Search NASA⌕ Search

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.

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

At least 991 records · Page 55

Observable optimization for precision theory: machine learning energy correlators

The practice of collider physics typically involves the marginalization of multi-dimensional collider data to uni-dimensional observables relevant for some physics task. In many cases, such as classification or anomaly detection, the observable can be arbitrarily complicated, such as the output of a neural network. However, for precision measurements, the observable must correspond to something computable systematically beyond the level of current simulation tools. In this work, we demonstrate that precision-theory-compatible observable space exploration can be systematized by using neural simulation-based inference techniques from machine learning. We illustrate this approach by exploring the space of marginalizations of the energy 3-point correlator to optimize sensitivity to the top quark mass. We first learn the energy-weighted probability density from simulation, then search in the space of marginalizations for an optimal triangle shape. Although simulations and machine learning are used in the process of observable optimization, the output is an observable definition which can be then computed to high precision and compared directly to data without any memory of the computations which produced it. We find that the optimal marginalization is isosceles triangles on the sphere with a side ratio approximately $1 : 1 : \sqrt{2}$ (i.e. right triangles) within the set of marginalizations we consider.

Jets and Jet Substructure↗

Analyzing Data Privacy for Edge Systems

Internet-of-Things (IoT)-based streaming applications are all around us. Currently, we are transitioning from IoT processing being performed on the cloud to the edge. While moving to the edge provides significant networking efficiency benefits, IoT edge computing creates significant data privacy concerns. We propose a methodology that can successfully privacy protect the continual data streams generated by sensors on the edge device. We implement local differential privacy on streaming data and incorporate Bayesian inference and Gaussian process to evaluate the privacy policy. We demonstrate our methodology on a real-world smart meter testbed and identify the optimal privacy protection settings.

Kotevska, Olivera↗

SPANet: Generalized permutationless set assignment for particle physics using symmetry preserving attention

The creation of unstable heavy particles at the Large Hadron Collider is the most direct way to address some of the deepest open questions in physics. Collisions typically produce variable-size sets of observed particles which have inherent ambiguities complicating the assignment of observed particles to the decay products of the heavy particles. Current strategies for tackling these challenges in the physics community ignore the physical symmetries of the decay products and consider all possible assignment permutations and do not scale to complex configurations. Attention based deep learning methods for sequence modelling have achieved state-of-the-art performance in natural language processing, but they lack built-in mechanisms to deal with the unique symmetries found in physical set-assignment problems. We introduce a novel method for constructing symmetry-preserving attention networks which reflect the problem's natural invariances to efficiently find assignments without evaluating all permutations. This general approach is applicable to arbitrarily complex configurations and significantly outperforms current methods, improving reconstruction efficiency between 19% - 35% on typical benchmark problems while decreasing inference time by two to five orders of magnitude on the most complex events, making many important and previously intractable cases tractable. A full code repository containing a general library, the specific configuration used, and a complete dataset release, are available at https://github.com/Alexanders101/SPANet

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Simple and Sophisticated Models of Taylor's Cylinder Impact

We present a study based on the simplest possible models for Taylor’s cylinder impact problem, in addition to examination of using convolutional neural networks (CNNs) to map cylinder profiles to strength calibrations. We find that the approximate treatments of Taylor and Hawkyard compare well with hydrodynamic simulations using an equivalent assumption of constant flow stress. The CNN models prove to be well suited to successfully infer parameterizations of the Preston-Tonks-Wallace model of plastic deformation based on the deformed profile of an impacted cylinder

36 MATERIALS SCIENCE↗

SafeDNN: Understanding and Verifying Neural Networks

The SafeDNN project at NASA Ames explores analysis techniques and tools to ensure that systems that use Deep Neural Networks (DNN) are safe, robust and interpretable. Research directions we are pursuing include: symbolic execution for DNN analysis, label-guided clustering to automatically identify input regions that are robust, parallel and compositional approaches to improve formal SMT-based verification, property inference and automated program repair for DNNs, adversarial training and detection, probabilistic reasoning for DNNs. In this talk I will highlight some of the research advances from SafeDNN, that were already published.

