SEARCH · Search NASA

Results for “Network inference”

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 811 records · Page 45

All sources NASA NTRS DOE OSTI

DOE OSTI2025

A hybrid neural architecture: Online attosecond x-ray characterization

The emergence of high-repetition-rate x-ray free-electron lasers (XFELs), such as SLAC’s LCLS-II, serves as our canonical example for autonomous controls that necessitate high-throughput diagnostics paired with streaming computational pipelines capable of single-shot analysis with extremely low latency. We present the deterministic characterization with an integrated parallelizable hybrid resolver architecture, a hybrid machine learning framework designed for fast, accurate analysis of XFEL diagnostics using angular streaking-based sinogram images. This architecture integrates convolutional neural networks and bidirectional long short-term memory models to denoise input, identify x-ray sub-spike features, and extract sub-spike relative delays with sub-30 attosecond temporal resolution. Deployed on low-latency hardware, it achieves over 10 kHz throughput with 168.3 μs inference latency, indicating scalability to 14 kHz with field-programmable gate array integration. By transforming regression tasks into classification problems and leveraging optimized error encoding, we achieve high precision with low-latency performance that is critical for real-time streaming event selection and experimental control feedback signals. This represents a key development in real-time control pipelines for next-generation autonomous science, generally, and high repetition-rate x-ray experiments in particular.

Accelerator Physics (physics.acc-ph)↗

DOE OSTI2020

3D-ReG: A 3D ReRAM-based Heterogeneous Architecture for Training Deep Neural Networks

Deep neural network (DNN) models are being expanded to a broader range of applications. The computational capability of traditional hardware platforms cannot accommodate the growth of model complexity. Among recent technologies to accelerate DNN, resistive memory (ReRAM)-based processing-in-memory (PIM) emerged as a promising solution for DNN inference due to its high efficiency for matrix-based computation. We face two major technical challenges in extending the use of ReRAM-based accelerators for training: (1) full-precision data is essential in back-propagation; (2) the need to support both feed-forward and back-propagation aggravates the data-movement burden. We propose a heterogeneous architecture named as 3D-ReG, which leverages full-precision GPU to ensure training accuracy and low-overhead 3D integration to provide low-cost data movements. Moreover, we introduce conservative and aggressive task-mapping schemes, which partition the computation phases in different ways to balance execution efficiency and training accuracy. We evaluate 3D-ReG implemented with two 3D integration technologies, through-silicon vias (TSVs) and monolithic inter-tier vias (MIVs), and compare them with GPU-only and PIM-only counterparts. Various GPU-only platforms using two main-memory technologies (DRAM, ReRAM) and three interconnect technologies (2D, TSV, MIV) are evaluated as well. Experimental results show that 3D-ReG can achieve on average 5.64× training speedup and 3.56× higher energy efficiency compared with the GPU with DRAM as main memory, at the cost of 0.05%–3.39% accuracy drop. We define a new metric, gain-loss ratio (GLR), which quantitatively evaluates the capability of a DNN training hardware in terms of the model accuracy and hardware efficiency. The results of our comparison show that the aggressive task-mapping scheme on MIV-based 3D-ReG outperforms the other methods.

Computer Science↗

NASA NTRS2015

Using Single-Scattering Albedo Spectral Curvature to Characterize East Asian Aerosol Mixtures

Spectral dependence of aerosol single-scattering albedo (SSA) has been used to infer aerosol composition. In particular, aerosol mixtures dominated by dust absorption will have monotonically increasing SSA with wavelength while that dominated by black carbon absorption has monotonically decreasing SSA spectra. However, by analyzing SSA measured at four wavelengths, 440, 675, 870, and 1020 nm from the Aerosol Robotic Network data set, we find that the SSA spectra over East Asia are frequently peaked at 675 nm. In these cases, we suggest that SSA spectral curvature, defined as the negative of the second derivative of SSA as a function of wavelength, can provide additional information on the composition of these aerosol mixtures. Aerosol SSA spectral curvatures for East Asia during fall and winter are considerably larger than those found in places primarily dominated by biomass burning or dust aerosols. SSA curvature is found to increase as the SSA magnitude decreases. The curvature increases with coarse mode fraction (CMF) to a CMF value of about 0.4, then slightly decreases or remains constant at larger CMF. Mie calculations further verify that the strongest SSA curvature occurs at approx. 40% dust fraction, with 10% scattering aerosol fraction. The nonmonotonic SSA spectral dependence is likely associated with enhanced absorption in the shortwave by dust, absorption by black carbon at longer wavelengths, and also the flattened absorption optical depth spectral dependence due to the increased particle size.

