Search NASA⌕ Search

SEARCH · Search NASA

Results for “Distributed Learning”

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 127 records · Page 7
All sourcesNASA NTRSDOE OSTI
DOE OSTI2021

Watch and learn—a generalized approach for transferrable learning in deep neural networks via physical principles

Transfer learning refers to the use of knowledge gained while solving a machine learning task and applying it to the solution of a closely related problem. Such an approach has enabled scientific breakthroughs in computer vision and natural language processing where the weights learned in state-of-the-art models can be used to initialize models for other tasks which dramatically improve their performance and save computational time. Here we demonstrate an unsupervised learning approach augmented with basic physical principles that achieves fully transferrable learning for problems in statistical physics across different physical regimes. By coupling a sequence model based on a recurrent neural network to an extensive deep neural network, we are able to learn the equilibrium probability distributions and inter-particle interaction models of classical statistical mechanical systems. Our approach, distribution-consistent learning, DCL, is a general strategy that works for a variety of canonical statistical mechanical models (Ising and Potts) as well as disordered interaction potentials. Using data collected from a single set of observation conditions, DCL successfully extrapolates across all temperatures, thermodynamic phases, and can be applied to different length-scales. This constitutes a fully transferrable physics-based learning in a generalizable approach.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2025

Dynamical Sketching for Enhanced Communication Efficiency in Federated Learning

Federated learning (FL) has revolutionized distributed machine learning by enabling collaborative model training without sharing local data. However, communication efficiency and privacy guarantees remain significant challenges. This paper introduces a dynamic sketching mechanism in FL, optimizing the trade-off between communication efficiency and model accuracy. By dynamically selecting the sketch matrix size, our approach adapts to the evolving characteristics of the data and the model, ensuring optimal performance across diverse scenarios. We leverage Bayesian optimization to systematically tune the sketch parameters, achieving an effective balance between resource efficiency and model performance. Experimental results on the MNIST dataset using a convolutional neural network (CNN) architecture validate the proposed method's efficiency and scalability. Our dynamic sketching approach significantly outperforms fixed-size sketching techniques, achieving higher compression ratios (up to 62x) and providing better privacy guarantees while maintaining high model accuracy. These findings highlight the robustness and versatility of our approach and make it a valuable solution for privacy-preserving, communication-efficient federated learning.

Afrose, Sharmin [ORNL]↗
DOE OSTI2024

MDLoader: A Hybrid Model-Driven Data Loader for Distributed Graph Neural Network Training

Scalable data management is essential for processing large scientific dataset on HPC platforms for distributed deep learning. In-memory distributed storage is preferred for its speed, enabling rapid, random, and frequent data access required by stochastic optimizers. Processes use one-sided or collective communication to fetch remote data, with optimal performance depending on (i) dataset characteristics, (ii) training scale, and (iii) interconnection network. Empirical analysis shows collective communication excels with larger mini-batch sizes and/or fewer processes, whereas one-sided communication outperforms at larger scales. We propose MDLoader, a hybrid in-memory data loader for distributed graph neural network training. MDLoader features a model-driven performance estimator that dynamically selects between one-sided and collective communication at the beginning of training using Tree of Parzen Estimators (TPE). Evaluations on NERSC Perlmutter and OLCF Summit show MDLoader outperforms single-backend loaders by up to 2.83 × and predicts the suitable communication method with 96.3% (Perlmutter) and 94.3% (Summit) success rate.

Bae, Jonghyun↗
DOE OSTI2025

FedOSAA: Improving Federated Learning with One-Step Anderson Acceleration

Federated learning (FL) is a distributed machine learning approach that enables multiple local clients and a central server to collaboratively train a model while keeping the data on their own devices. First-order methods, particularly those incorporating variance reduction techniques, are the most widely used FL algorithms due to their simple implementation and stable performance. However, these methods tend to be slow and require a large number of communication rounds to reach the global minimizer. We propose FedOSAA, a novel approach that preserves the simplicity of first-order methods while achieving the rapid convergence typically associated with second-order methods. Our approach applies one Anderson acceleration (AA) step following classical local updates based on first-order methods with variance reduction, such as FedSVRG and SCAFFOLD, during local training. This AA step is able to leverage curvature information from the history points and gives a new update that approximates the Newton-GMRES direction, thereby significantly improving the convergence. We establish a local linear convergence rate to the global minimizer of FedOSAA for smooth and strongly convex loss functions. Numerical comparisons show that FedOSAA substantially improves the communication and computation efficiency of the original first-order methods, achieving performance comparable to second-order methods like GIANT.