Corina Pasareanu↗

Business Traveler Behavioral Shifts from Ride-Hailing: A Before-After Case Study

In this case study, the cost and behavioral impacts resulting from the adoption of transportation network companies (TNCs) over traditional rental vehicle use are analyzed for one individual's typical business travel. The travel consists of business-related trips to Columbus, Ohio, over the course of several months from late 2016 through mid 2017. The case study analyzes twenty such trips, ten with rental cars serving the primary transportation and ten with TNCs as the primary mode of transporation. The analysis quantifies the differences in mobility, cost, convenience, and sustainability for a single subject. Quantified metrics include vehicle miles traveled, cost of transportation, and behavioral changes such as hotel, service preferences, and alterred work routines. Opinions and attitudes towards the use of TNCs in lieue of rental cars are also documented. Alhtough a single-subject study is not enough data to make inferences over a larger population, this case study provides a basis for a larger study, providing insight into a possible measureable impacts of TNC adoption at scale.

33 ADVANCED PROPULSION SYSTEMS↗

Cyber-Physical Trade-Offs in Distributed Detection Networks

We consider a network of sensors that measure the scalar intensity due to the background or a source combined with background, inside a two-dimensional monitoring area. The sensor measurements may be random due to the underlying nature of the source and background or due to sensor errors or both. The detection problem is infer the presence of a source of unknown intensity and location based on sensor measurements. In the conventional approach, detection decisions are made at the individual sensors, which are then combined at the fusion center, for example using the majority rule. With increased communication and computation costs, we show that a more complex fusion algorithm based on measurements achieves better detection performance under smooth and non-smooth source intensity functions, Lipschitz conditions on probability ratios and a minimum packing number for the state-space. We show that these conditions for trade-offs between the cyber costs and physical detection performance are applicable for two detection problems: (i) point radiation sources amidst background radiation, and (ii) sources and background with Gaussian distributions.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

U.S. NO₂ trends (2005–2013): EPA air quality system (AQS) data versus improved observations from the Ozone Monitoring Instrument (OMI)

Emissions of nitrogen oxides (NO x ) and, subsequently, atmospheric levels of nitrogen dioxide (NO₂) have decreased over the U.S. due to a combination of environmental policies and technological change. Consequently, NO₂ levels have decreased by 30–40% in the last decade. We quantify NO₂ trends (2005–2013) over the U.S. using surface measurements from the U.S. Environmental Protection Agency (EPA) Air Quality System (AQS) and an improved tropospheric NO₂ vertical column density (VCD) data product from the Ozone Monitoring Instrument (OMI) on the Aura satellite.We demonstrate that the current OMI NO₂ algorithm is of sufficient maturity to allow a favorable correspondence of trends and variations in OMI and AQS data. Our trend model accounts for the non-linear dependence of NO₂ concentration on emissions associated with the seasonal variation of the chemical lifetime, including the change in the amplitude of the seasonal cycle associated with the significant change in NO x emissions that occurred over the last decade. The direct relationship between observations and emissions becomes more robust when one accounts for these non-linear dependencies. We improve the OMI NO₂ standard retrieval algorithm and, subsequently, the data product by using monthly vertical concentration profiles, a required algorithm input, from a high-resolution chemistry and transport model (CTM) simulation with varying emissions (2005-2013). The impact of neglecting the time-dependence of the profiles leads to errors in trend estimation, particularly in regions where emissions have changed substantially. For example, trends calculated from retrievals based on time-dependent profiles offer 18% more instances of significant trends and up to 15% larger total NO₂ reduction versus the results based on profiles for 2005. Using a CTM, we explore the theoretical relation of the trends estimated from NO₂ VCDs to those estimated from ground-level concentrations. The model-simulated trends in VCDs strongly correlate with those estimated from surface concentrations (r = 0.83, N = 355). We then explore the observed correspondence of trends estimated from OMI and AQS data. We find a significant, but slightly weaker, correspondence (i.e., r = 0.68, N = 208) than predicted by the model and discuss some of the important factors affecting the relationship, including known problems (e.g., NO z interferents) associated with the AQS data. This significant correspondence gives confidence in trend and surface concentration estimates from OMI VCDs for locations, such as the majority of the U.S. and globe, that are not covered by surface monitoring networks. Using our improved trend model and our enhanced OMI data product, we find that both OMI and AQS data show substantial downward trends from 2005 to 2013, with an average reduction of 38% for each over the U.S. The annual reduction rates inferred from OMI and AQS measurements are larger (–4.8 ± 1.9%/yr, –3.7 ± 1.5%/yr) from 2005 to 2008 than 2010 to 2013 (–1.2 ± 1.2%/yr, –2.1 ± 1.4%/yr). We quantify NO₂ trends for major U.S. cities and power plants; the latter suggest larger negative trend (–4.0 ± 1.5%/yr) between 2005 and 2008 and smaller or insignificant changes (–0.5 ± 1.2%/yr) during 2010-2013.

