Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Deep Neural Networks”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 919 records · Page 51

Emerging applications: Neuromorphic computing and reservoir computing

The emergence of doped hafnium oxide (HfO 2 )-based ferroelectric films has enabled highly scalable and silicon-compatible ferroelectric devices, opening new frontiers in neuromorphic and reservoir computing. Among these, ferroelectric field-effect transistors (FeFETs) are particularly promising due to their analog memory characteristics and unique polarization dynamics. These properties make FeFETs ideal candidates for artificial synapses in neuromorphic architectures, supporting deep neural networks and spiking neural networks based on leaky-integrate-and-fire (LIF) mechanisms. Beyond neuromorphic computing, FeFETs also play a crucial role in physical reservoir computing, leveraging their intrinsic nonlinear and history-dependent behavior for efficient real-time learning. This approach offers significant advantages for time-series processing and edge artificial intelligence (AI) applications, addressing the growing need for energy-efficient computing. As a result, this article explores the principles, key demonstrations, and future potential of FeFET-based neuromorphic and reservoir computing, highlighting their impact on next-generation AI hardware.

36 MATERIALS SCIENCE↗

Simultaneous prediction of structural properties in epitaxially–grown GaN with quantum and conventional multi–output learning algorithms

Hundreds of GaN thin film crystal plasma–assisted molecular beam epitaxy synthesis experiment records spanning two decades were organized into a dataset correlating the growth experiment design parameters with discrete, binary determinations of crystallinity and surface morphology. Conventional data science techniques as well as both quantum and classical multi–output supervised machine learning algorithms were implemented to investigate the relationships between the operating parameter data and the structural figures of merit. Correlation coefficients, decision tree nodes, p–values, and SHAP values all support substrate temperature and gallium effusion cell conditions as being statistically significant for simultaneously influencing GaN crystallinity and surface morphology. Here, a conventional deep neural network learned best from the data, followed by a quantum–classical hybrid gradient boosting algorithm. When combined with calculations of uncertainty intervals based on VennAbers predictors, machine learning predictions of both structural properties show good agreement with results reported in published experimental literature.

36 MATERIALS SCIENCE↗

Towards Query-Efficient Black-Box Adversary with Zeroth-Order Natural Gradient Descent

Despite the great achievements of the modern deep neural networks (DNNs), the vulnerability/robustness of state-of-the-art DNNs raises security concerns in many application domains requiring high reliability. Various adversarial attacks are proposed to sabotage the learning performance of DNN models. Among those, the black-box adversarial attack methods have received special attentions owing to their practicality and simplicity. Black-box attacks usually prefer less queries in order to maintain stealthy and low costs. However, most of the current black-box attack methods adopt the first-order gradient descent method, which may come with certain deficiencies such as relatively slow convergence and high sensitivity to hyper-parameter settings. In this paper, we propose a zeroth-order natural gradient descent (ZO-NGD) method to design the adversarial attacks, which incorporates the zeroth-order gradient estimation technique catering to the black-box attack scenario and the second-order natural gradient descent to achieve higher query efficiency. The empirical evaluations on image classification datasets demonstrate that ZO-NGD can obtain significantly lower model query complexities compared with state-of-the-art attack methods.

Zhao, Pu↗

Role of physics in physics-informed machine learning

Physical systems are characterized by inherent symmetries, one of which is encapsulated in the units of their parameters and system states. These symmetries enable a lossless order-reduction, e.g., via dimensional analysis based on the Buckingham theorem. Despite the latter's benefits, machine learning (ML) strategies for the discovery of constitutive laws seldom subject experimental and/or numerical data to dimensional analysis. We demonstrate the potential of dimensional analysis to significantly enhance the interpretability and generalizability of ML-discovered secondary laws. Our numerical experiments with creeping fluid flow past solid ellipsoids show how dimensional analysis enable both deep neural networks and sparse regression reproduce old results, e.g., Stokes law for a sphere, and generate new ones, e.g., an expression for an ellipsoid misaligned with the flow direction. Furthermore, our results suggest the need to incorporate other physics-based symmetries and invariances into ML-based techniques for equation discovery.

