Results for “hybrid learning” — Page 30

NASA NTRS2014

MARGInS: Model-Based Analysis of Realizable Goals in Systems

Under NASAs Constellation effort, the Exploration Technology Development Program funded research toward a system validation capability that applied machine learning and test-case generation techniques to the analysis of black-box system behavior. The behavior analysis capability scaled to spaces of hundreds of input parameters and tens of thousands of test cases. Aerospace systems at the vehicle level, especially those systems which contain some level of autonomy, are best described by hybrid and non-linear mathematics. Even simplified models of such systems need parameter dimensionalities in the hundreds or thousands of parameters in order to capture sufficient fidelity. The System Safety Assessments (such as those described in the SAE ARP 4761A Safety Assessment Process guidelines) for these systems are prone to errorinteractions between the vehicles subsystems are complex, and can display emergent behaviors. NASA captured this new analysis in the Model-based Analysis of Realizable Goals in Systems (MARGInS) tool and applied it to the Pad Abort 1 (PA-1) simulation as part of the independent validation and verification cycle before the PA-1 flight test in May of 2010. MARGInS evaluated the adherence of the high-fidelity simulation to its requirements, and deter- mined the margins to failure from the expected nominal input conditions. Following the PA-1 test, the capabilities within the MARGInS framework have been extended with sophisticated statistical and white-box test case generation techniques and applied to other NASA missions. The frame- work now includes a critical factors analysis that was applied to NASAs Orion simulation and design. NASAs Aeronautics Research Mission Directorate (ARMD) leveraged the existing MARGInS framework for work on aviation safety for civil transport vehicles and for research on autonomy issues. The NASA ARMD effort created a time series output prediction capability that has been used to characterize trajectories for a plane with an adaptive control system, and a safety boundary detection capability that has been applied to an air traffic control concept of operation for the Federal Aviation Administration. The statistical and machine- learning based techniques within MARGInS have been successfully combined with concolic execution to improve the coverage of a critical unit by driving system-level inputs. The use case driving the concolic execution and MARGInS integration was inspired by the Air France 447 disaster in which the loss of a critical functionality (the airspeed calculation from the pitot tubes) led to loss of the entire plane with the people aboard. To illustrate capabilities and limitations, we will highlight the analyses for the applications listed above. We will then discuss the future plans for MARGInS and its interfaces with other tools.

Validation↗

DOE OSTI2021

HAM: Hotspot-Aware Manager for Improving Communications with 3D-Stacked Memory

merging High-Performance Computing (HPC) workloads, such as graph analytics, machine learning, and big data science, are data-intensive. Data-intensive workloads usually present fine-grained memory accesses with limited or no data locality, and thus incur frequent cache misses and low utilization of memory bandwidth. 3D-stacked memory devices such as Hybrid Memory Cube (HMC) and High Bandwidth Memory (HBM) can provide significantly higher bandwidth than conventional memory modules. However, the traditional interfaces and optimization methods for JEDEC DDR devices do not allow to fully exploit the potential performance of 3D-stacked memory with the massive amount of irregular memory accesses of data-intensive applications. In this paper, we propose a novel Hotspot-Aware Manager (HAM) infrastructure for 3D-stacked memory devices capable of optimizing memory access streams via request aggregation, hotspot detection, and in-memory prefetching. %and an associated hotspot-aware page policy. We present the HAM design and implementation, and simulate it on a system using RISC-V embedded cores with attached HMC devices. We extensively evaluate HAM with over 12 benchmarks and applications representing diverse irregular memory access patterns. The results show that, on average, HAM reduces redundant requests by 37.51\% and increases the prefetch buffer hit rate by 4.2 times, compared to a baseline streaming prefetcher. On the selected benchmark set, HAM provides performance gains of 21.81\% in average (up to 34.28\%) and power savings of 35.07\% over a standard 3D-stacked memory.

