Search NASA⌕ Search

SEARCH · Search NASA

Results for “Network Models”

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 919 records · Page 51

Large Scale Caching and Streaming of Training Data for Online Deep Learning

The training of deep neural network models on large data remains a difficult problem, despite progress towards scalable techniques. In particular, there is a mismatch between the random but predetermined order in which AI flows select training samples and the streaming I/O patterns for which traditional HPC data storage (e.g., parallel file systems) are designed. In addition, as more data are obtained, it is feasible neither simply to train learning models incrementally, due to catastrophic forgetting (i.e., bias towards new samples), nor to train frequently from scratch, due to prohibitive time and/or resource constraints. In this paper, we study data management techniques that combine caching and streaming with rehearsal support in order to enable efficient access to training samples in both offline training and continual learning. We revisit state-of-art streaming approaches based on data pipelines that transparently handle prefetching, caching, shuffling, and data augmentation, and discuss the challenges and opportunities that arise when combining these methods with data-parallel training techniques. We also report on preliminary experiments that evaluate the I/O overheads involved in accessing the training samples from a parallel file system (PFS) under several concurrency scenarios, highlighting the impact of the PFS on the design of the data pipelines.

data pipelines↗

Towards On-Chip Learning for Low Latency Reasoning with End-to-End Synthesis

The Software Defined Architectures (SODA) Synthesizer is an open-source compiler-based tool able to automatically generate domain-specialized systems targeting Application-Specific Integrated Circuits (ASICs) or Field Programmable Gate Arrays (FPGAs) starting from high-level programming. SODA is composed of a frontend, SODA-OPT, which leverages the multilevel intermediate representation (MLIR) framework to interface with productive programming tools (e.g., machine learning frame-works), identify kernels suitable for acceleration, and perform high-level optimizations, and of a state-of-the-art high-level synthesis backend, Bambu from the PandA framework, to generate custom accelerators. One specific application of the SODA Synthesizer is the generation of accelerators to enable ultra-low latency inference and control on autonomous systems for scientific discovery (e.g., electron microscopes, sensors in particle accelerators, etc.). This paper provides an overview of the flow in the context of the generation of accelerators for edge processing to be integrated in transmission electron microscopy (TEM) devices, focusing on use cases from precision material synthesis. We show the tool in action with an example of design space exploration for inference on reconfigurable devices with a conventional deep neural network model (LeNet). Finally, we discuss the research directions and opportunities enabled by SODA in the area of autonomous control for scientific experimental workflows.

Castellana, Vito G.↗

Mechanical Characterization of Electrolyzer Membranes and Components Under Compression

Proton-exchange membrane (PEM) water electrolysis is a promising technology for producing clean hydrogen by electrochemically splitting water when paired with renewable energy sources. A major roadblock to improving electrolyzer durability is the mechanical degradation of the cell components, which requires an understanding of their mechanical response under device-relevant conditions. However, there is a lack of studies on the mechanical characterization of the PEM and other components, as well as and their interactions. This study aims to address this gap by using a custom-designed testing apparatus to investigate the mechanics of electrolyzer components in uniaxial compression at 25 and 80 °C. Findings show stress-strain response of components have a varying degree of nonlinearity owing to their distinct deformation mechanisms and morphologies, from porous structures to polymers. These results are used to develop an expression for compressive stress-strain response of Nafion membranes and then analyze the deformation of components under applied pressure by using a 1-D spring network model of cell assembly. This work provides a new understanding of mechanical responses of the electrolyzer membrane and cell components, which can help assess material design and cell assembly strategies for improved electrolyzer durability.

08 HYDROGEN↗

AirflowNetwork [SWR 19-45]

The AirflowNetwork software provides the ability to simulate air movement and the resulting transport of contaminants and other materials within a building. The performance of air distribution systems, including supply and return leaks, and multizone airflows driven by outdoor wind and forced air during HVAC system operation are simulated using a nodal network representation of the building. The pressure network model implemented here is based on George Walton's AIRNET and was modified for inclusion in the EnergyPlus building energy modeling software. The algorithm were then extracted from EnergyPlus into a stand-alone package to facility use without the building energy model and more extensive testing and validation of the underlying operations.