Feng, Xue [University of California, Davis]↗
DOE OSTI2023

Risk-Informed Condition Evaluation of Solar-centered Energy Generation and Distribution Networks through Bayesian Learning and Inference

We develop a methodology based on Bayesian inference over Probabilistic Graphical Models (PGMs) to understand and quantify risk in solar-centered grids using targeted measurements and learned system behavior. Being non-prescriptive but, rather, able to infer system behavior and, ultimately, address risk queries from data, our machine learning-type paradigm is tailored for diverse topologies and threat scenarios often associated with distributed energy generation and photovoltaic distributed energy resources (PV-DERs) in particular. We describe algorithmic processes for: (i) learning the structure of PGMs that result from attack-prone PV-DER-proliferated distribution systems, (ii) quantifying cause-effect relationships, and (iii) evaluating risk queries based on diverse evidence. The contributions are illustrated on a residential grid subject to output impairment attacks on its PV-DER infrastructure.

24 POWER TRANSMISSION AND DISTRIBUTION↗
DOE OSTI2021

Restoring Critical Loads In Resilient Distribution Systems using A Curriculum Learned Controller

In this paper, we propose a curriculum learned reinforcement learning (RL) controller to facilitate distribution system critical load restoration (CLR), leveraging RL's fast online response and its outstanding optimal sequential control capability. Like many grid control problems, CLR is complicated due to the large control action space and renewable uncertainty in a heavily constrained non-linear environment with strong intertemporal dependency. The nature of the problem oftentimes causes the RL policy to converge to a poor-performing local optimum if learned directly. To overcome this, we design a two-stage curriculum in which the RL agent will learn generation control and load restoration decision under different scenarios progressively. Via curriculum learning, the trained RL controller is expected to achieve a better control performance, with critical loads restored as rapidly and reliably as possible. Using the IEEE 13-bus test system, we illustrate the performance of the RL controller trained by the proposed curriculum-based method.

curriculum learning↗
DOE OSTI2020

Machine Learning-Based Prediction of Distribution Network Voltage and Sensor Allocation

Increasing penetration levels of fast-varying energy resources might negatively affect power system operation. At the same time, sensor deployment throughout distribution networks improves system awareness and enables the development of new and advanced voltage control solutions. Such control techniques rely on accurate prediction in anticipation of voltage violation scenarios. This paper analyzes various approaches to voltage prediction in a distribution system, and it is shown that combining multiple techniques into a single regressor improves its predictive power. Moreover, a two-step regressor is proposed in which initial predictions based on a global regressor are refined by local regressors; in this case, prediction errors decrease significantly. Additionally, a clustering approach is employed to perform sensor allocation so that only the most influential buses are selected for monitoring without diminishing prediction accuracy.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2020

Machine Learning-Based Prediction of Distribution Network Voltage and Sensors Allocation: Preprint

Increasing penetration of fast-varying energy resources may negatively affect power systems' operation. At the same time, sensor deployment throughout distribution networks improves system awareness and enables the development of new and advanced voltage control solutions. Such control techniques rely on accurate prediction in anticipation to voltage violation scenarios. This paper analyzes various approaches for voltage prediction in a distribution system; it is shown that combining multiple techniques into a single regressor improves its predictive power. Moreover, a two-step regressor is proposed, where initial predictions based on a global regressor are refined by local regressors; in this case, prediction errors decrease significantly. Additionally, a clustering approach is employed for performing sensor allocation, so that only the most influential buses are selected for monitoring without diminishing prediction accuracy.

24 POWER TRANSMISSION AND DISTRIBUTION↗
DOE OSTI2020

Machine Learning-Based Prediction of Distribution Network Voltage and Sensor Allocation

Increasing penetration of fast-varying energy resources may negatively affect power systems' operation. At the same time, sensor deployment throughout distribution networks improves system awareness and enables the development of new and advanced voltage control solutions. Such control techniques rely on accurate prediction in anticipation to voltage violation scenarios. This paper analyzes various approaches for voltage prediction in a distribution system; it is shown that combining multiple techniques into a single regressor improves its predictive power. Moreover, a two-step regressor is proposed, where initial predictions based on a global regressor are refined by local regressors; in this case, prediction errors decrease significantly. Additionally, a clustering approach is employed for performing sensor allocation, so that only the most influential buses are selected for monitoring without diminishing prediction accuracy.

61 RADIATION PROTECTION AND DOSIMETRY↗
DOE OSTI2021

Restoring Critical Loads in Resilient Distribution Systems Using a Curriculum Learned Controller: Preprint

In this paper, we propose a curriculum learned reinforcement learning (RL) controller to facilitate distribution system critical load restoration (CLR), leveraging RL's fast online response and its outstanding optimal sequential control capability. Like many grid control problems, CLR is complicated due to the large control action space and renewable uncertainty in a heavily constrained non-linear environment with strong intertemporal dependency. The nature of the problem oftentimes causes the RL policy to converge to a poor-performing local optimum if learned directly. To overcome this, we design a two-stage curriculum in which the RL agent will learn generation control and load restoration decision under different scenarios progressively. Via curriculum learning, the trained RL controller is expected to achieve a better control performance, with critical loads restored as rapidly and reliably as possible. Using the IEEE 13-bus test system, we illustrate the performance of the RL controller trained by the proposed curriculum-based method.