54 ENVIRONMENTAL SCIENCES↗

Interspecies cross-feeding orchestrates carbon degradation in the rumen ecosystem

Because of their agricultural value, there is a great body of research dedicated to understanding the microorganisms responsible for rumen carbon degradation. However, we lack a holistic view of the microbial food web responsible for carbon processing in this ecosystem. Here, we sampled rumen-fistulated moose, allowing first of its kind access to rumen microbial communities actively degrading woody plant biomass in real-time. We resolved 1,193 viral contigs and 77 unique, near-complete microbial metagenome-assembled genomes (MAGs), many of which lacked previous metabolic insights. Plant-derived metabolites were measured with NMR and carbohydrate microarrays to quantify the carbon nutrient landscape. Network analyses directly linked measured metabolites to expressed proteins from these unique MAGs, revealing a genome-resolved three-tiered carbohydrate fueled trophic system. This provided a glimpse into microbial specialization into functional guilds defined by specific metabolites. To validate our proteomic inferences, the catalytic activity of a polysaccharide utilization locus from a highly connected metabolic hub genome was confirmed using heterologous gene expression. Viral detected proteins and linkages to microbial hosts demonstrate that phage are active controllers of rumen ecosystem function. In conclusion, our findings elucidate the microbial and viral members, as well as their metabolic interdependencies, that support in situ carbon degradation in the rumen ecosystem.

59 BASIC BIOLOGICAL SCIENCES↗

Independent saturation of three TrpRS subsites generates a partially assembled state similar to those observed in molecular simulations

Two new crystal structures of Bacillus stearothermophilus tryptophanyl-tRNA synthetase (TrpRS) afford evidence that a closed interdomain hinge angle requires a covalent bond between AMP and an occupant of either pyrophosphate or tryptophan subsite. They also are within experimental error of a cluster of structures observed in a nonequilibrium molecular dynamics simulation showing partial active-site assembly. Further, the highest energy structure in a minimum action pathway computed by using elastic network models for Open and Pretransition state (PreTS) conformations for the fully liganded TrpRS monomer is intermediate between that simulated structure and a partially disassembled structure from a nonequilibrium molecular dynamics trajectory for the unliganded PreTS. These mutual consistencies provide unexpected validation of inferences drawn from molecular simulations.

36 MATERIALS SCIENCE↗

ANALYSIS OF QUIET-SUN INTERNETWORK MAGNETIC FIELDS BASED ON LINEAR POLARIZATION SIGNALS

We present results from the analysis of Fe I 630 nm measurements of the quiet Sun taken with the spectropolarimeter of the Hinode satellite. Two data sets with noise levels of 1.2 Multiplication-Sign 10{sup -3} and 3 Multiplication-Sign 10{sup -4} are employed. We determine the distribution of field strengths and inclinations by inverting the two observations with a Milne-Eddington model atmosphere. The inversions show a predominance of weak, highly inclined fields. By means of several tests we conclude that these properties cannot be attributed to photon noise effects. To obtain the most accurate results, we focus on the 27.4% of the pixels in the second data set that have linear polarization amplitudes larger than 4.5 times the noise level. The vector magnetic field derived for these pixels is very precise because both circular and linear polarization signals are used simultaneously. The inferred field strength, inclination, and filling factor distributions agree with previous results, supporting the idea that internetwork (IN) fields are weak and very inclined, at least in about one quarter of the area occupied by the IN. These properties differ from those of network fields. The average magnetic flux density and the mean field strength derived from the 27.4% of the field of view with clear linear polarization signals are 16.3 Mx cm{sup -2} and 220 G, respectively. The ratio between the average horizontal and vertical components of the field is approximately 3.1. The IN fields do not follow an isotropic distribution of orientations.