DeGraw, Jason↗

ccp137/SUGAR

This package (SUrface-wave Grader with ARtificial intelligence or SUGAR) automatically assigns a quality score to surface-wave seismograms (SAC format) using a trained artificial neural network model (included). Specifically, the python script 01_apply_ann.py calculates probability scores for a list of SAC files. You may consider seismograms with probability scores larger than 0.5 as acceptable data. Note no scores will be given to seismograms that do not pass an initial check (e.g., insufficient number of data points). See https://doi.org/10.1002/essoar.10507941.3 for more details.

Chai, Chengping [Oak Ridge National Lab. (ORNL), O↗

Neural Networks for Supercapacitor Modeling

This software implements a one-dimensional model of a supercapacitor and calibrates it to experimental data. A system of coupled partial differential equation models the salt concentration and the ionic and electronic potentials. The term in the equation modeling the capacitance contains a neural network, whose parameters are optimized to fit the provided experimental data. The software is written in python using the open-source library JAX to calculate the gradients needed for the optimization.

Salazar de Troya, Miquel↗

BlendPATH (Blending Pipeline Analysis Tool for Hydrogen) [SWR-24-10]

BlendPATH provides case-by-case techno-economic analysis for potential projects where hydrogen is blended into a natural gas transmission pipeline. BlendPATH estimates 1) the transmission pipeline modifications and operating conditions necessary to blend hydrogen to a user specified volume faction of hydrogen in pipeline gas and 2) the incremental capital and operating expenses to prepare said transmission pipeline for hydrogen blending. BlendPATH is developed in Python and requires SAInt, a underlying commercial natural gas pipeline network modeling software, to run. Pipeline modification and operation condition estimation is guided by ASME B31.12. The intent of this software is to target application for projects in the initial project assessment stage and provide the user with the capability of assess promising opportunities before the use proceeds with further detailed pipeline evaluation based on a probable economic outcome.

Kee, Jamie↗

MalGen

MalGen is helper code that includes a script for loading data in format for PyTorch neural network models. It also includes example models. Sandia National Laboratories is a multimission laboratory managed and operated by National Technology & Engineering Solutions of Sandia, LLC, a wholly owned subsidiary of Honeywell International Inc., for the U.S. Department of Energy’s National Nuclear Security Administration under contract DE-NA0003525. SAND2021-3078 O

Johnson, NicholasT↗

ORNL-Chi-Geometry

Library for benchmarking neural network models on classification tasks for chirality detection in atomistic structures of organic compounds.

Weaver, Rylie [Oak Ridge National Laboratory (ORNL↗

fast3

Code to train neural network models for binding affinity prediction

Kim, Hyojin [Lawrence Livermore National Laborator↗

Simulation as a sustainable trading zone: Aiming at intergenerational justice

The paper, drawing on the example of simulation codes used in nuclear physics and high-energy physics, seeks to call attention to the ethical implications of discontinuing support for simulation codes and the loss of knowledge embodied in them. Predicated on the concept of trading zones and actor network models, the paper addresses the problem of extinction of simulation codes and attempts to understand their evolution and development within those frameworks. We show that simulation codes of closed type develop to the level of creoles, becoming local languages and standards of scientific centers and disappearing as their few main developers leave, whereas codes of open types become universal languages, imposing problemsolving patterns on the entire community and crowding out other codes. The paper suggests that because of simulations’ reliance on tacit knowledge, practices entrenched in codes cannot be exhaustively explicated or transmitted through writing alone; on the contrary, the life cycle of a simulation code is determined by the life cycle of its trading zone. We examine the extent to which both of these phenomena pose a risk to the preservation of knowledge. Bearing upon intergenerational ethics, we draw analogies between the pure intergenerational problem (PIP) and the problem of preserving the knowledge implemented in simulation codes and transmitting it to future generations. We argue that for the complete transfer of knowledge, it is necessary to develop and maintain inhabitability and sustainability of simulation trading zones in a controllable way, at least until the demand for these codes is warranted to cease in the future.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

The Artificial Intelligence Ontology: LLM-Assisted Construction of AI Concept Hierarchies