61 RADIATION PROTECTION AND DOSIMETRY↗
DOE OSTI2022

A Q-Learning-Based Approach for Distributed Beam Scheduling in mmWave Networks

We consider the problem of distributed downlink beam scheduling and power allocation for millimeter-Wave (mmWave) cellular networks where multiple base stations (BSs) belonging to different service operators share the same unlicensed spectrum with no centralized control or explicit coordination among them. Our goal is to design efficient distributed beam scheduling and power allocation schemes such that the network-level payoff, defined as the weighted sum of the total throughput and a power penalization term, can be maximized. To this end, we propose a distributed scheduling approach to power allocation and adaptation for efficient interference management over the shared spectrum by modeling each BS as an independent Q-learning agent. As a baseline, we compare the proposed approach to the state-of-the-art non-cooperative game-based approach which was previously developed for the same problem. We conduct extensive experiments under various scenarios to verify the effect of multiple factors on the performance of both approaches. Experiment results show that the proposed approach adapts well to different interference conditions by learning from experience and thus can achieve higher payoff than the game-based approach. The proposed approach can also be integrated into our previously developed Lyapunov stochastic optimization framework for the purpose of network utility maximization with optimality guarantee. As a result, the weights in the payoff function can be automatically and optimally determined by the virtual queue values from the sub-problems derived from the Lyapunov optimization framework.

99 GENERAL AND MISCELLANEOUS↗
DOE OSTI2021

Learning-Accelerated ADMM for Distributed DC Optimal Power Flow

We propose a novel data-driven method to accelerate the convergence of Alternating Direction Method of Multipliers (ADMM) for solving distributed DC optimal power flow (DC-OPF) where lines are shared between independent network partitions. Using previous observations of ADMM trajectories for a given system under varying load, the method trains a recurrent neural network (RNN) to predict the converged values of dual and consensus variables. Given a new realization of system load, a small number of initial ADMM iterations is taken as input to infer the converged values and directly inject them into the iteration. We empirically demonstrate that the online injection of these values into the ADMM iteration accelerates convergence by a significant factor for partitioned 14-, 118- and 2848-bus test systems under differing load scenarios. The proposed method has several advantages: it maintains the security of private decision variables inherent in consensus ADMM; inference is fast and so may be used in online settings; RNN-generated predictions can dramatically improve time to convergence but, by construction, can never result in infeasible ADMM subproblems; it can be easily integrated into existing software implementations. While we focus on the ADMM formulation of distributed DC-OPF in this paper, the ideas presented are naturally extended to other distributed optimization problems.

alternating direction method of multipliers↗
NASA NTRS2002

Lessons Learned In Developing Multiple Distributed Planning Systems for the International Space Station

The planning processes for the International Space Station (ISS) Program are quite complex. Detailed mission planning for ISS on-orbit operations is a distributed function. Pieces of the on-orbit plan are developed by multiple planning organizations, located around the world, based on their respective expertise and responsibilities. The "pieces" are then integrated to yield the final detailed plan that will be executed onboard the ISS. Previous space programs have not distributed the planning and scheduling functions to this extent. Major ISS planning organizations are currently located in the United States (at both the NASA Johnson Space Center (JSC) and NASA Marshall Space Flight Center (MSFC)), in Russia, in Europe, and in Japan. Software systems have been developed by each of these planning organizations to support their assigned planning and scheduling functions. Although there is some cooperative development and sharing of key software components, each planning system has been tailored to meet the unique requirements and operational environment of the facility in which it operates. However, all the systems must operate in a coordinated fashion in order to effectively and efficiently produce a single integrated plan of ISS operations, in accordance with the established planning processes. This paper addresses lessons learned during the development of these multiple distributed planning systems, from the perspective of the developer of one of the software systems. The lessons focus on the coordination required to allow the multiple systems to operate together, rather than on the problems associated with the development of any particular system. Included in the paper is a discussion of typical problems faced during the development and coordination process, such as incompatible development schedules, difficulties in defining system interfaces, technical coordination and funding for shared tools, continually evolving planning concepts/requirements, programmatic and budget issues, and external influences. Techniques that mitigated some of these problems will also be addressed, along with recommendations for any future programs involving the development of multiple planning and scheduling systems. Many of these lessons learned are not unique to the area of planning and scheduling systems, so may be applied to other distributed ground systems that must operate in concert to successfully support space mission operations.