Wang, Xi↗

DOE OSTI2023

Digital Modeling on Large Kernel Metamaterial Neural Network

Deep neural networks (DNNs) utilized recently are physically deployed with computational units (e.g., CPUs and GPUs). Such a design might lead to a heavy computational burden, significant latency, and intensive power consumption, which are critical limitations in applications such as Internet of Things (IoT), edge computing, and usage of drones. Recent advances in optical computational units (e.g., metamaterial) have shed light on energy-free and light-speed neural networks. However, the digital design of the metamaterial neural network (MNN) is fundamentally limited by its physical limitations, such as precision, noise, and bandwidth during fabrication. Moreover, the unique advantages of MNN’s (e.g., light-speed computation) are not fully explored via standard 3×3 convolution kernels. In this paper, we propose a novel large kernel metamaterial neural network (LMNN) that maximizes the digital capacity of the state-of-the-art (SOTA) MNN with model re-parametrization and network compression, while also considering the optical limitation explicitly. The new digital learning scheme can maximize the learning capacity of MNN while modeling the physical restrictions of meta-optics. With the proposed LMNN, the computation cost of the convolutional front-end can be offloaded to fabricated optical hardware. The experimental results on two publicly available datasets demonstrate that the optimized hybrid design improved classification accuracy while reducing computational latency. In conclusion, the development of the proposed LMNN is a promising step towards the ultimate goal of energy-free and light-speed AI.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2023

DeePMD-kit v2: A software package for deep potential models

DeePMD-kit is a powerful open-source software package that facilitates molecular dynamics simulations using machine learning potentials known as Deep Potential (DP) models. This package, which was released in 2017, has been widely used in the fields of physics, chemistry, biology, and material science for studying atomistic systems. The current version of DeePMD-kit offers numerous advanced features, such as DeepPot-SE, attention-based and hybrid descriptors, the ability to fit tensile properties, type embedding, model deviation, DP-range correction, DP long range, graphics processing unit support for customized operators, model compression, non-von Neumann molecular dynamics, and improved usability, including documentation, compiled binary packages, graphical user interfaces, and application programming interfaces. This article presents an overview of the current major version of the DeePMD-kit package, highlighting its features and technical details. Additionally, this article presents a comprehensive procedure for conducting molecular dynamics as a representative application, benchmarks the accuracy and efficiency of different models, and discusses ongoing developments.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2021

Machine Learning for a-posteriori model-observed data fusion to enhance predictive value of ESM output

The proposed research is consistent with focal areas 2 and 3: when successful, this effort will enhance the predictive value of Earth System Model (ESM) output by identifying and quantifying its systematic deviations from available observations and by correcting and recalibrating to existing data. The product is a hybrid process & data-driven modeling tool whose estimation can also bring insights through pattern discovery recognition on model deficiencies on the one hand and data gaps on the other.

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2022

Computer Science Research Needs for Parallel Discrete Event Simulation (PDES)

Historically, scientific computing efforts have demonstrated the clear need for, and effective use of, supercomputing with traditional time-stepped simulations. Nevertheless, there are several areas in the mission spaces of the U.S. Department of Energy and other agencies waiting to tap advanced computing research using a different, discrete event style of modeling, simulation, and analysis. These span a wide spectrum of applications including energy grid resilience, urban planning and policy, transportation science, building technologies, emergency response and planning, environmental impact analysis, computational epidemiology, Internet communications, cyber security, and cyber-physical systems, to name only a few. Even within traditional scientific applications, the role of discrete event modes of execution is increasing in the form of new event-based mathematical solvers such as quantized state integration methods and discrete-continuous hybrid system solvers. Co-design of advanced supercomputing hardware systems is another area that exploits discrete event simulation at its core for effective analyses. Complex systems, entity behaviors and interconnections play a significant role in all these applications, which are mapped to large-scale models with discrete event formulations. To make advancements in all the aforementioned scientific areas, many technical aspects need to be more thoroughly studied and deeply understood in parallel discrete event simulation (PDES). The unique dynamics inherent in a discrete event modeling approach, by their very nature, intersect and influence the entire stack of the computing system, including (a) the unique nature of the instruction sets exercised in PDES workloads without a predominance of high-precision floating point operations, (b) virtual time-constrained multi-threaded execution of many logical processes per processor, (c) extremely variable and difficult to predict network traffic characteristics, (d) interfaces and inter-dependencies with machine learning and artificial intelligence codes at higher software layers, and (e) highly challenging load balancing needs, especially in effectively accounting for accelerated/extremely heterogeneous computing in current and future high-performance computing systems. Efficient and accurate parallel execution of PDES workloads is also dominated by challenges in dealing with their asynchronous concurrency fundamentally present at the model level. Conservative synchronization, optimistic/speculative synchronization, and their hybrid schemes open new questions in fundamental computer science with respect to reversibility of computation and prediction (lookahead) of behaviors inherent within model codes. On the implementation front, there are relatively few scalable, general-purpose parallel discrete event simulators in the world, and even fewer have been studied on emerging hardware platforms. To enable scientific advances using PDES, the research needs in computer science must also be pursued and met in the intersection of the algorithmic and hardware-aware aspects of scalable PDES engines. This report is aimed at capturing a computer science-oriented view of this important area of research in PDES, presenting a sample of important applications with their inherent discrete event technology elements. Needs are outlined in core areas of parallel discrete event research as well as cross-cutting directions in computer science research that positively impact scientific advancements across several important application areas. A selection of priority research opportunities in advanced computing for PDES is identified to serve as reference for key research topics and their order of importance for scientific advancements.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2025