curvature↗

DOE OSTI2024

GrainGNN: A dynamic graph neural network for predicting 3D grain microstructure

We propose GrainGNN, a surrogate model for the evolution of polycrystalline grain structure under rapid solidification conditions in metal additive manufacturing. High fidelity simulations of solidification microstructures are typically performed using multicomponent partial differential equations (PDEs) with moving interfaces. The inherent randomness of the PDE initial conditions (grain seeds) necessitates ensemble simulations to predict microstructure statistics, e.g., grain size, aspect ratio, and crystallographic orientation. Here, currently such ensemble simulations are prohibitively expensive and surrogates are necessary.In GrainGNN, we use a dynamic graph to represent interface motion and topological changes due to grain coarsening. We use a reduced representation of the microstructure using hand-crafted features; we combine pattern finding and altering graph algorithms with two neural networks, a classifier (for topological changes) and a regressor (for interface motion). Both networks have an encoder-decoder architecture; the encoder has a multi-layer transformer long-short-term-memory architecture; the decoder is a single layer perceptron.We evaluate GrainGNN by comparing it to high-fidelity phase field simulations for in-distribution and out-of-distribution grain configurations for solidification under laser power bed fusion conditions. GrainGNN results in 80%–90% pointwise accuracy; and nearly identical distributions of scalar quantities of interest (QoI) between phase field and GrainGNN simulations compared using Kolmogorov-Smirnov test. GrainGNN's inference speedup (PyTorch on single x86 CPU) over a high-fidelity phase field simulation (CUDA on a single NVIDIA A100 GPU) is 150×–2000× for 100-initial grain problem. Further, using GrainGNN, we model the formation of 11,600 grains in 220 seconds on a single CPU core.

36 MATERIALS SCIENCE↗

DOE OSTI2025

Real Time implementation of Artificial Intelligence compression algorithm for High-Speed Streaming Readout signals

The new generation of high-energy physics experiments plans to acquire data in streaming mode. With this approach, it is possible to access the information of the whole detector (organized in time slices) for optimal and lossless triggering of data acquisitions. With this approach, data rates, especially in large detectors, are often very high, and the network is likely to be the bottleneck for the entire Streaming Read Out system. The aim of this work is to study the implementation of a lossy compression algorithm based on Artificial Intelligence: an Autoencoder. With Machine Learning it is possible to achieve a high compression ratio and fast inference time with only a small degradation of the signals, almost negligible for the specific application. This work explores different configurations of the Autoencoder and the implementation on different hardware. Different Autoencoder configurations are explored to find the best trade-off between compression ratio and reconstruction loss, both for signals and energy spectrum. Different hardware implementations are also explored to find the best platform to achieve real-time performance for the specific application.

Rossi, Fabio (ORCID:0009000385713885)↗

DOE OSTI2025

Comprehensive Neural Posterior Estimation for Galaxy-Galaxy Strong Lensing

We present a deep learning model based on neural posterior estimation (NPE) for comprehensive extraction of astrophysical parameters from galaxy-scale strong gravitational lenses. The unprecedentedly large amount of galaxy-scale strong lenses expected in future cosmological surveys (${\cal O}(10^5)$) promises to enable valuable statistical constraints in various studies ranging from galaxy formation to the nature of dark matter, but it also poses a significant challenge for traditional modelling pipelines. To this end, our automated model includes several new, state-of-the-art features and approaches leveraging the framework of simulation-based inference (SBI). We infer a total of 20 parameters describing the mass and light profiles of both lens and source galaxies, using simulated raw multi-band data modelled under noise and observing conditions expected by the Legacy Survey of Space and Time (LSST), with its summary statistics generated by a residual network. We examine the efficacy of multi-band data in extracting nearly 20 model parameters simultaneous from strong lensing images including lens light. Finally, We perform a comprehensive set of diagnostics for SBI models, evaluating the model's prediction accuracy, stability, and uncertainty quantification.

