Search NASA⌕ Search

SEARCH · Search NASA

Results for “machine learning optimization”

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 415 records · Page 23
All sourcesNASA NTRSDOE OSTI
DOE OSTI2025

When more data hurts: Optimizing data coverage while mitigating diversity-induced underfitting in an ultrafast machine-learned potential

Machine-learned interatomic potentials (MLIPs) are becoming an essential tool in materials modeling. However, optimizing the generation of training data used to parametrize the MLIPs remains a significant challenge. This is because MLIPs can fail when encountering local environments too different from those present in the training data. The difficulty of determining a priori the environments that will be encountered during molecular dynamics simulation necessitates diverse, high-quality training data. Here, this study investigates how training data diversity affects the performance of MLIPs using the Ultra-Fast force field (UF 3 ) to model amorphous silicon nitride. We employ expert and autonomously generated data to create the training data and fit four force field variants to subsets of the data. Our findings reveal a critical balance in training data diversity: insufficient diversity hinders generalization, while excessive diversity can exceed the MLIP's learning capacity, reducing simulation accuracy. Specifically, we found that the UF 3 variant trained on a subset of the training data, in which nitrogen-rich structures were removed, offered vastly better prediction and simulation accuracy than any other variant. By comparing these UF 3 variants, we highlight the nuanced requirements for creating accurate MLIPs, emphasizing the importance of application-specific training data to achieve optimal performance in modeling complex material behaviors.

ab initio molecular dynamics↗
DOE OSTI2024

Analytical ab initio hessian from a deep learning potential for transition state optimization

Identifying transition states—saddle points on the potential energy surface connecting reactant and product minima—is central to predicting kinetic barriers and understanding chemical reaction mechanisms. In this work, we train a fully differentiable equivariant neural network potential, NewtonNet, on thousands of organic reactions and derive the analytical Hessians. By reducing the computational cost by several orders of magnitude relative to the density functional theory (DFT) ab initio source, we can afford to use the learned Hessians at every step for the saddle point optimizations. We show that the full machine learned (ML) Hessian robustly finds the transition states of 240 unseen organic reactions, even when the quality of the initial guess structures are degraded, while reducing the number of optimization steps to convergence by 2–3× compared to the quasi-Newton DFT and ML methods. All data generation, NewtonNet model, and ML transition state finding methods are available in an automated workflow.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2020

Automated detector simulation and reconstruction parametrization using machine learning

Rapidly applying the effects of detector response to physics objects (e.g. electrons, muons, showers of particles) is essential in high energy physics. Presently available tools for the transformation from truth-level physics objects to reconstructed detector-level physics objects involve manually defining resolution functions. These resolution functions are typically derived in bins of variables that are correlated with the resolution (e.g. pseudorapidity and transverse momentum). This process is time consuming, requires manual updates when detector conditions change, and can miss important correlations. Machine learning offers a way to automate the process of building these truth-to-reconstructed object transformations and can capture complex correlation for any given set of input variables. Such machine learning algorithms, with sufficient optimization, could have a wide range of applications: improving phenomenological studies by using a better detector representation, allowing for more efficient production of Geant4 simulation by only simulating events within an interesting part of phase space, and studies on future experimental sensitivity to new physics.

47 OTHER INSTRUMENTATION↗
DOE OSTI2026

Surrogate modeling of Monte Carlo radiation transport with convolutional neural networks for shielding optimization

Here, we present a machine learning (ML)-based surrogate model using convolutional neural networks (CNN) designed to emulate the attenuation of neutron fields as they pass through various shielding materials. This model can compute the outgoing neutron flux almost instantaneously and achieves reasonable accuracy compared to traditional Monte Carlo (MC)-based codes, which are computationally intensive. This emulator alleviates the complexity of neutron radiation transport through shielding materials by reducing the dimensionality and enables shielding optimization for a known radiation environment. This optimization process, which would have taken an unrealistic timeline due to several complex radiation transport simulations, can now be achieved in minutes, thus increasing computational capabilities in radiation shielding assessment. We demonstrate the applications of this emulator in computing effective dose rates and optimizing shielding solutions for a heavy-ion accelerator facility, such as the Facility for Rare Isotope Beams, where secondary neutrons produced via beam interactions dominate the radiation environment.

