Search NASA⌕ Search

SEARCH · Search NASA

Results for “Distributed training”

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 163 records · Page 9

Artificial Intelligence Medical Support for Long-Duration Space Missions

We envision an artificial intelligence (AI) based system that will provide support and recommendations to the crew medical officer (CMO) and ground flight surgeon during long-duration space missions. Such a system would be pretrained on the knowledgebase of clinical knowledge on Earth, minimizing the amount of Earth data that needs to be transferred into space. Then during deployment, the system would be constantly refined through active learning from diverse streams of data from sensors in the spacecraft, data collected daily from individual astronauts, and human-in-the-loop feedback from the crew. The model could be interrogated for predictions and recommendations on personalized crew health based on the overall status of the spacecraft, medicinal stores, and status of other crew members. Adaptation techniques would be used to incorporate spaceflight data that have very different distributions from the training data due to the extreme environment. Edge computing and the most advanced neuromorphic processing would enable computation in scenarios with low power and bandwidth, while dimensionality reduction would be employed to ensure that the input data streams from spaceflight are as small as possible. In order to realize this long-term vision, several hardware and software aspects need to be developed and assembled. First, models pretrained on Earth biomedical data would need to be evaluated for predictive accuracy, and the best one selected. That model would need to be adapted to learn from diverse, sparse, and inconsistently measured data streams, as well as human-in-the-loop feedback. A data integration, standardization, and dimensionality reduction methodology would need to be developed to handle all data types and feed them into the model. Once the software and data infrastructure is developed, it would need to be integrated with small footprint compute processors and tested in high-radiation, high-vibration, unregulated temperature situations. As a short-term goal, we recommend to focus on the development of the data and model software structure. Several large language models (LLM) already exist that have been trained on Earth biomedical and clinical knowledgebases, including BioMedLLM, Med-PaLM, SPOKE LLM, and Foresight. These models need to be evaluated for accuracy and the best one chosen for a proof-of-concept structure, while maintaining awareness of the accelerating AI field and incorporating any newly improved model architectures as needed. Then, we recommend to develop a database of synthetic data types to mimic the diverse data streams that are expected in a long-duration space mission. This should include environmental and microbial data from the spacecraft, non-invasive data from wearables and point-of-care devices employed by astronauts, and more invasive molecular and physiological monitoring of clinical and biomarker data from astronauts. The data standardization methodology should be developed, and these data streams used to refine the clinical LLM. Several scenarios should be developed that could plausibly come up in a long-duration space mission, and changes or aberrations introduced to the data at specific times to mimic these scenarios. Then, question and answer tasks should be designed to interrogate the model for predictions and recommendations, with acceptable answers already identified.

Artificial Intelligence↗

Detecting Unclassified Electromagnetic Signals for Secure Wireless Communication Using Open Set Recognition

We developed multiple machine learning methods for the detection and classification of new wireless communication waveforms, which is critical for targeted attacks in wireless networks and electronic warfare. Our machine learning models are capable of dynamically detecting security threats in near real time through our advanced open set recognition (OSR) approach. This model has demonstrated significant improvements in the detection of unknown waveforms, thereby enhancing the security and reliability of mission critical communications. Our approach to detecting uncertain security threats is novel; we advanced OSR techniques by incorporating domain knowledge of wireless signals. Specifically, we combined time and frequency domain model features to enhance the model’s performance. Utilizing an OSR approach eliminates the need for training data to be distributed similarly to the deployment environment and removes the requirement for the training set to contains all possible threat classes. This is crucial because it is often infeasible to determine and characterize all potential security threats in advance. Our model were trained on simulated data, generated in partnership with the University at Albany, State of New York. The data set contained a diverse array of wireless signals, including those with additive white Gaussian noise and multipath signals, with and without line of sight. This comprehensive training set allowed us to optimize our models to detect unknown waveforms under various challenging scenarios, such as low signal-to-noise ratios. By training on various waveforms, varying signal-to-noise ratio, and different sample sizes under normal conditions, our models were fine tuned to perform effectively in challenging environments.

