained) differential equation describing mass transport and reaction in two example tissues. The network is described as being explicit in the sense that the network is trained using macroscale concentrations and gradients of concentration as components of the feature space rather than incorporating them as part of a constraint in the optimization process. Network training and solutions to the macroscale transport equations were computed for two different tissues. The two tissue types (brain and liver) exhibit markedly different geometrical complexity and spatial scale (cell size and sample size). The upscaled solutions for the average concentration are compared with numerical solutions derived from the microscale concentration fields by a posteriori averaging. There are three outcomes of this work of particular note. 1) Our overall approach results in an upscaled nonlinear PDE. The PDE is closed using a neural network, and our approach results in the definition of the classical effectiveness factor for effecting closure. 2) We identify particular source terms for the closure problem that are important for representing the structure of the closure. These source terms involve macroscale concentrations and their gradients. We adopt these source terms to use as explicit features in the learning algorithm. We find the trained networks that include the macroscale source terms generate models that are able to predict the correction factor with increased fidelity over those that do not. 3) We find that the trained network exhibits good generalizability, and it is able to predict the effectiveness factor with high fidelity for realistically-structured tissues despite the significantly different scale and geometrical complexity of the two example tissue types. This latter result emphasizes our purposeful connection between conventional averaging methods with the use of machine learning for closure; this contrasts with some machine learning methods for upscaling where the exact form of the macroscale equation remains unknown.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2022

Deep Hierarchical Super Resolution for Scientific Data

We present a novel technique for hierarchical super resolution (SR) with neural networks (NNs), which upscales volumetric data represented with an octree data structure to a high-resolution uniform gridwith minimal seam artifacts on octree node boundaries. Our method uses existing state-of-the-art SR models and adds flexibility to upscale input data with varying levels of detail across the domain, instead of only uniform grid data that are supported in previous approaches.The key is to use a hierarchy of SR NNs, each trained to perform 2x SR between two levels of detail, with a hierarchical SR algorithm that minimizes seam artifacts by starting from the coarsest level of detail and working up.We show that our hierarchical approach outperforms baseline interpolation and hierarchical upscaling methods, and demonstrate the usefulness of our proposed approach across three use cases including data reduction using hierarchical downsampling+SR instead of uniform downsampling+SR, computation savings for hierarchical finite-time Lyapunov exponent field calculation, and super-resolving low-resolution simulation results for a high-resolution approximation visualization.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2021

Computationally Efficient Multiscale Neural Networks Applied to Fluid Flow in Complex 3D Porous Media

Abstract The permeability of complex porous materials is of interest to many engineering disciplines. This quantity can be obtained via direct flow simulation, which provides the most accurate results, but is very computationally expensive. In particular, the simulation convergence time scales poorly as the simulation domains become less porous or more heterogeneous. Semi-analytical models that rely on averaged structural properties (i.e., porosity and tortuosity) have been proposed, but these features only partly summarize the domain, resulting in limited applicability. On the other hand, data-driven machine learning approaches have shown great promise for building more general models by virtue of accounting for the spatial arrangement of the domains’ solid boundaries. However, prior approaches building on the convolutional neural network (ConvNet) literature concerning 2D image recognition problems do not scale well to the large 3D domains required to obtain a representative elementary volume (REV). As such, most prior work focused on homogeneous samples, where a small REV entails that the global nature of fluid flow could be mostly neglected, and accordingly, the memory bottleneck of addressing 3D domains with ConvNets was side-stepped. Therefore, important geometries such as fractures and vuggy domains could not be modeled properly. In this work, we address this limitation with a general multiscale deep learning model that is able to learn from porous media simulation data. By using a coupled set of neural networks that view the domain on different scales, we enable the evaluation of large ( $$>512^3$$ > 512 3 ) images in approximately one second on a single graphics processing unit. This model architecture opens up the possibility of modeling domain sizes that would not be feasible using traditional direct simulation tools on a desktop computer. We validate our method with a laminar fluid flow case using vuggy samples and fractures. As a result of viewing the entire domain at once, our model is able to perform accurate prediction on domains exhibiting a large degree of heterogeneity. We expect the methodology to be applicable to many other transport problems where complex geometries play a central role.

36 MATERIALS SCIENCE↗

DOE OSTI2021

Probabilistic Deep Learning Approach to Automate the Interpretation of Multi-phase Diffraction Spectra

Autonomous synthesis and characterization of inorganic materials requires the automatic and accurate analysis of X-ray diffraction spectra. For this task, we designed a probabilistic deep learning algorithm to identify complex multi-phase mixtures. At the core of this algorithm lies an ensemble convolutional neural network trained on simulated diffraction spectra, which are systematically augmented with physics-informed perturbations to account for artifacts that can arise during experimental sample preparation and synthesis. In this work, larger perturbations associated with off-stoichiometry are also captured by supplementing the training set with hypothetical solid solutions. Spectra containing mixtures of materials are analyzed with a newly developed branching algorithm that utilizes the probabilistic nature of the neural network to explore suspected mixtures and identify the set of phases that maximize confidence in the prediction. Our model is benchmarked on simulated and experimentally measured diffraction spectra, showing exceptional performance with accuracies exceeding those given by previously reported methods based on profile matching and deep learning. We envision that the algorithm presented here may be integrated in experimental workflows to facilitate the high-throughput and autonomous discovery of inorganic materials.

36 MATERIALS SCIENCE↗

DOE OSTI2026

Deep learning for time series forecasting: a survey of recent advances

Time series forecasting plays a critical role in numerous real-world applications, such as finance, healthcare, transportation, and scientific computing. In recent years, deep learning has become a powerful tool for modeling complex temporal patterns and improving forecasting accuracy. This survey provides an overview of recent deep learning approaches for time series forecasting, involving various architectures including RNNs, CNNs, GNNs, transformers, large language models, MLP-based models, and diffusion models. We first identify key challenges in the field, such as temporal dependency, efficiency, and cross-variable dependency, which drive the development of forecasting techniques. Then, the general advantages and limitations of each architecture are discussed to contextualize their adaptation in time series forecasting. Furthermore, we highlight promising design trends like multi-scale modeling, decomposition, and frequency-domain techniques, which are shaping the future of the field. This paper serves as a compact reference for researchers and practitioners seeking to understand the current landscape and future trajectory of deep learning in time series forecasting.

97 MATHEMATICS AND COMPUTING↗