accelerator shielding↗
DOE OSTI2025

Randomized Federated Learning Methods for Nonsmooth, Nonconvex, and Hierarchical Optimization (Final Technical Report)

This final technical report summarizes the outcomes of a DOE-funded project on federated scientific machine learning (FL) under nonsmooth, nonconvex, and hierarchical optimization settings. The project develops new mathematical models, algorithms, and theoretical guarantees for decentralized stochastic, bilevel, and minimax optimization problems arising in DOE mission-relevant applications. A unified framework of randomized and zeroth-order federated optimization methods is introduced, providing provable convergence, communication efficiency, and sample-complexity guarantees. The report documents algorithmic design, theoretical analysis, and empirical validation of the proposed federated learning methods. The project also contributes to workforce development through graduate training and dissemination of results via publications and seminars.

97 MATHEMATICS AND COMPUTING↗
NASA NTRSDate not supplied

Machine Learning for Dynamic Test Sensor Placement

There are multiple different algorithms to perform modal test sensor placement optimization: effective independence, residual kinetic energy, iterative Guyan reduction, genetic algorithms, or a brute-force methodology. However, any of these methods may be computationally expensive, especially for structural models with a large number of degrees of freedom. Given the high-cost and the need to optimize the solution, modal sensor placement is a great application for machine learning (ML) algorithms. In this paper, we will apply ML algorithms to determine the optimal sensor locations for simple and complex structures. We will also discuss the benefits and drawbacks of using machine learning over other sensor placement algorithms.

Kelsey Buckles↗
DOE OSTI2022

Active meta-learning for predicting and selecting perovskite crystallization experiments

Autonomous experimentation systems use algorithms and data from prior experiments to select and perform new experiments in order to meet a specified objective. In most experimental chemistry situations, there is a limited set of prior historical data available, and acquiring new data may be expensive and time consuming, which places constraints on machine learning methods. Active learning methods prioritize new experiment selection by using machine learning model uncertainty and predicted outcomes. Meta-learning methods attempt to construct models that can learn quickly with a limited set of data for a new task. Here in this paper, we applied the model-agnostic meta-learning (MAML) model and the Probabilistic LATent model for Incorporating Priors and Uncertainty in few-Shot learning (PLATIPUS) approach, which extends MAML to active learning, to the problem of halide perovskite growth by inverse temperature crystallization. Using a dataset of 1870 reactions conducted using 19 different organoammonium lead iodide systems, we determined the optimal strategies for incorporating historical data into active and meta-learning models to predict reaction compositions that result in crystals. We then evaluated the best three algorithms (PLATIPUS and active-learning k-nearest neighbor and decision tree algorithms) with four new chemical systems in experimental laboratory tests. With a fixed budget of 20 experiments, PLATIPUS makes superior predictions of reaction outcomes compared to other active-learning algorithms and a random baseline.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2026

A Decomposition-Based Learn-To-Optimize Approach with Feasibility Layer Assistance for Sub-Hourly Unit Commitment

Sub-hourly unit commitment (UC) with 15-min intervals is gaining significant attention as a way to respond rapidly to the fluctuations in electricity supply and demand introduced by renewable resources. However, the increased temporal resolution and complex inter-temporal dependencies pose substantial computational challenges for traditional optimization methods. To this end, this paper explores a decomposition-based learn-to-optimize approach. Building on recent advances in machine learning, our method revisits the long- overlooked Lagrangian relaxation framework, which is a classical decomposition technique that enables tractable subproblem solving. These smaller subproblems are inherently well-suited for machine learning, as their reduced dimensionality and structural regularity allow predictive models to efficiently learn and generalize solution patterns. We thus propose a generic predictive model, which embeds Gated Recurrent Units (GRUs) and Attention in the encoder-decoder structure, and integrate a rule-based feasibility layer to capture temporal dependencies, reduce training effort, and improve feasibility w.r.t. unit-level constraints. Our method has been validated on the IEEE 118-bus system, demonstrating promising performance in solving sub-hourly UC problems efficiently and feasibly.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