Zhao, Roy J. [Chicago U., KICP]↗

DOE OSTI2022

Seismic Waveform Inversion Capability on Resource-Constrained Edge Devices

Seismic full wave inversion (FWI) is a widely used non-linear seismic imaging method used to reconstruct subsurface velocity images, however it is time consuming, has high computational cost and depend heavily on human interaction. Recently, deep learning has accelerated it’s use in several data-driven techniques, however most deep learning techniques suffer from overfitting and stability issues. In this work, we propose an edge computing-based data-driven inversion technique based on supervised deep convolutional neural network to accurately reconstruct the subsurface velocities. Deep learning based data-driven technique depends mostly on bulk data training. In this work, we train our deep convolutional neural network (DCN) (UNet and InversionNet) on the raw seismic data and their corresponding velocity models during the training phase to learn the non-linear mapping between the seismic data and velocity models. The trained network is then used to estimate the velocity models from new input seismic data during the prediction phase. The prediction phase is performed on a resource-constrained edge device such as Raspberry Pi. Raspberry Pi provides real-time and on-device computational power to execute the inference process. In addition, we demonstrate robustness of our models to perform inversion in the presence on noise by performing both noise-aware and no-noise training and feeding the resulting trained models with noise at different signal-to-noise (SNR) ratio values. We make great efforts to achieve very feasible inference times on the Raspberry Pi for both models. Specifically, the inference times per prediction for UNet and InversionNet models on Raspberry Pi were 22 and 4 s respectively whilst inference times for both models on the GPU were 2 and 18 s which are very comparable. Finally, we have designed a user-friendly interactive graphical user interface (GUI) to automate the model execution and inversion process on the Raspberry Pi.

Manu, Daniel (ORCID:0000000154982677)↗

NASA NTRS2012

The Kinematic and Microphysical Control of Storm Integrated Lightning Flash Extent

Objective: To investigate the kinematic and microphysical control of lightning properties, particularly those that may govern the production of nitrogen oxides (NOx) in thunderstorms, such as flash rate, type (intracloud [IC] vs. cloud-to-ground [CG] ) and extent. Data and Methodology: a) NASA MSFC Lightning Nitrogen Oxides Model (LNOM) is applied to North Alabama Lightning Mapping Array (NALMA) and Vaisala National Lightning Detection Network(TradeMark) (NLDN) observations following ordinary convective cells through their lifecycle. b) LNOM provides estimates of flash type, channel length distributions, lightning segment altitude distributions (SADs) and lightning NOx production profiles (Koshak et al. 2012). c) LNOM lightning characteristics are compared to the evolution of updraft and precipitation properties inferred from dual-Doppler (DD) and polarimetric radar analyses of UAHuntsville Advanced Radar for Meteorological and Operational Research (ARMOR, Cband, polarimetric) and KHTX (S-band, Doppler).

Carey, Lawrence D.↗

NASA NTRS2015

Spatiotemporal Coherence of Mean and Extreme Aerosol Particle Events over Eastern North America as Observed from Satellite

Quantifying spatiotemporal scales of coherence and variability of aerosol particle properties provides (i) metrics for evaluating observational data sets and model simulations, (ii) inferences regarding dominant causes of variability, and (iii) information that is necessary to determine their climate forcing and air quality impacts. Accordingly, analyses of columnar aerosol optical depth (AOD) and Ångstrom exponent (AE) from the MODerate resolution Imaging Spectroradiometer (MODIS) on the Aqua and Terra satellites, and AErosol RObotic NETwork (AERONET) stations are used to evaluate the spatially averaged (Level-3) MODIS products, and to characterize the spatiotemporal scales of variability (and coherence) of aerosol particles over eastern North America (2000-2013). Results show a high degree of consistency in AOD retrievals from the different data sets but larger discrepancies in AE estimates, highest mean AOD and lowest day-to-day variability during the summer, and largest scales of spatial coherence in summer and fall. Consistent with the scales of spatial coherence from MODIS data, in both data sets power spectra indicate AOD variability is manifest primarily on synoptic and annual time scales. Conversely, AE variability in MODIS data is primarily focused on seasonal, semiannual, and annual time scales, and there is an additional mode of AE variability at approx. 30 days in the AERONET measurements. The frequency of cooccurrence of extreme AOD values (>local 90th percentile) decreases to below 50% at approx. 150 km from a central grid cell, but is above that expected by random chance over almost all of eastern North America, indicating supra-regional scale extreme events.