NNL.Fe.qSNAP-ZBL.2024.1: A Fe Spectral Neighbor Analysis Potential for Radiation Damage Simulations

The NNL.Fe.qSNAP-ZBL.2024.1 machine-learned potential (MLP) has been generated to support the development of an elemental body-centered cubic (BCC) Fe athermal recombination corrected neutron damage model and simulations of primary recoil atom (PRA) cascades in BCC Fe. This MLP is a quadratic spectral neighbor analysis potential (qSNAP) hybridized with the universal Ziegler-Beirsack-Littmark (ZBL) potential at short-range and is named according to Naval Nuclear Laboratory MLP naming conventions (NNL.material-system.MLP-type.year.version). Training set calculations for Fe are presented along with the subsequent MLP fitting procedure. A key criterion of the fitting procedure is that ZBL describes the short-range interaction with minimal impact on the MLP. The MLP is compared to density functional theory (DFT) predicted properties relevant to radiation damage simulation, including threshold displacement energies, for validation. The NNL.Fe.qSNAP-ZBL.2024.1 potential is considered suitable for molecular dynamics (MD) simulations of radiation defects up to 800 K and PRA cascades in BCC Fe up to around 10 keV. The potential can additionally be used on a limited basis for recoils of 10–20 keV, within which range the emergence of structures outside the training set in cascade simulations may cause system instabilities.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

NASA NTRSDate not supplied

Hyperspectral Sounder Spectral Fingerprinting: Using Machine Learning Techniques to Enhance Model-Based Physical Inversion

Different retrieval algorithms have been developed to process top-of-atmosphere (TOA) spectral radiance data provided by hyperspectral infrared sounder missions. Those algorithms are either optimal estimation method (OEM) based schemes with radiative transfer calculation involved in the retrieval process, or machine learning based methods that allow ultra-efficient data procession but lack of radiometric consistency validation based on the directly measured information. Combining both approaches leverages their respective technical advantages, leading to more accurate results. This study introduces a hyperspectral sounder fingerprinting algorithm to explore this hybrid approach. This approach involves the use of a spectral information-based classification method to identify an reference geophysical state and the corresponding radiative kernel. This enables the efficient retrieval of geophysical variables of interest through a radiative kernel-based linear inversion procedure. The fingerprinting method has been applied to analyze a decade-long hyperspectral sounder data record.