99 - GENERAL AND MISCELLANEOUS↗

Lion Cub: Minimizing Communication Overhead in Distributed Lion

Communication overhead is a key challenge in distributed deep learning, especially on slower Ethernet intercon nects, and given current hardware trends, communication is likely to become a major bottleneck. While gradient compression techniques have been explored for SGD and Adam, the Lion optimizer has the distinct advantage that its update vectors are the output of a sign operation, enabling straightforward quantization. However, simply compressing updates for communication and using techniques like majority voting fails to lead to end-to-end speedups due to inefficient communication algorithms and reduced convergence. We analyze three factors critical to distributed learning with Lion: optimizing communication methods, identifying effective quantization methods, and assessing the necessity of momentum synchronization. Our findings show that quantization techniques adapted to Lion and selective momentum synchronization can significantly reduce communication costs while maintaining convergence. We combine these into Lion Cub, which enables up to 5x speedups in end-to-end training compared to Lion. This highlights Lion’s potential as a communication-efficient solution for distributed training.

97 MATHEMATICS AND COMPUTING↗

pnnl/frequency_sensitivity

This research code base includes functions to compute statistics of image distributions in (spatial DFT) frequency space, train deep learning image classifiers on these distributions with variable depth and weight decay, and finally measure the sensitivity of the trained models to perturbations along (spatial DFT) frequency components.

Central, PNNL Developer↗

Back Propagation Neural Networks for Predicting Ultimate Strengths of Unidirectional Graphite/Epoxy Tensile Specimens

The research presented herein demonstrates the feasibility of predicting ultimate strengths in simple composite structures through a neural network analysis of their acoustic emission (AE) amplitude distribution data. A series of eleven ASTM D-3039 unidirectional graphite/epoxy tensile samples were loaded to failure to generate the amplitude distributions for this analysis. A back propagation neural network was trained to correlate the AE amplitude distribution signatures generated during the first 25% of loading with the ultimate strengths of the samples. The network was trained using two sets of inputs: (1) the statistical parameters obtained from a Weibull distribution fit of the amplitude distribution data, and (2) the event frequency (amplitude) distribution itself. The neural networks were able to predict ultimate strengths with a worst case error of -8.99% for the Weibull modeled amplitude distribution data and 3.74% when the amplitude distribution itself was used to train the network. The principal reason for the improved prediction capability of the latter technique lies in the ability of the neural network to extract subtle features from within the amplitude distribution.

Walker, James L.↗

Out-of-distribution generalization for learning quantum dynamics

Generalization bounds are a critical tool to assess the training data requirements of Quantum Machine Learning (QML). Recent work has established guarantees for in-distribution generalization of quantum neural networks (QNNs), where training and testing data are drawn from the same data distribution. However, there are currently no results on out-of-distribution generalization in QML, where we require a trained model to perform well even on data drawn from a different distribution to the training distribution. Here, we prove out-of-distribution generalization for the task of learning an unknown unitary. In particular, we show that one can learn the action of a unitary on entangled states having trained only product states. Since product states can be prepared using only single-qubit gates, this advances the prospects of learning quantum dynamics on near term quantum hardware, and further opens up new methods for both the classical and quantum compilation of quantum circuits.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Bayesian quantum state reconstruction with a learning-based tuned prior

We demonstrate machine-learning-enhanced Bayesian quantum state tomography on near-term intermediate-scale quantum hardware. Our approach to selecting prior distributions leverages pre-trained neural networks incorporating measurement data and enables improved inference times over standard prior distributions.

Regmi, Sangita↗

Know Your Space: Inlier and Outlier Construction for Calibrating Medical OOD Detectors

This software offers methods and functions for training calibrated out-of-distribution detectors for medical image classification tasks. It includes functionalities for training, synthesizing data augmentations, calibration, and out-of-distribution detection. Developed using PyTorch, this software is compatible with standard neural network architectures used for imaging data. Additionally, it provides capabilities to compute evaluation metrics for assessing the performance and quality of the detectors.