Levelized cost of charging of extreme fast charging with stationary LMO/LTO batteries

Extreme DC fast charging for electric vehicles (EVs) could be competitive with the internal combustion engine refueling experience and enable longer-distance travel, which could help with EV adoption and decarbonization, but these systems have high capital costs and extremely variable high-power demands. Behind-the-meter systems (BTMS) could support extreme-fast-charging (XFC) stations to increase nationwide adoption of EVs. Here, this study examines the optimal break-even levelized cost of charging (LCOC) across 96 BTMS scenarios to enable low-wait XFC stations providing 200 miles of charge in 10 min. This research simulates LCOC via synthetic XFC-capable EV loads, machine-learned battery life models from testing data, and nonlinear optimal controls, co-minimizing complex utility costs and battery replacements. An aggregate optimal BTMS design treating each EV load as equal likely gives an optimal LCOC per utility rate, the average of which is $\$$0.59/kWh. In addition, the sensitivity of optimal and off-optimal design factors, the long-life LMO/LTO chemistry, and optimized controls are analyzed. The battery control model, based on battery stressors to compare chemistries, optimizes LMO/LTO resting state of charge and cycle depth without compromising cost reduction, which enables greater flexibility in operation. The LCOC savings due to replacement reduction are small, up to $\$$0.035/kWh (6%), with an average of $\$$0.02/kWh (3.5%). Compared with gasoline stations, the aggregate XFC station design achieves comparable speed, experience of service, and cost at $\$$3.81/gal gasoline, showing that EVs can replace gasoline vehicles even for longer-distance travel.

25 ENERGY STORAGE↗
DOE OSTI2020

Determining Multi-Component Phase Diagrams with Desired Characteristics Using Active Learning

Herein, we demonstrate how to predict and experimentally validate phase diagrams for multi-component systems from a high-dimensional virtual space of all possible phase diagrams involving several elements based on small existing experimental data. The experimental data for bulk phases for known systems represents a sampling from this space, and screening the space allows multi-component phase diagrams with given design criteria to be built. This approach uses machine learning methods to predict phase diagrams and Bayesian experimental design to minimize experiments for refinement and validation, all within an active learning loop. The approach is proven by predicting and synthesizing the ferroelectric ceramic system (1-ω)(Ba 0.61 Ca 0.28 Sr 0.11 TiO 3 )-ω(BaTi 0.888 Zr 0.0616 Sn 0.0028 Hf 0.0476 O 3 ) with a relatively high transition temperature and triple point, as well as the NiTi-based pseudo-binary phase diagram (1-ω)(Ti 0.309 Ni 0.485 Hf 0.20 Zr 0.006 )-ω(Ti 0.309 Ni 0.485 Hf 0.07 Zr 0.068 Nb 0.068 ) designed for high transition temperature (ω ≤ 1). Each phase diagram is validated and optimized through only three new experiments. The complexity of these compounds is beyond the reach of today’s computational methods.

36 MATERIALS SCIENCE↗
DOE OSTI2022

Robust Decentralized Learning Using ADMM With Unreliable Agents

Many signal processing and machine learning problems can be formulated as consensus optimization problems which can be solved efficiently via a cooperative multi-agent system. However, the agents in the system can be unreliable due to a variety of reasons: noise, faults and attacks. Providing erroneous updates leads the optimization process in a wrong direction, and degrades the performance of distributed machine learning algorithms. This paper considers the problem of decentralized learning using ADMM in the presence of unreliable agents. First, we rigorously analyze the effect of erroneous updates (in ADMM learning iterations) on the convergence behavior of the multi-agent system. We show that the algorithm linearly converges to a neighborhood of the optimal solution under certain conditions and characterize the neighborhood size analytically. Next, we provide guidelines for network design to achieve a faster convergence to the neighborhood. Here, we also provide conditions on the erroneous updates for exact convergence to the optimal solution. Finally, to mitigate the influence of unreliable agents, we propose ROAD , a robust variant of ADMM, and show its resilience to unreliable agents with an exact convergence to the optimum.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2025