97 MATHEMATICS AND COMPUTING↗

Data-Driven Modeling and Correction of Vehicle Dynamics

We develop a data-driven framework for learning and correcting nonautonomous vehicle dynamics. Physics-based vehicle models are often simplified for tractability and therefore exhibit inherent model-form uncertainty, motivating the need for data-driven correction. Moreover, nonautonomous dynamics are governed by time-dependent control inputs, which pose challenges in learning predictive models directly from temporal snapshot data. To address these, we reformulate the vehicle dynamics via a local parameterization of the time-dependent inputs, yielding a modified system composed ofa sequence of local parametric dynamical systems. Here, we approximate these parametric systems using two complementary approaches. First, we employ the dimension reduction and interpolation in parameter space (DRIPS) methodology to construct efficient linear surrogate models, equipped with lifted observable spaces and manifold-based operator interpolation. This enables data-efficient learning of vehicle models whose dynamics admit accurate linear representations in the lifted spaces. Second, for more strongly nonlinear systems, we employ flow map learning (FML), a deep neural network (DNN) approach that approximates the parametric evolution map without requiring special treatment of nonlinearities. We further extend FML with a transfer-learning-based model correction procedure, enabling the correction of misspecified prior models using only a sparse set of high-fidelity or experimental measurements, without assuming a prescribed form for the correction term. Through a suite of numerical experiments on unicycle, simplified bicycle, and slip-based bicycle models, we demonstrate that DRIPS offers robust and highly data-efficient learning of nonautonomous vehicle dynamics, while FML provides expressive nonlinear modeling and effective correction of model-form errors under severe data scarcity.

data-driven modeling↗

A Wrapper to Use a Machine-Learning-Based Algorithm for Earthquake Monitoring

Seismology is one of the main sciences used to monitor volcanic activity worldwide. Fast, efficient, and accurate seismicity detectors are crucial to assess the activity level of a volcano in near–real time and to issue timely warnings. Traditional real–time seismic processing software uses phase onset pickers followed by a phase association algorithm to declare an event and estimate its location. The pickers typically do not identify whether the detected phase is a P or S arrival, which can have a negative impact on hypocentral location quality and complicates phase association. We implemented the deep–neural–network–based method PhaseNet to identify in real time P and S seismic waves on data from one– and three–component seismometers. We tuned the Earthworm binder_ew associator module to use the phase identification from PhaseNet to detect and locate the events, which we archive in a SeisComP3 database. We assessed the performance of the algorithm by comparing the results with existing catalogs built to monitor seismic and volcanic activity in Mayotte and the Lesser Antilles region. Our algorithm, which we refer to as PhaseWorm, showed promising results in both contexts and clearly outperformed the previous automatic method implemented in Mayotte. As a result, this innovative real–time processing system is now operational for seismicity monitoring in Mayotte and Martinique.

58 GEOSCIENCES↗

jaxhps: An elliptic PDE solver built with machine learning in mind

Elliptic partial differential equations (PDEs) can model many physical phenomena, such as electrostatics, acoustics, wave propagation, and diffusion. In scientific machine learning settings, a high-throughput PDE solver may be required to generate a training dataset, run in the inner loop of an iterative algorithm, or interface directly with a deep neural network. To provide value to machine learning users, such a PDE solver must be compatible with standard automatic differentiation frameworks, scale efficiently when run on graphics processing units (GPUs), and maintain high accuracy for a large range of input parameters. We have designed the jaxhps package with these use-cases in mind by implementing a highly efficient and accurate solver for elliptic problems with native hardware acceleration and automatic differentiation support.