MODIS↗

DOE OSTI2022

Tracking the location of a road-constrained radioactive source with a network of detectors

Not Available

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

DOE OSTI2024

Bridging the gap between experiments and simulations using machine learning

The physics of inertial confinement fusion is rich and complex. Simulation codes that are used to design experiments are computationally expensive and lack the predictive capability required for extensive parameter exploration in search of a high-performing design for laser direct drive. In this work we use deep learning to build a fast emulator of experiments. To facilitate the development of the deep-learning model, an autoencoder is used to reduce the dimensionality of the input space. Two deep learning models are developed. One model is trained on a vast array of simulation data and is subsequently calibrated to expensive and limited experimental data using a technique known as “transfer learning.” The other model is trained on a statistical model and is subsequently calibrated using experimental data. A comparative study of the two predictive models is carried out. The models potentially reproduce key experimental observables with high accuracy and unprecedented inference times relative to those achieved with simulation codes. These models facilitate rapid exploration of a high dimensional input parameter space.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

NASA NTRS1989

REDEX: The ranging equipment diagnostic expert system

REDEX, an advanced prototype expert system that diagnoses hardware failures in the Ranging Equipment (RE) at NASA's Ground Network tracking stations is described. REDEX will help the RE technician identify faulty circuit cards or modules that must be replaced, and thereby reduce troubleshooting time. It features a highly graphical user interface that uses color block diagrams and layout diagrams to illustrate the location of a fault. A semantic network knowledge representation technique was used to model the design structure of the RE. A catalog of generic troubleshooting rules was compiled to represent heuristics that are applied in diagnosing electronic equipment. Specific troubleshooting rules were identified to represent additional diagnostic knowledge that is unique to the RE. Over 50 generic and 250 specific troubleshooting rules have been derived. REDEX is implemented in Prolog on an IBM PC AT-compatible workstation. Block diagram graphics displays are color-coded to identify signals that have been monitored or inferred to have nominal values, signals that are out of tolerance, and circuit cards and functions that are diagnosed as faulty. A hypertext-like scheme is used to allow the user to easily navigate through the space of diagrams and tables. Over 50 graphic and tabular displays have been implemented. REDEX is currently being evaluated in a stand-alone mode using simulated RE fault scenarios. It will soon be interfaced to the RE and tested in an online environment. When completed and fielded, REDEX will be a concrete example of the application of expert systems technology to the problem of improving performance and reducing the lifecycle costs of operating NASA's communications networks in the 1990's.

Luczak, Edward C.↗

NASA NTRS1989

REDEX - The ranging equipment diagnostic expert system

Luczak, Edward C.↗

DOE OSTI2022

Inferring topological transitions in pattern-forming processes with self-supervised learning

Abstract The identification of transitions in pattern-forming processes are critical to understand and fabricate microstructurally precise materials in many application domains. While supervised methods can be useful to identify transition regimes, they need labels, which require prior knowledge of order parameters or relevant microstructures describing these transitions. Instead, we develop a self-supervised, neural-network-based approach that does not require predefined labels about microstructure classes to predict process parameters from observed microstructures. We show that assessing the difficulty of solving this inverse problem can be used to uncover microstructural transitions. We demonstrate our approach by automatically discovering microstructural transitions in two distinct pattern-forming processes: the spinodal decomposition of a two-phase mixture and the formation of binary-alloy microstructures during physical vapor deposition of thin films. This approach opens a path forward for discovering unseen or hard-to-discern transitions and ultimately controlling complex pattern-forming processes.