Exploring scenarios for enhanced fuel compression and performance on the National Ignition Facility with machine-learning-aided design techniques

Recent fusion experiments on the National Ignition Facility (NIF) have achieved ignition, producing multi-MJ fusion yields for input laser energies of roughly 2 MJ [Abu-Shawareb et al., Phys. Rev. Lett. 132, 065102 (2024)]. Building on the success of the target designs that have achieved ignition, we explore new implosion scenarios predicted to generate significantly more compression of the dense DT ice layer and correspondingly higher yields while preserving many of the key physics characteristics of present-day ignition designs. Our main result is a novel 3-shock implosion scheme that effectively minimizes the shock-induced entropy in the dense, accelerating DT shell and maximizes the resulting fuel compression subject to a fixed leading shock strength consistent with present-day ignition experiments, which is necessary to melt the crystalline high-density carbon ablator. Compared to the first NIF experiment to fulfill Lawson's ignition criterion, shot N210808 [Abu-Shawareb et al., Phys. Rev. Lett. 129, 075001 (2022)], our design exhibits a 40% increase in simulated peak areal density (ρR) and a 5× increase in 1D fusion yield using a 4% lighter ablator and identical DT payloads. We also present a complete integrated 2D hohlraum design and laser pulse specifications capable of generating the desired 3-shock drive and maintaining control of the low-mode capsule implosion symmetry, where the increase in simulated 2D yield relative to N210808 is > 10×. This new implosion regime was discovered with help from a machine-learning-enabled capsule design optimization framework. We outline the workflow this automated tool uses to identify improved design candidates by running several rounds of capsule simulations, constructing a surrogate model mapping input variations to key physics output quantities, and querying the resulting statistical model to propose adjustments to the x-ray drive and capsule to reach a set of physics objectives prescribed by the designer.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2025

Higher-order factorization machine for accurate surrogate modeling in material design

Efficient and robust optimization is important in material science for identifying optimal structural parameters and enhancing material performance. Surrogate-based active learning algorithms have recently gained great attention for their ability to efficiently navigate large, high-dimensional design spaces. Among surrogate models, 2 nd -order factorization machine (FM) models are widely employed as the surrogate model in active learning algorithms due to their balance between simplicity and effectiveness. However, their quadratic nature limits their capacity to capture complex, higher-order interactions among variables, often leading to suboptimal solutions. To overcome this limitation, we propose an active learning scheme integrating a 3 rd -order FM model, capable of modeling three-variable interactions and more intricate relationships in material systems. We comprehensively evaluate the surrogate modeling performance of the 3 rd -order FM case using various objective functions. Furthermore, we examine the optimization reliability and efficiency of the 3 rd -order FM-based active learning in a real-world material design task (e.g., nanophotonic structures for transparent radiative cooling). Our study shows that the 3 rd -order FM outperforms the 2 nd -order model in both surrogate accuracy and optimization performance, highlighting higher-order models’ promises for material design and optimization problems.

Factorization machine↗
DOE OSTI2024

Improving ADAM through an implicit-explicit (IMEX) time-stepping approach

The ADAM optimizer, often used in machine learning for neural network training, corresponds to an underlying ordinary differential equation (ODE) in the limit of very small learning rates. Here, this work shows that the classical ADAM algorithm is a first-order implicit-explicit (IMEX) Euler discretization of the underlying ODE. Employing the time discretization point of view, we propose new extensions of the ADAM scheme obtained by using higher-order IMEX methods to solve the ODE. Based on this approach, we derive a new optimization algorithm for neural network training that performs better than classical ADAM on several regression and classification problems.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2021