Narayanaswamy, VivekSivaraman↗

Out-of-distribution generalization for learning quantum dynamics

Abstract Generalization bounds are a critical tool to assess the training data requirements of Quantum Machine Learning (QML). Recent work has established guarantees for in-distribution generalization of quantum neural networks (QNNs), where training and testing data are drawn from the same data distribution. However, there are currently no results on out-of-distribution generalization in QML, where we require a trained model to perform well even on data drawn from a different distribution to the training distribution. Here, we prove out-of-distribution generalization for the task of learning an unknown unitary. In particular, we show that one can learn the action of a unitary on entangled states having trained only product states. Since product states can be prepared using only single-qubit gates, this advances the prospects of learning quantum dynamics on near term quantum hardware, and further opens up new methods for both the classical and quantum compilation of quantum circuits.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

A Neural Network Aero Design System for Advanced Turbo-Engines

An inverse design method calculates the blade shape that produces a prescribed input pressure distribution. By controlling this input pressure distribution the aerodynamic design objectives can easily be met. Because of the intrinsic relationship between pressure distribution and airfoil physical properties, a Neural Network can be trained to choose the optimal pressure distribution that would meet a set of physical requirements. Neural network systems have been attempted in the context of direct design methods. From properties ascribed to a set of blades the neural network is trained to infer the properties of an 'interpolated' blade shape. The problem is that, especially in transonic regimes where we deal with intrinsically non linear and ill posed problems, small perturbations of the blade shape can produce very large variations of the flow parameters. It is very unlikely that, under these circumstances, a neural network will be able to find the proper solution. The unique situation in the present method is that the neural network can be trained to extract the required input pressure distribution from a database of pressure distributions while the inverse method will still compute the exact blade shape that corresponds to this 'interpolated' input pressure distribution. In other words, the interpolation process is transferred to a smoother problem, namely, finding what pressure distribution would produce the required flow conditions and, once this is done, the inverse method will compute the exact solution for this problem. The use of neural network is, in this context, highly related to the use of proper optimization techniques. The optimization is used essentially as an automation procedure to force the input pressure distributions to achieve the required aero and structural design parameters. A multilayered feed forward network with back-propagation is used to train the system for pattern association and classification.

Sanz, Jose M.↗

Power Grid Contingency Analysis with Machine Learning: A Brief Survey and Prospects

We briefly review previous applications of machine learning (ML) in power grid analyses and introduce our ongoing effort toward developing a generative-adversarial (GA) model for fast and reliable grid contingency analyses. According to our review, the persisting limitation of traditional ML techniques in grid analyses is the need for an exhaustive amount of training data for model generalization and accurate predictions. GA models overcome this limitation by first learning true data distribution from a small training set, from which new samples assimilating true data are generated with some variations. Subsequently, GA models can transfer learn or super-generalize with increased accuracy, that is, accurately predict n - (k + 2) contingencies from a small n - k training set and generated n - (k + 1) data. The joint effort between Idaho National Lab and Florida State University strives to develop a zero-shot and deep learning-based contingency analysis tool, named Smart Contingency Analysis Neural Network (SCANN), by leveraging the aforementioned advantages of GA models. The basic architecture of SCANN stems from the Latent Encoding of Atypical Perturbations network combined with an adversarial network, and it is designed to generate imbalanced power flow data from learned true data distributions for prediction purposes. Here we also introduce the abstract concept of resilience-chaos plots, a new resilience characterization tool proposed to complement SCANN by aiding in the assessment of large amounts of high-order contingency predictions.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Decentralized Distributed Proximal Policy Optimization (DD-PPO) for High Performance Computing Scheduling on Multi-User Systems

Resource allocation in High Performance Computing (HPC) environments presents a complex and multifaceted challenge for job scheduling algorithms. Beyond the efficient allocation of system resources, schedulers must account for and optimize multiple performance metrics, including job wait time and system throughput. Traditional heuristic-based scheduling algorithms increasingly struggle and lack the efficiency needed to meet the demands and address the complexity and scale of modern HPC systems. Consequently, recent research efforts have focused on leveraging advancements in Artificial Intelligence (AI) and Deep Learning (DL), particularly Reinforcement Learning (RL), to develop more adaptable and intelligent scheduling strategies. Previous RL-based scheduling approaches have explored a range of algorithms, from Deep Q-Networks (DQN) to Proximal Policy Optimization (PPO), and more recently, hybrid methods that integrate Graph Neural Networks (GNNs) with RL techniques. However, a common limitation across these methods is their reliance on relatively small datasets, with few methods being evaluated using large-scale, multi-million-job trace datasets representative of real-world HPC workloads. Moreover, existing RL schedulers face scalability issues due to centralized policy updates, which hinder training efficiency and performance when applied to large datasets. This study introduces a novel RL-based scheduler utilizing Decentralized Distributed Proximal Policy Optimization (DD-PPO) algorithm, which supports large-scale distributed training across multiple workers without requiring parameter synchronization at every step. By eliminating reliance on centralized updates to a shared policy, the DD-PPO scheduler enhances scalability, training efficiency, and sample utilization. Experimental validation using a large real-world dataset containing over 11.5 million job traces collected from petascale HPC systems over six years assesses the influence of dataset scale on training effectiveness and compares DD-PPO performance to traditional and advanced scheduling approaches. The experimental results demonstrate improved scheduling performance in comparison to both heuristic-based schedulers and existing RL-based scheduling algorithms.