Abram, Marcin↗

DOE OSTI2024

Uncertainty-quantification-enabled inversion of nuclear responses

Nuclear quantum many-body methods rely on integral transform techniques to infer properties of electroweak response functions from ground-state expectation values. Retrieving the energy dependence of these responses is highly nontrivial, especially for quantum Monte Carlo methods, as it requires inverting the Laplace transform, a notoriously ill-posed problem. Here, in this work, we propose an artificial neural network architecture suitable for accurate response function reconstruction with precise estimation of the uncertainty of the inversion. We demonstrate the capabilities of this new architecture benchmarking it against maximum entropy and previously developed neural network methods designed for a similar task, paying particular attention to its robustness noise in the Euclidean

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

DOE OSTI2021

Proteome specialization of anaerobic fungi during ruminal degradation of recalcitrant plant fiber

The rumen harbors a complex microbial mixture of archaea, bacteria, protozoa, and fungi that efficiently breakdown plant biomass and its complex dietary carbohydrates into soluble sugars that can be fermented and subsequently converted into metabolites and nutrients utilized by the host animal. While rumen bacterial populations have been well documented, only a fraction of the rumen eukarya are taxonomically and functionally characterized, despite the recognition that they contribute to the cellulolytic phenotype of the rumen microbiota. To investigate how anaerobic fungi actively engage in digestion of recalcitrant fiber that is resistant to degradation, we resolved genome-centric metaproteome and metatranscriptome datasets generated from switchgrass samples incubated for 48 h in nylon bags within the rumen of cannulated dairy cows. Across a gene catalog covering anaerobic rumen bacteria, fungi and viruses, a significant portion of the detected proteins originated from fungal populations. Intriguingly, the carbohydrate-active enzyme (CAZyme) profile suggested a domain-specific functional specialization, with bacterial populations primarily engaged in the degradation of hemicelluloses, whereas fungi were inferred to target recalcitrant cellulose structures via the detection of a number of endo- and exo-acting enzymes belonging to the glycoside hydrolase (GH) family 5, 6, 8, and 48. Notably, members of the GH48 family were amongst the highest abundant CAZymes and detected representatives from this family also included dockerin domains that are associated with fungal cellulosomes. A eukaryote-selected metatranscriptome further reinforced the contribution of uncultured fungi in the ruminal degradation of recalcitrant fibers. These findings elucidate the intricate networks of in situ recalcitrant fiber deconstruction, and importantly, suggest that the anaerobic rumen fungi contribute a specific set of CAZymes that complement the enzyme repertoire provided by the specialized plant cell wall degrading rumen bacteria.

Hagen, Live↗

DOE OSTI2022

Toward FAIR Representations of Microbial Interactions

Despite an ever-growing number of data sets that catalog and characterize interactions between microbes in different environments and conditions, many of these data are neither easily accessible nor intercompatible. These limitations present a major challenge to microbiome research by hindering the streamlined drawing of inferences across studies.

59 BASIC BIOLOGICAL SCIENCES↗

NASA NTRS2012

Model-Based Method for Sensor Validation

Fault detection, diagnosis, and prognosis are essential tasks in the operation of autonomous spacecraft, instruments, and in situ platforms. One of NASA s key mission requirements is robust state estimation. Sensing, using a wide range of sensors and sensor fusion approaches, plays a central role in robust state estimation, and there is a need to diagnose sensor failure as well as component failure. Sensor validation can be considered to be part of the larger effort of improving reliability and safety. The standard methods for solving the sensor validation problem are based on probabilistic analysis of the system, from which the method based on Bayesian networks is most popular. Therefore, these methods can only predict the most probable faulty sensors, which are subject to the initial probabilities defined for the failures. The method developed in this work is based on a model-based approach and provides the faulty sensors (if any), which can be logically inferred from the model of the system and the sensor readings (observations). The method is also more suitable for the systems when it is hard, or even impossible, to find the probability functions of the system. The method starts by a new mathematical description of the problem and develops a very efficient and systematic algorithm for its solution. The method builds on the concepts of analytical redundant relations (ARRs).

Vatan, Farrokh↗