Data-Driven Strategies for Accelerated Materials Design

The ongoing revolution of the natural sciences by the advent of machine learning and artificial intelligence sparked significant interest in the material science community in recent years. The intrinsically high dimensionality of the space of realizable materials makes traditional approaches ineffective for large-scale explorations. Modern data science and machine learning tools developed for increasingly complicated problems are an attractive alternative. An imminent climate catastrophe calls for a clean energy transformation by overhauling current technologies within only several years of possible action available. Tackling this crisis requires the development of new materials at an unprecedented pace and scale. For example, organic photovoltaics have the potential to replace existing silicon-based materials to a large extent and open up new fields of application. In recent years, organic light-emitting diodes have emerged as state-of-the-art technology for digital screens and portable devices and are enabling new applications with flexible displays. Reticular frameworks allow the atom-precise synthesis of nanomaterials and promise to revolutionize the field by the potential to realize multifunctional nanoparticles with applications from gas storage, gas separation, and electrochemical energy storage to nanomedicine. In the recent decade, significant advances in all these fields have been facilitated by the comprehensive application of simulation and machine learning for property prediction, property optimization, and chemical space exploration enabled by considerable advances in computing power and algorithmic efficiency. In this Account, we review the most recent contributions of our group in this thriving field of machine learning for material science. We start with a summary of the most important material classes our group has been involved in, focusing on small molecules as organic electronic materials and crystalline materials. Specifically, we highlight the data-driven approaches we employed to speed up discovery and derive material design strategies. Subsequently, our focus lies on the data-driven methodologies our group has developed and employed, elaborating on high-throughput virtual screening, inverse molecular design, Bayesian optimization, and supervised learning. We discuss the general ideas, their working principles, and their use cases with examples of successful implementations in data-driven material discovery and design efforts. Furthermore, we elaborate on potential pitfalls and remaining challenges of these methods. Finally, we provide a brief outlook for the field as we foresee increasing adaptation and implementation of large scale data-driven approaches in material discovery and design campaigns.

36 MATERIALS SCIENCE↗
DOE OSTI2020

Data optimization for large batch distributed training of deep neural networks

Distributed training in deep learning (DL) is common practice as data and models grow. The current practice for distributed training of deep neural networks faces the challenges of communication bottlenecks when operating at scale, and model accuracy deterioration with an increase in global batch size. Present solutions focus on improving message exchange efficiency as well as implementing techniques to tweak batch sizes and models in the training process. The loss of training accuracy typically happens because the loss function gets trapped in a local minima. We observe that the loss landscape minimization is shaped by both the model and training data and propose a data optimization approach that utilizes machine learning to implicitly smooth out the loss landscape resulting in fewer local minima. Our approach filters out data points which are less important to feature learning, enabling us to speed up the training of models on larger batch sizes to improved accuracy.

Gahlot, Shubhankar↗
DOE OSTI2022

Uncertainty-aware mixed-variable machine learning for materials design

Abstract Data-driven design shows the promise of accelerating materials discovery but is challenging due to the prohibitive cost of searching the vast design space of chemistry, structure, and synthesis methods. Bayesian optimization (BO) employs uncertainty-aware machine learning models to select promising designs to evaluate, hence reducing the cost. However, BO with mixed numerical and categorical variables, which is of particular interest in materials design, has not been well studied. In this work, we survey frequentist and Bayesian approaches to uncertainty quantification of machine learning with mixed variables. We then conduct a systematic comparative study of their performances in BO using a popular representative model from each group, the random forest-based Lolo model (frequentist) and the latent variable Gaussian process model (Bayesian). We examine the efficacy of the two models in the optimization of mathematical functions, as well as properties of structural and functional materials, where we observe performance differences as related to problem dimensionality and complexity. By investigating the machine learning models’ predictive and uncertainty estimation capabilities, we provide interpretations of the observed performance differences. Our results provide practical guidance on choosing between frequentist and Bayesian uncertainty-aware machine learning models for mixed-variable BO in materials design.

36 MATERIALS SCIENCE↗