97 MATHEMATICS AND COMPUTING↗

Final Report for File System Support for Burst Buffers on HPC Systems

Distributed burst buffers are a promising storage architecture for handling I/O workloads for exascale computing. As they are being deployed on more supercomputers, a file system that efficiently manages these burst buffers for fast I/O operations carries great consequence. Over the past year, FSU team has undertaken several efforts to design, prototype and evaluate distributed file systems for burst buffers on HPC systems. These include MetaKV: a Key-Value Store for Metadata Management of Distributed Burst Buffers, a user-level file system with multiple backends, and a specialized file system for large datasets of deep neural networks. Our progress for these respective efforts are elaborated further in this report.

97 MATHEMATICS AND COMPUTING↗

Exploring Deep Learning and Sparse Matrix Format Selection

We proposed to explore the use of Deep Neural Networks (DNN) for addressing the longstanding barriers. The recent rapid progress of DNN technology has created a large impact in many fields, which has significantly improved the prediction accuracy over traditional machine learning techniques in image classifications, speech recognitions, machine translations, and so on. To some degree, these tasks resemble the decision makings in many HPC tasks, including the aforementioned format selection for SpMV and linear solver selection. For instance, sparse matrix format selection is akin to image classification—such as, to tell whether an image contains a dog or a cat; in both problems, the right decisions are primarily determined by the spatial patterns of the elements in an input. For image classification, the patterns are of pixels, and for sparse matrix format selection, they are of non-zero elements. DNN could be naturally applied if we regard a sparse matrix as an image and the format selection or solver selection as classification problems.

97 MATHEMATICS AND COMPUTING↗

Advanced Modeling of Ions in Solutions, on Surfaces, and in Biological Environments (Final Scientific/Technical Report)

The research supported by this grant focused on modeling liquid water from basic microscopic theory and on developing advanced methodologies for molecular simulation and electronic structure calculation of large and disordered systems. Water is arguably the most important molecule on earth: without it there would be no life, as we know it. Water does not behave as a simple liquid. Understanding its unique and anomalous properties at the molecular level has been the focus of intense study during the entire last century and still is a major goal of research today. To address the problem we adopted a multipronged approach that included methodological and algorithmic developments as well as simulations of important model systems. We made progress on several key issues. (i) We implemented accurate electronic structure schemes for modeling water from quantum mechanical theory that make accessible in simulations processes driven by formation and breaking of chemical bonds. We studied in this way neat bulk water and simple solvated ions. We could understand, in particular, why hydronium, i.e. the positive water ion, diffuses faster than its negative counterpart, the hydroxyl, solving a 200-year-old ionic mystery of water, as reported in a news article featuring our study on Chemistry World, the magazine of the Royal Society of Chemistry (U.K.). (ii) Using advanced statistical mechanical techniques we studied water in the deeply undercooled regime, which may hold the clue to understanding many of the water anomalies. These conditions are currently only accessible to simulations based on empirical force fields. We found that deeply undercooled water, as described by the popular force field ST2, possesses a secondary critical point, below which it would spontaneously separate into two liquids, one of low-density (LDL) and one of high-density (HDL), much in the same way in which water and oil phase-separate at room temperature. The secondary critical point is supported indirectly by many experiments, but rapid ice nucleation has so far hindered direct observation. In related studies we gained new insight on the connection of the two metastable liquids with the two corresponding amorphous ice phases, LDA and HDA, which are observed experimentally. (iii) We developed and implemented several new algorithms for computing the electronic structure of large systems and to accelerate the calculation of electron exchange energies. We applied these methodologies to model nanomaterials, such as nanoribbons and nanoflakes, and to study water absorption and splitting on nanomaterials. (iv) We developed new local order descriptors. (v) In the last year of the project we developed a methodology based on deep neural networks to construct model interatomic potentials that can accurately reproduce ab-initio quantum mechanical simulations at the computational cost of empirical force fields. These potentials will make possible to access the size and time scales that are currently only accessible with empirical force fields without loosing the predictive power of ab-initio simulations.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Exploring Flexible Communications for Streamlining DNN Ensemble Training Pipelines