AI↗

Iterative self-organizing SCEne-LEvel sampling (ISOSCELES) for large-scale building extraction

Convolutional neural networks (CNN) provide state-of-the-art performance in many computer vision tasks, including those related to remote-sensing image analysis. Successfully training a CNN to generalize well to unseen data, however, requires training on samples that represent the full distribution of variation of both the target classes and their surrounding contexts. With remote sensing data, acquiring a sufficiently representative training set is a challenge due to both the inherent multi-modal variability of satellite or aerial imagery and the general high cost of labeling data. To address this challenge, we have developed ISOSCELES, an Iterative Self-Organizing SCEne LEvel Sampling method for hierarchical sampling of large image sets. Using affinity propagation, ISOSCELES automates the selection of highly representative training images. Compared to random sampling or using available reference data, the distribution of the training is principally data driven, reducing the chance of oversampling uninformative areas or undersampling informative ones. In comparison to manual sample selection by an analyst, ISOSCELES exploits descriptive features, spectral and/or textural, and eliminates human bias in sample selection. Using a hierarchical sampling approach, ISOSCELES can obtain a training set that reflects both between-scene variability, such as in viewing angle and time of day, and within-scene variability at the level of individual training samples. We verify the method by demonstrating its superiority to stratified random sampling in the challenging task of adapting a pre-trained model to a new image and spatial domain for country-scale building extraction. Using a pair of hand-labeled training sets comprising 1,987 sample image chips, a total of 496,000,000 individually labeled pixels, we show, across three distinct model architectures, an increase in accuracy, as measured by F1-score, of 2.2–4.2%.

42 ENGINEERING↗

Neural operator prediction of linear instability waves in high-speed boundary layers

Here, we investigate if neural operators can predict the linear evolution of instability waves in high-speed boundary layers. To this end, we extend the design of the DeepOnet to ensure accurate and robust predictions, and also to perform data assimilation. In particular, we train DeepONet to take as inputs an upstream disturbance and a downstream location of interest, and to provide as output the perturbation field downstream in the boundary layer. DeepONet thus approximates the linearized and parabolized Navier-Stokes operator for this flow. For successful application to the high-speed boundary layer problem, we add sample weighting and Fourier input features to the regular DeepONet formulation. Once trained, the DeepOnet can perform fast and accurate predictions of the downstream disturbances within the range of training frequencies (inside the distribution). In addition, we show that DeepONet can solve the inverse problem, where downstream wall measurements are adopted as input, and a trained network can predict the upstream disturbances that led to these observations. This capability, along with the forward predictions, allows us to perform a full data assimilation cycle efficiently: starting from wall-pressure data, we predict the upstream disturbance using the inverse DeepONet and its evolution using the forward DeepONet. Finally, we introduce three new metrics to benchmark the training, evaluation and break-even cost of neural operators.

42 ENGINEERING↗

Assessments of epistemic uncertainty using Gaussian stochastic weight averaging for fluid-flow regression