The Artificial Intelligence Ontology (AIO) is a systematization of artificial intelligence (AI) concepts, methodologies, and their interrelations. Developed via manual curation, with the additional assistance of large language models (LLMs), AIO aims to address the rapidly evolving landscape of AI by providing a comprehensive framework that encompasses both technical and ethical aspects of AI technologies. The primary audience for AIO includes AI researchers, developers, and educators seeking standardized terminology and concepts within the AI domain. We use the term “branches” for classes, and their subclasses, in our ontology that are subclasses of owl:Thing. AIO contains eight branches: Bias, Layer, Machine Learning Task, Mathematical Function, Model, Network, Preprocessing, and Training Strategy, each designed to support the modular composition of AI methods and facilitate a deeper understanding of deep learning architectures and ethical considerations in AI. AIO uses the Ontology Development Kit (ODK) for its creation and maintenance, with its content being more easily updated through AI-driven curation support. This approach not only ensures the ontology's relevance amidst the fast-paced advancements in AI but also significantly enhances its utility for researchers, developers, and educators by simplifying the integration of new AI concepts and methodologies. The ontology's utility is demonstrated through the annotation of AI methods data in a catalog of AI research publications and the integration into the BioPortal ontology resource, highlighting its potential for cross-disciplinary research. The AIO ontology is open source and is available on GitHub ( https://w3id.org/aio/ ) and BioPortal ( https://bioportal.bioontology.org/ontologies/AIO ).

Joachimiak, Marcin P. [Biosystems Data Science Dep↗

Network inference reveals novel connections in pathways regulating growth and defense in the yeast salt response

Cells respond to stressful conditions by coordinating a complex, multi-faceted response that spans many levels of physiology. Much of the response is coordinated by changes in protein phosphorylation. Although the regulators of transcriptome changes during stress are well characterized in Saccharomyces cerevisiae , the upstream regulatory network controlling protein phosphorylation is less well dissected. Here, we developed a computational approach to infer the signaling network that regulates phosphorylation changes in response to salt stress. We developed an approach to link predicted regulators to groups of likely co-regulated phospho-peptides responding to stress, thereby creating new edges in a background protein interaction network. We then use integer linear programming (ILP) to integrate wild type and mutant phospho-proteomic data and predict the network controlling stress-activated phospho-proteomic changes. The network we inferred predicted new regulatory connections between stress-activated and growth-regulating pathways and suggested mechanisms coordinating metabolism, cell-cycle progression, and growth during stress. We confirmed several network predictions with co-immunoprecipitations coupled with mass-spectrometry protein identification and mutant phospho-proteomic analysis. Results show that the cAMP-phosphodiesterase Pde2 physically interacts with many stress-regulated transcription factors targeted by PKA, and that reduced phosphorylation of those factors during stress requires the Rck2 kinase that we show physically interacts with Pde2. Conclusively, our work shows how a high-quality computational network model can facilitate discovery of new pathway interactions during osmotic stress.

59 BASIC BIOLOGICAL SCIENCES↗

Evaluating the limitations of Bayesian metabolic control analysis

Bayesian Metabolic Control Analysis (BMCA) is a promising framework for inferring metabolic control coefficients in data-limited scenarios, combining Bayesian inference with linear-logarithmic (lin-log) rate laws. These metabolic control coefficients quantify how changes in enzyme activities affect steady-state fluxes and metabolite concentrations across a metabolic network. However, its predictive accuracy and limitations remain underexplored. This study systematically evaluates BMCA’s ability to infer elasticity values, flux control coefficients (FCC), and concentration control coefficients (CCC) under varying data availability conditions using three synthetic metabolic network models. We demonstrate that BMCA predictions are highly dependent on the inclusion of flux and enzyme concentration data, with the omission of these datasets leading to severe inaccuracies. In our synthetic, enzyme-perturbation datasets, external metabolite concentrations had minimal impact and, in some cases, their exclusion improved predictions; when external-nutrient perturbations were introduced and those concentrations were observed, gains were at most modest. Additionally, we find that posterior estimation with both ADVI and HMC can underestimate large-magnitude elasticities in our synthetic settings, with ADVI showing somewhat higher variance under strong up-regulation; thus, recovering |elasticity| ≳ 1.5 remains challenging regardless of the inference engine. ADVI also fails to accurately infer allosteric interactions, even when regulatory effects are strong. While BMCA maintains reasonable accuracy in partially recovering the rankings of the highest FCC values, its estimates of absolute values remain constrained by prior assumptions and data limitations. Our findings reveal the BMCA algorithm’s strengths and weaknesses, providing guidance on its application in metabolic engineering, and highlighting the need for methodological refinements to enhance its predictive capabilities.