Wan Wu↗

DOE OSTI2025

A Digital Twin Framework Utilizing Machine Learning for Robust Predictive Maintenance: Enhancing Tire Health Monitoring

We introduce a novel digital twin (DT) framework for the predictive maintenance of long-term physical systems. Using monitoring tire health as an application, we show how the DT framework can be used to enhance automotive safety and efficiency, and how the technical challenges can be overcome using a three-step approach. First, to manage the data complexity over a long operation span, we employ data reduction techniques to concisely represent physical tires using historical performance and usage data. Relying on these data, for fast real-time prediction, we train a transformer-based model offline on our concise dataset to predict future tire health over time, represented as remaining casing potential (RCP). Based on our architecture, our model quantifies both epistemic and aleatoric uncertainties, providing reliable confidence intervals around predicted RCP. Second, to incorporate real-time data, we update the predictive model in the DT framework, ensuring its accuracy throughout its lifespan with the aid of hybrid modeling and the use of the discrepancy function. Third, to assist decision-making in predictive maintenance, we implement a tire state decision algorithm, which strategically determines the optimal timing for tire replacement based on RCP forecasted by our transformer model. This approach ensures that our DT accurately predicts system health, continually refines its digital representation, and supports predictive maintenance decisions. Furthermore, our framework effectively embodies a physical system, leveraging big data and machine learning (ML) for predictive maintenance, model updates, and decision-making.

advanced computing infrastructure↗

DOE OSTI2021

MULTICOM2 open-source protein structure prediction system powered by deep learning and distance prediction

Protein structure prediction is an important problem in bioinformatics and has been studied for decades. However, there are still few open-source comprehensive protein structure prediction packages publicly available in the field. In this paper, we present our latest open-source protein tertiary structure prediction system—MULTICOM2, an integration of template-based modeling (TBM) and template-free modeling (FM) methods. The template-based modeling uses sequence alignment tools with deep multiple sequence alignments to search for structural templates, which are much faster and more accurate than MULTICOM1. The template-free (ab initio or de novo) modeling uses the inter-residue distances predicted by DeepDist to reconstruct tertiary structure models without using any known structure as template. In the blind CASP14 experiment, the average TM-score of the models predicted by our server predictor based on the MULTICOM2 system is 0.720 for 58 TBM (regular) domains and 0.514 for 38 FM and FM/TBM (hard) domains, indicating that MULTICOM2 is capable of predicting good tertiary structures across the board. It can predict the correct fold for 76 CASP14 domains (95% regular domains and 55% hard domains) if only one prediction is made for a domain. The success rate is increased to 3% for both regular and hard domains if five predictions are made per domain. Moreover, the prediction accuracy of the pure template-free structure modeling method on both TBM and FM targets is very close to the combination of template-based and template-free modeling methods. This demonstrates that the distance-based template-free modeling method powered by deep learning can largely replace the traditional template-based modeling method even on TBM targets that TBM methods used to dominate and therefore provides a uniform structure modeling approach to any protein. Finally, on the 38 CASP14 FM and FM/TBM hard domains, MULTICOM2 server predictors (MULTICOM-HYBRID, MULTICOM-DEEP, MULTICOM-DIST) were ranked among the top 20 automated server predictors in the CASP14 experiment. After combining multiple predictors from the same research group as one entry, MULTICOM-HYBRID was ranked no. 5. The source code of MULTICOM2 is freely available at https://github.com/multicom-toolbox/multicom/tree/multicom_v2.0 .

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2023

LCA-PyTorch

LCA-PyTorch is a code repository which contains PyTorch implementations of the Locally Competitive Algorithm (LCA), which is a biologically-plausible sparse coding model. LCA-PyTorch allows for the training, testing, and analysis of single layer LCA networks, multi-layer LCA networks, and hybrid LCA-based deep neural network models on a wide variety of applications and data types. LCA-PyTorch was developed in Python, a high-level programming language that takes advantage of the Python ecosystem of high-quality open-source packages for machine learning. LCA-PyTorch interfaces heavily with the open-source PyTorch Python package.