Parallel training of a Deep Neural Network (DNN) ensemble on a cluster of nodes is a common practice to train multiple models in order to construct a model with a higher prediction accuracy. Existing ensemble training pipelines can perform a great deal of redundant operations, resulting in unnecessary CPU usage, or even poor pipeline performance. In order to remove these redundancies, we need pipelines with more communication flexibility than existing DNN frameworks can provide. This project investigates a series of designs to improve pipeline flexibility and adaptivity, while also increasing performance. We implement our designs using Tensorflow with Horovod, and test it using several large DNNs in a large scale GPU cluster, Titan supercomputer at Oak Ridge National Lab. Our results show that the CPU time spent during training is reduced by 2-11X. Furthermore, our implementation can achieve up to 10X speedups when CPU core limits are imposed. Our best pipeline also reduces the average power draw of the ensemble training process by 5-16% when compared to the baseline.

97 MATHEMATICS AND COMPUTING↗

Integrated Cyber/Physical Grid Resiliency Modeling

This project explored coupling modeling and analysis methods from multiple domains to address complex hybrid (cyber and physical) attacks on mission critical infrastructure. Robust methods to integrate these complex systems are necessary to enable large trade-space exploration including dynamic and evolving cyber threats and mitigations. Reinforcement learning employing deep neural networks, as in the AlphaGo Zero solution, was used to identify "best" (or approximately optimal) resilience strategies for operation of a cyber/physical grid model. A prototype platform was developed and the machine learning (ML) algorithm was made to play itself in a game of 'Hurt the Grid'. This proof of concept shows that machine learning optimization can help us understand and control complex, multi-dimensional grid space. A simple, yet high-fidelity model proves that the data have spatial correlation which is necessary for any optimization or control. Our prototype analysis showed that the reinforcement learning successfully improved adversary and defender knowledge to manipulate the grid. When expanded to more representative models, this exact type of machine learning will inform grid operations and defense - supporting mitigation development to defend the grid from complex cyber attacks! This same research can be expanded to similar complex domains.

97 MATHEMATICS AND COMPUTING↗

User-level File Systems Specialized for HPC Workloads Year-End Report

High-performance computing (HPC) clusters attract the attention of Deep Learning (DL) training users due to the clusters' powerful computation capabilities. While there have been many existing efforts to enable deep neural works to leverage the powerful CPU and GPU processors from leadership high-performance computing (HPC) systems, large-scale deep learning with larger datasets requires efficient I/O support from the underlying file and storage systems. As some current and upcoming HPC clusters have large on-node memory or are equipped with NVMe SSD on compute nodes, more distributed DL trainings are considered to leverage those storages to store dataset for efficient dataset access. Our project goal is to design a specialized DL-oriented file system to improve datasets loading performance of any DL training applications. Over the past year, the team at Florida State University has performed research activities in three different aspects, including 1) completing a specialized memory-based I/O framework (DeepIO) for improving dataset loading performance on DL applications; 2) proposing a more generalized file system (DLFS) for DL applications on node-local SSDs; 3) completing the BeeGFS performance evaluation project for Deep Neural Networks.

97 MATHEMATICS AND COMPUTING↗

Physics-Informed Machine Learning with Application to Solid Oxide Fuel Cell System Modeling and Optimization

Traditional data-driven machine learning (ML) techniques were combined with the physics-based SOFC-MP model toward improving SOFC system-level performance prediction. Four different physics-informed ML methods were proposed to demonstrate better prediction accuracy compared with traditional ML methods. First, it was preferred that the existing Kriging-based surrogate model not be entirely abandoned, so errors between Kriging regression prediction and true solution are learned using ML. The goal was to find the pattern of the error distribution as a function of model input so discrepancies between the Kriging prediction and true solution could be reduced by adding the additional ML-trained error term to the existing prediction. Second, deep neural networks (DNN) coupled with the mass balance model (MBM) have been developed to significantly decrease the reduced order model (ROM) prediction error with fewer training data compared to the traditional Kriging-based ROM and traditional DNN regression approaches. Third, DNN regression coupled with a neural networks (NN) classifier and MBM was developed. It provides superior performance on classifying the physically operational conditions for natural gas fuel cell (NGFC) than traditional classification approaches. Finally, two kinds of DNN transferring were studied and tested. One was transferring DNN for low-fidelity model data to a high-fidelity model. The second one was transferring DNN from state-of-the-art NGFC to advanced NGFC. These physic-informed ML frameworks can serve as a solid foundation for estimating U.S. Department of Energy-targeted fuel cell performance and exhibit great potential to be employed to other engineering applications.