Maxwell, Theresa G.↗
NASA NTRS2002

Lessons Learned in Developing Multiple Distributed Planning Systems for the International Space Station

The planning processes for the International Space Station (ISS) Program are quite complex. Detailed mission planning for ISS on-orbit operations is a distributed function. Pieces of the on-orbit plan are developed by multiple planning organizations, located around the world, based on their respective expertise and responsibilities. The pieces are then integrated to yield the final detailed plan that will be executed onboard the ISS. Previous space programs have not distributed the planning and scheduling functions to this extent. Major ISS planning organizations are currently located in the United States (at both the NASA Johnson Space Center (JSC) and NASA Marshall Space Flight Center (MSFC)), in Russia, in Europe, and in Japan. Software systems have been developed by each of these planning organizations to support their assigned planning and scheduling functions. Although there is some cooperative development and sharing of key software components, each planning system has been tailored to meet the unique requirements and operational environment of the facility in which it operates. However, all the systems must operate in a coordinated fashion in order to effectively and efficiently produce a single integrated plan of ISS operations, in accordance with the established planning processes. This paper addresses lessons learned during the development of these multiple distributed planning systems, from the perspective of the developer of one of the software systems. The lessons focus on the coordination required to allow the multiple systems to operate together, rather than on the problems associated with the development of any particular system. Included in the paper is a discussion of typical problems faced during the development and coordination process, such as incompatible development schedules, difficulties in defining system interfaces, technical coordination and funding for shared tools, continually evolving planning concepts/requirements, programmatic and budget issues, and external influences. Techniques that mitigated some of these problems will also be addressed, along with recommendations for any future programs involving the development of multiple planning and scheduling systems. Many of these lessons learned are not unique to the area of planning and scheduling systems, so may be applied to other distributed ground systems that must operate in concert to successfully support space mission operations.

Maxwell, Theresa G.↗
DOE OSTI2021

K_Road (KRoad)

KRoad is a 2-dimensional, lightweight, vehicle driver simulator that uses 3-dimensional vehicle driving model considerations, i.e. state-of-the-art friction models, and enables the driver to make control decisions with less computation time and programmer effort required than is with full-fidelity 3d simulators. KRoad is fully compatible with OpenAI gym, and contains a factored gym framework for composing gyms from various components (process, reward function, observation function, termination conditions, etc.). KRoad is additionally compatible with the RLlib distributed reinforcement learning environment, running under the Ray distributed python framework.

Tripp, CharlesEdison [National Renewable Energy La↗
DOE OSTI2022

On the Verification of Deep Reinforcement Learning Solution for Intelligent Operation of Distribution Grids

Capabilities of deep reinforcement learning (DRL) in obtaining fast decision policies in high dimensional and stochastic environments have led to its extensive use in operational research, including the operation of distribution grids with high penetration of distributed energy resources (DER). However, the feasibility and robustness of DRL solutions are not guaranteed for the system operator, and hence, those solutions may be of limited practical value. This paper proposes an analytical method to find feasibility ellipsoids that represent the range of multi-dimensional system states in which the DRL solution is guaranteed to be feasible. Empirical studies and stochastic sampling determine the ratio of the discovered to the actual feasible space as a function of the sample size. In addition, the performance of logarithmic, linear, and exponential penalization of infeasibility during the DRL training are studied and compared in order to reduce the number of infeasible solutions

Hosseini, Mohammad Mehdi↗
DOE OSTI2020

Learning-Accelerated ADMM for Distributed DC Optimal Power Flow

We suggest a novel data-driven method to accelerate the convergence of Alternating Direction Method of Multipliers (ADMM) for solving distributed DC optimal power flow (DC-OPF) where lines are shared between independent network partitions. Using previous observations of ADMM trajectories for a given system under varying load, the method trains a recurrent neural network (RNN) to predict the converged values of dual and consensus variables. Given a new realization of system load, a small number of initial ADMM iterations is taken as input to infer the converged values and directly inject them into the iteration. We empirically demonstrate that the online injection of these values into the ADMM iteration accelerates convergence by a significant factor for partitioned 14-, 118-and 2848-bus test systems under differing load scenarios. The proposed method has several advantages: it maintains the security of private decision variables inherent in consensus ADMM; inference is fast and so may be used in online settings; RNN-generated predictions can dramatically improve time to convergence but, by construction, can never result in infeasible ADMM subproblems; it can be easily integrated into existing software implementations. While we focus on the ADMM formulation of distributed DC-OPF in this paper, the ideas presented are naturally extended to other distributed optimization problems.

24 POWER TRANSMISSION AND DISTRIBUTION↗