Teti, Michael↗

DOE OSTI2026

When does global attention help: a unified empirical study on atomistic graph learning

Graph neural networks (GNNs) are widely used as surrogates for costly experiments and first-principles simulations to study the behavior of compounds at atomistic scale, and their architectural complexity is constantly increasing to enable the modeling of complex physics. While most recent GNNs combine more traditional message passing neural networks (MPNNs) layers to model short-range interactions with more advanced graph transformers (GTs) with global attention mechanisms to model long-range interactions, it is still unclear when global attention mechanisms provide real benefits over well-tuned MPNN layers due to inconsistent implementations, features, or hyperparameter tuning. We introduce the first unified, reproducible benchmarking framework–built on HydraGNN–that enables seamless switching among four controlled model classes: MPNN, MPNN with chemistry/topology encoders, GPS-style hybrids of MPNN with global attention, and fully fused localglobal models with encoders. Using seven diverse open-source datasets for benchmarking across regression and classification tasks, we systematically isolate the contributions of message passing, global attention, and encoder-based feature augmentation. Our study shows that encoder-augmented MPNNs form a robust baseline, while fused localglobal models yield the clearest benefits for properties governed by long-range interaction effects. We further quantify the accuracycompute trade-offs of attention, reporting its overhead in memory. Together, these results establish the first controlled evaluation of global attention in atomistic graph learning and provide a reproducible testbed for future model development.

Equivariant graph neural networks↗

DOE OSTI2022

An LSTM Deep Learning Network for ¿Background Radiation Prediction

Determination of appropriate background radiation is important in any measurement application. Environmental radiation monitors and monitors used to assess dose to individuals outside of controlled areas are particularly susceptible to changes in readings due to fluctuations in the environmental conditions. These fluctuations (e.g. radon progeny concentrations) lead to changes in the detector response in the actual radiation environment, and they need to be taken into account when extracting the net operational doses. Work has been ongoing to apply advances in Deep Learning and Artificial Intelligence to account for changes in detector responses based on environmental parameters; in particular, a Long-Short Term Memory (LSTM) Deep Learning architecture has been utilized to incorporate time-series data into a prediction model. In this work, we present the current status of the project to predict radiation measurements based on meteorological conditions and air packet trajectories extracted using the National Oceanic and Atmospheric Administration's (NOAA) Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT4) model.

Degtiarenko, Pavel↗

DOE OSTI2024

Causality Extraction from Nuclear Licensee Event Reports Using a Hybrid Framework

In this paper, we proposed a hybrid framework for causality detection and extraction in LER documents. The main contributions are summarized below. We 1) built an LER corpus with 20,129 text samples for causality analysis, 2) developed an interactive tool was developed for labeling cause-effect pairs, 3) built a deep learning-based approach for causal relation detection, and 4) built a knowledge-based cause-effect extraction approach.

99 GENERAL AND MISCELLANEOUS↗

NASA NTRSDate not supplied

Transient Optimization of an Electrified Gas Turbine Engine Using Machine Learning

Gas turbine engines are designed with sufficient margin to prevent stall under normal operating conditions throughout their life. This compromise ensures that during rapid accelerations, compressor operation remains stable, but at the cost of efficiency and thrust responsiveness. The design margin encompasses multiple sources of uncertainty and systematic deviances from the operating line, the largest of which is the transient allowance. This set-aside accounts for the temporary incoordination of the engine spools during an acceleration while still enabling it to meet the certification requirement to accelerate from low to high power within a specified time, and without experiencing overtemperature, surge, stall, or other detrimental factors. Electrification of the powertrain provides the opportunity to address this reserve and truly optimize the design. The addition of electric machines inherent in hybrid propulsion concepts offers a means to interact with the engine shafts such that the necessary margin can be reduced, which can positively impact the engine design. By adjusting the amount of power extracted from or injected to the engine spools by the electric machines during transient operation, excursions from the operating line can be minimized. Past work using a dynamic engine model has shown that optimization of the fuel flow schedule during acceleration can reduce the required margin while still meeting the time requirement, and results are further improved when combined with power injection and extraction. The current work uses machine learning through a genetic algorithm to address the problem holistically by concurrently optimizing the electric machine power command and fuel flow acceleration schedule using an updated, higher fidelity version of the original engine model.