59 BASIC BIOLOGICAL SCIENCES↗

Utah FORGE Project 3-2417: Simulations for Distributed Acoustic Sensing Strain Signatures as an Indicator of Fracture Connectivity

This dataset encompasses simulations of strain signatures from both hydraulically connected and "near-miss" fractures in enhanced geothermal systems (EGS). The files and results are presented from the perspective of digital acoustic sensing's (DAS) potential to differentiate the two fracture types. This dataset was acquired by the FOGMORE R&D project (Fiber Optic Geophysical MOnitoring of Reservoir Evolution), under Utah FORGE R&D Project 3-2417. Included are simulation and results via MatLab and COMSOL files, as well as a thesis and paper summarizing the results. Some stimulated fractures may be incomplete, approaching but not intersecting the production well. These "near-miss" fractures can be addressed in future stimulation stages or re-stimulated to complete the connection. We propose the use of fiber optic distributed acoustic sensing (DAS) as a method by which near-miss stimulated fractures may be identified and distinguished from hydraulically connected fractures. The low-frequency sub-nanostrain signatures of both complete and near-miss fractures in DAS data are simulated in this study using a hydrogeomechanical discrete fracture network model. The spatial distribution of strain was found to be an accurate indicator. However, this indicator must be evaluated in the context of DAS gauge length and spatial sampling. These simulations are a precursor to tests conducted at FORGE in 2023.

15 GEOTHERMAL ENERGY↗

The genotype-phenotype landscape of an allosteric protein

Allostery is a fundamental biophysical mechanism that underlies cellular sensing, signaling, and metabolism. Yet a quantitative understanding of allosteric genotype-phenotype relationships remains elusive. Here, we report the large-scale measurement of the genotype-phenotype landscape for an allosteric protein: the lac repressor from Escherichia coli , LacI. Using a method that combines long-read and short-read DNA sequencing, we quantitatively measure the dose-response curves for nearly 10 5 variants of the LacI genetic sensor. The resulting data provide a quantitative map of the effect of amino acid substitutions on LacI allostery and reveal systematic sequence-structure-function relationships. We find that in many cases, allosteric phenotypes can be quantitatively predicted with additive or neural-network models, but unpredictable changes also occur. For example, we were surprised to discover a new band-stop phenotype that challenges conventional models of allostery and that emerges from combinations of nearly silent amino acid substitutions.

59 BASIC BIOLOGICAL SCIENCES↗

Measuring, Predicting and Visualizing Short-Term Change in Word Representation and Usage in VKontakte Network

Language in social media is extremely dynamic: new concepts emerge, trend and disappear, while the meaning of existing concepts can fluctuate over time. This work addresses a novel task of measuring and predicting meaning shift – context or semantic representation change from surface level dynamics e.g., word frequencies observed in social media streams. We study the relationship between language dynamics and short-term context change on a corpus of VKontakte posts collected during the Russia-Ukraine crisis in 2014 – 2015. We train a neural network model to predict short-term shifts in meaning from previous meaning as well as word dynamics. We demonstrate that short-term representation shift can be accurately predicted up to several weeks in advance. Moreover, we visualize this short-term context shift for the example words and demonstrate the practical side of our approach to discover and track meaning of newly emerging terms in social media during crises.

distributional semantics, word embeddings, social ↗

Bayesian inference of Stochastic reaction networks using Multifidelity Sequential Tempered Markov Chain Monte Carlo

Stochastic reaction network models are often used to explain and predict the dynamics of gene regulation in single cells.These models usually involve several parameters, such as the kinetic rates of chemical reactions, that are not directly measurable and must be inferred from experimental data. Bayesian inference provides a rigorous probabilistic frame-work for identifying these parameters by finding a posterior parameter distribution that captures their uncertainty.Traditional computational methods for solving inference problems such as Markov Chain Monte Carlo methods based on classical Metropolis-Hastings algorithm involve numerous serial evaluations of the likelihood function, which in turn requires expensive forward solutions of the chemical master equation (CME). We propose an alternate approach based on a multifidelity extension of the Sequential Tempered Markov Chain Monte Carlo (ST-MCMC) sampler. This algorithm is built upon Sequential Monte Carlo and solves the Bayesian inference problem by decomposing it into a sequence of efficiently solved subproblems that gradually increase both model fidelity and the influence of the observed data. We reformulate the finite state projection (FSP) algorithm, a well-known method for solving the CME, to produce a hierarchy of surrogate master equations to be used in this multifidelity scheme. To determine the appropriate fidelity, we introduce a novel information-theoretic criteria that seeks to extract the most information about the ultimate Bayesian posterior from each model in the hierarchy without inducing significant bias. This novel sampling scheme is tested with high performance computing resources using biologically relevant problems.

97 MATHEMATICS AND COMPUTING↗