79 ASTRONOMY AND ASTROPHYSICS↗

Multibaseline gravitational wave radiometry

We present a statistic for the detection of stochastic gravitational wave backgrounds (SGWBs) using radiometry with a network of multiple baselines. We also quantitatively compare the sensitivities of existing baselines and their network to SGWBs. We assess how the measurement accuracy of signal parameters, e.g., the sky position of a localized source, can improve when using a network of baselines, as compared to any of the single participating baselines. The search statistic itself is derived from the likelihood ratio of the cross correlation of the data across all possible baselines in a detector network and is optimal in Gaussian noise. Specifically, it is the likelihood ratio maximized over the strength of the SGWB and is called the maximized-likelihood ratio (MLR). One of the main advantages of using the MLR over past search strategies for inferring the presence or absence of a signal is that the former does not require the deconvolution of the cross correlation statistic. Therefore, it does not suffer from errors inherent to the deconvolution procedure and is especially useful for detecting weak sources. In the limit of a single baseline, it reduces to the detection statistic studied by Ballmer [Classical Quantum Gravity 23, S179 (2006).] and Mitra et al.[Phys. Rev. D 77, 042002 (2008).]. Unlike past studies, here the MLR statistic enables us to compare quantitatively the performances of a variety of baselines searching for a SGWB signal in (simulated) data. Although we use simulated noise and SGWB signals for making these comparisons, our method can be straightforwardly applied on real data.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

BitGNN: Unlocking the Performance Potential of Binary Graph Neural Networks on GPUs

Graph Neural Networks (GNNs) have shown compelling results in many graph-based learning tasks. They are, however, time-consuming. Recent work has shown a promising direction in improving GNN speed and shrinking the size — network binarization, which binarizes network values and operations. Prior work, however, mainly focused on algorithm designs, leaving it open on how to fully materialize the performance potential. This work fills the gap by proposing techniques to best map binary GNNs and their computations to fit the nature of bit manipulations, optimizations and algorithms to maximize BSpMM kernel efficiency, and solutions to other factors influencing the end-to-end time on GPUs. Results on real-world graphs show that the proposed techniques outperform state of-the-art binary GNN implementations by 21-67× with little accuracy loss.

Chen, Jou-An↗

WHISPER: Wireless Home Identification and Sensing Platform for Energy Reduction

Many regions of the world benefit from heating, ventilating, and air-conditioning (HVAC) systems to provide productive, comfortable, and healthy indoor environments, which are enabled by automatic building controls. Due to climate change, population growth, and industrialization, HVAC use is globally on the rise. Unfortunately, these systems often operate in a continuous fashion without regard to actual human presence, leading to unnecessary energy consumption. As a result, the heating, ventilation, and cooling of unoccupied building spaces makes a substantial contribution to the harmful environmental impacts associated with carbon-based electric power generation, which is important to remedy. For our modern electric power system, transitioning to low-carbon renewable energy is facilitated by integration with distributed energy resources. Automatic engagement between the grid and consumers will be necessary to enable a clean yet stable electric grid, when integrating these variable and uncertain renewable energy sources. We present the WHISPER (Wireless Home Identification and Sensing Platform for Energy Reduction) system to address the energy and power demand triggered by human presence in homes. The presented system includes a maintenance-free and privacy-preserving human occupancy detection system wherein a local wireless network of battery-free environmental, acoustic energy, and image sensors are deployed to monitor homes, record empirical data for a range of monitored modalities, and transmit it to a base station. Several machine learning algorithms are implemented at the base station to infer human presence based on the received data, harnessing a hierarchical sensor fusion algorithm. Results from the prototype system demonstrate an accuracy in human presence detection in excess of 95%; ongoing commercialization efforts suggest approximately 99% accuracy. Using machine learning, WHISPER enables various applications based on its binary occupancy prediction, allowing situation-specific controls targeted at both personalized smart home and electric grid modernization opportunities.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Binary Black Hole Mergers in the First Advanced LIGO Observing Run

The first observational run of the Advanced LIGO detectors, from September 12, 2015 to January 19, 2016, saw the first detections of gravitational waves from binary black hole mergers. In this paper we present full results from a search for binary black hole merger signals with total masses up to 100M solar mass and detailed implications from our observations of these systems. Our search, based on general-relativistic models of gravitational wave signals from binary black hole systems, unambiguously identified two signals, GW150914 and GW151226, with a significance of greater than 5 alpha over the observing period. It also identified a third possible signal, LVT151012, with substantially lower significance, which has a 87 probability of being of astrophysical origin. We provide detailed estimates of the parameters of the observed systems. Both GW150914 and GW151226 provide an unprecedented opportunity to study the two-body motion of a compact-object binary in the large velocity, highly nonlinear regime. We do not observe any deviations from general relativity, and place improved empirical bounds on several high-order post-Newtonian coefficients. From our observations we infer stellar-mass binary black hole merger rates lying in the range 9-240 Gpc-3 yr-1. These observations are beginning to inform astrophysical predictions of binary black hole formation rates, and indicate that future observing runs of the Advanced detector network will yield many more gravitational wave detections.