Stall Margin↗

NASA NTRSDate not supplied

Transient Optimization of an Electrified Gas Turbine Engine Using Machine Learning

Gas turbine engines are designed with sufficient margin to prevent stall under normal operating conditions throughout their life. This compromise ensures that during rapid accelerations, compressor operation remains stable, but at the cost of efficiency and thrust responsiveness. The design margin encompasses multiple sources of uncertainty and systematic deviances from the operating line, the largest of which is the transient allowance. This set-aside accounts for the temporary incoordination of the engine spools during an acceleration while still enabling it to meet the certification requirement to accelerate from low to high power within a specified time, and without experiencing overtemperature, surge, stall, or other detrimental factors. Electrification of the powertrain provides the opportunity to address this reserve and truly optimize the design. The addition of electric machines inherent in hybrid propulsion concepts offers a means to interact with the engine shafts such that the necessary margin can be reduced, which can positively impact the engine design. By adjusting the amount of power extracted from or injected to the engine spools by the electric machines during transient operation, excursions from the operating line can be minimized. Past work using a dynamic engine model has shown that optimization of the fuel flow schedule during acceleration can reduce the required margin while still meeting the time requirement, and results are further improved when combined with power injection and extraction. The current work uses machine learning through a genetic algorithm to address the problem holistically by concurrently optimizing the electric machine power command and fuel flow acceleration schedule using an updated, higher fidelity version of the original engine model.

Stall Margin↗

NASA NTRS2024

Transient Optimization of an Electrified Gas Turbine Engine Using Machine Learning

Gas turbine engines are designed with sufficient margin to prevent stall under normal operating conditions throughout their life. This compromise ensures that during rapid accelerations, compressor operation remains stable, but at the cost of efficiency and thrust responsiveness. The design margin encompasses multiple sources of uncertainty and systematic deviances from the operating line, the largest of which is the transient allowance. This set-aside accounts for the temporary incoordination of the engine spools during an acceleration while still enabling it to meet the certification requirement to accelerate from low to high power within a specified time, and without experiencing overtemperature, surge, stall, or other detrimental factors. Electrification of the powertrain provides the opportunity to address this reserve and truly optimize the design. The addition of electric machines inherent in hybrid propulsion concepts offers a means to interact with the engine shafts such that the necessary margin can be reduced, which can positively impact the engine design. By adjusting the amount of power extracted from or injected to the engine spools by the electric machines during transient operation, excursions from the operating line can be minimized. Past work using a dynamic engine model has shown that optimization of the fuel flow schedule during acceleration can reduce the required margin while still meeting the time requirement, and results are further improved when combined with power injection and extraction. The current work uses machine learning through a genetic algorithm to address the problem holistically by concurrently optimizing the electric machine power command and fuel flow acceleration schedule using an updated, higher fidelity version of the original engine model.

Stall Margin↗

DOE OSTI2022

Evolution of Programmatic Asset Lifecycle Planning at MESA

Early on in 2018 Sandia recognized the Microsystems Engineering, Science and Applications (MESA) Programmatic Asset Lifecycle Planning capability to be unpredictable, inconsistent, reactive, and unable to provide strong linkage to the sponsor's needs. The impetus for this report is to share learnings from MESA's journey towards maturing this capability. This report describes re-building the foundational elements of MESA's Programmatic Asset Lifecycle Planning capability using a risk-based, Multi-Criteria Decision Analysis (MCDA) approach. To begin, MESA's decades-old Piano Chart + Ad Hoc Hybrid Methodology is described with a narrative of its strengths and weaknesses. Then its replacement, the MCDA /Analytical Hierarchy Process, is introduced with a discussion of its strengths and weaknesses. To generate a realistic Programmatic Asset Lifecycle Planning budget outlook, MESA used its rolling 20-year Extended Life Program Plan (MELPP) as a baseline. The new MCDA risk-based prioritization methodology implements DOE/NNSA guidelines for prioritization of DOE activities and provides a reliable, structured framework for combining expert judgement and stakeholder preferences according to an established scientific technique. An in-house Hybrid Decision Support System (HDSS) software application was developed to facilitate production of several key deliverables. The application enables analysis of the prioritization decisions with charts to display and provide linkage of MESA's funding requests to the stakeholders' priorities, strategic objectives, nuclear deterrence programs, MESA priorities, and much more.

29 ENERGY PLANNING, POLICY, AND ECONOMY↗