Here, we use Gaussian stochastic weight averaging (SWAG) to assess the epistemic uncertainty associated with neural-network-based function approximation relevant to fluid flows. SWAG approximates a posterior Gaussian distribution of each weight, given training data, and a constant learning rate. Having access to this distribution, it is able to create multiple models with various combinations of sampled weights, which can be used to obtain ensemble predictions. The average of such an ensemble can be regarded as the 'mean estimation', whereas its standard deviation can be used to construct 'confidence intervals', which enable us to perform uncertainty quantification (UQ) with regard to the training process of neural networks. We utilize representative neural-network-based function approximation tasks for the following cases: (i) a two-dimensional circular-cylinder wake; (ii) the DayMET dataset (maximum daily temperature in North America); (iii) a three-dimensional square-cylinder wake; and (iv) urban flow, to assess the generalizability of the present idea for a wide range of complex datasets. SWAG-based UQ can be applied regardless of the network architecture, and therefore, we demonstrate the applicability of the method for two types of neural networks: (i) global field reconstruction from sparse sensors by combining convolutional neural network (CNN) and multi-layer perceptron (MLP); and (ii) far-field state estimation from sectional data with two-dimensional CNN. We find that SWAG can obtain physically-interpretable confidence-interval estimates from the perspective of epistemic uncertainty. This capability supports its use for a wide range of problems in science and engineering.

97 MATHEMATICS AND COMPUTING↗

A Novel Data-based Stochastic Distribution Control for Non-Gaussian Stochastic Systems

In this note, we present a novel data-based approach to investigate the non-Gaussian stochastic distribution control problem. As the motivation of this note, the existing methods have been summarised regarding to the drawbacks, for example, neural network weights training for unknown stochastic distribution and so on. To overcome these disadvantages, a new transformation for dynamic probability density function is given by kernel density estimation using interpolation. Based upon this transformation, a representative model has been developed while the stochastic distribution control problem has been transformed into an optimisation problem. Then, data-based direct optimisation and identification-based indirect optimisation have been proposed. In addition, the convergences of the presented algorithms are analysed and the effectiveness of these algorithms has been evaluated by numerical examples. In summary, the contributions of this note are as follows: 1) a new data-based probability density function transformation is given; 2) the optimisation algorithms are given based on the presented model; and 3) a new research framework is demonstrated as the potential extensions to the existing stochastic distribution control.

42 ENGINEERING↗

DeepFreeze: Towards Scalable Asynchronous Checkpointing of Deep Learning Models

In the age of big data, deep learning has emerged as a powerful tool to extract insight and exploit its value, both in industry and scientific applications. One common pattern emerging in such applications is frequent checkpointing of the state of the learning model during training, needed in a variety of scenarios: analysis of intermediate states to explain features and correlations with training data, exploration strategies involving alternative models that share a common ancestor, knowledge transfer, resilience, etc. However, with increasing size of the learning models and popularity of distributed data-parallel training approaches, simple checkpointing techniques used so far face several limitations: low serialization performance, blocking I/O, stragglers due to the fact that only a single process is involved in checkpointing. This paper proposes a checkpointing technique specifically designed to address the aforementioned limitations, introducing efficient asynchronous techniques to hide the overhead of serialization and I/O, and distribute the load over all participating processes. Experiments with two deep learning applications (CANDLE and ResNet) on a pre-Exascale HPC platform (Theta) shows significant improvement over state-of-art, both in terms of checkpointing duration and runtime overhead.

checkpointing↗

Guidelines for creating artificial neural network empirical interatomic potential from first-principles molecular dynamics data under specific conditions and its application to $α$-Ag 2 Se

First-principles molecular dynamics (FPMD) simulations are highly accurate, but due to their high calculation cost, the computational scale is often limited to hundreds of atoms and few picoseconds under specific temperature and pressure conditions. As such, we present here the guidelines for creating artificial neural network empirical interatomic potential (ANN potential) trained with such a limited FPMD data, which can perform long time scale MD simulations at least under the same conditions. The FPMD data for training are prepared on the basis of the convergence of radial distribution function [$g(r)$]. While training the ANN using total energy and atomic forces of the FPMD data, the error of pressure is also monitored and minimized. To create further robust potential, we add a small amount of FPMD data to reproduce the interaction between two atoms that are close to each other. ANN potentials for $α$-Ag 2 Se were created as an application example, and it has been confirmed that not only $g(r)$ and mean square displacements but also the specific heat requiring a long time scale simulation matched the FPMD and the experimental values. In addition, the MD simulation using the ANN potential achieved over 10 4 acceleration over the FPMD one. The guidelines proposed here mitigate the creation difficulty of the ANN potential, and a lot of FPMD data sleeping on the hard disk after the research may be put on the front stage again.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