Abbott, B. P.↗

PM 2.5 Concentrations over Major Metropolitan Regions Inferred from Airborne High Spectral Resolution Lidar Measurements Using Machine Learning Regression

We use measurements of near-surface aerosol backscatter, extinction, and depolarization acquired by four NASA Langley Research Center airborne High Spectral Resolution Lidars (HSRLs) to develop a machine learning regression methodology to infer PM2.5 concentrations at the surface and aloft. These airborne HSRL measurements were acquired over major metropolitan regions in the United States and Asia during more than 170 flights since 2010. Hourly surface PM2.5 measurements from the EPA air quality system and similar networks in other countries acquired within 10 km and 15 minutes of these near-surface HSRL measurements are used to train models that compute PM2.5 concentrations from the HSRL measurements. We examine several regression methods and find that exponential Gaussian Process algorithms consistently give the best performance in terms of the lowest root-mean-square (RMS) errors and the highest correlations. Model performance varies significantly depending on various combinations of HSRL aerosol measurements (e.g., aerosol backscatter, extinction, depolarization, backscatter color ratios, lidar ratios, aerosol optical thickness) and retrievals (e.g., mixed layer height, aerosol type) used in the regressions. Models that use near-surface measurements of aerosol backscatter and aerosol intensive properties such as depolarization, backscatter color ratio, and lidar ratio typically give the best performance with RMS errors around 4 mg/m3 and correlation coefficients above 0.9. HSRL measurements were often acquired when the aircraft flew systematic “raster-scan” patterns for several hours over these cities. These flight patterns enabled measurements of the spatial, temporal, and vertical variabilities in the distributions of aerosol backscatter and aerosol intensive properties and allowed us to derive the corresponding variabilities in PM2.5 concentrations. We present examples of such variabilities over urban areas in the United States as well as Asia. We describe also how the distribution of surface PM2.5 varies with aerosol type and use these retrievals to examine model simulations of surface PM2.5 in these metropolitan regions. We also discuss how this methodology may be applied to measurements from satellite lidars such as CALIOP on CALIPSO and ATLID on EarthCARE.

lidar↗

Inferring Instantaneous, Multivariate and Nonlinear Sensitivities for the Analysis of Feedback Processes in a Dynamical System: Lorenz Model Case Study

A new approach is presented for the analysis of feedback processes in a nonlinear dynamical system by observing its variations. The new methodology consists of statistical estimates of the sensitivities between all pairs of variables in the system based on a neural network modeling of the dynamical system. The model can then be used to estimate the instantaneous, multivariate and nonlinear sensitivities, which are shown to be essential for the analysis of the feedbacks processes involved in the dynamical system. The method is described and tested on synthetic data from the low-order Lorenz circulation model where the correct sensitivities can be evaluated analytically.

Aires, Filipe↗

Reliable extrapolation of deep neural operators informed by physics or sparse observations

Deep neural operators can learn nonlinear mappings between infinite-dimensional function spaces via deep neural networks. As promising surrogate solvers of partial differential equations (PDEs) for real-time prediction, deep neural operators such as deep operator networks (DeepONets) provide a new simulation paradigm in science and engineering. Pure data-driven neural operators and deep learning models, in general, are usually limited to interpolation scenarios, where new predictions utilize inputs within the support of the training set. However, in the inference stage of real-world applications, the input may lie outside the support, i.e., extrapolation is required, which may result to large errors and unavoidable failure of deep learning models. Here, we address this challenge of extrapolation for deep neural operators. First, we systematically investigate the extrapolation behavior of DeepONets by quantifying the extrapolation complexity, via the 2-Wasserstein distance between two function spaces and propose a new strategy of bias–variance trade-off for extrapolation with respect to model capacity. Subsequently, we develop a complete workflow, including extrapolation determination, and we propose five reliable learning methods that guarantee a safe prediction under extrapolation by requiring additional information—the governing PDEs of the system or sparse new observations. The proposed methods are based on either fine-tuning a pre-trained DeepONet or multifidelity learning. We demonstrate the effectiveness of the proposed framework for various types of parametric PDEs. Furthermore, our systematic comparisons provide practical guidelines for selecting a proper extrapolation method depending on the available information, desired accuracy, and required inference speed.

42 ENGINEERING↗