30 DIRECT ENERGY CONVERSION↗

DNN Approximation of Nonlinear Finite Element Equations

We investigate the potential of applying (D)NN ((deep) neural networks) for approximating nonlinear mappings arising in the finite element discretization of nonlinear PDEs (partial differential equations). As an application, we apply the trained DNN to replace the coarse nonlinear operator thus avoiding the need to visit the fine level discretization in order to evaluate the actions of the true coarse nonlinear operator. The feasibility of the studied approach is demonstrated in a two level FAS (full approximation scheme) used to solve a nonlinear diffusion-reaction PDE.

97 MATHEMATICS AND COMPUTING↗

Multimodal Data Fusion via Entropy Minimization

The use of gradient-based data-driven models to solve a range of real-world remote sensing problems can in practice be limited by the uniformity of available data. Use of data from disparate sensor types, resolutions, and qualities typically requires compromises based on assumptions that are made prior to model training and may not necessarily be optimal given over-arching objectives. For example, while deep neural networks (NNs) are state-of-the-art in a variety of target detection problems, training them typically requires either limiting the training data to a subset over which uniformity can be enforced or training independent models which subsequently require additional score fusion. The method we introduce here seeks to leverage the benefits of both approaches by allowing correlated inputs from different data sources to co-influence preferred model solutions, while maintaining flexibility over missing and mismatching data. In this work we propose a new data fusion technique for gradient updated models based on entropy minimization and experimentally validate it on a hyperspectral target detection dataset. We demonstrate superior performance compared to currently available techniques using a range of realistic data scenarios, where available data has limited spacial overlap and resolution.

97 MATHEMATICS AND COMPUTING↗

Redox transistors based on TiO 2 for analogue neuromorphic computing

The ability to train deep neural networks on large data sets have made significant impacts onto artificial intelligence, but consume significant amounts of energy due to the need to move information from memory to logic units. In-memory "neuromorphic" computing presents an alternative framework that processes information directly on memory elements. In-memory computing has been limited by the poor performance of the analogue information storage element, often phase-change memory or memristors. To solve this problem, we developed two types of "redox transistors" using TiO 2 (anatase) which stores analogue information states through the electrochemical concentration of dopants in the crystal. The first type of redox transistor uses lithium as the electrochemical dopant ion, and its key advantage is low operating voltage. The second uses oxygen vacancies as the dopant, which is CMOS compatible and can retain state even when scaled to nanosized dimensions. Both devices offer significant advantages in terms of predictable analogue switching over conventional filamentary-based devices, and provide a significant advance in developing materials and devices for neuromorphic computing.

36 MATERIALS SCIENCE↗

Cognition at the Point of Sensing

Over the last 15 years, compressive sensing techniques have been developed which have the potential to greatly reduce the amount of data collected by systems while preserving the amount of information obtained. A cost of this efficiency is that a computationally-intensive optimization routine must be used to put the sensed data into a form that a person can interpret. At the same time, machine learning techniques have experienced tremendous growth as well. Machines have demonstrated the ability learn how to effectively perform tasks such as detection and classification at speeds much faster than humanly possible. Our goal in this project was to study the feasibility of using compressive sensing systems "at the edge." That is, how can compressive sensing sensors be deployed such that information is created at the remote sensor rather than sending raw data to a central processing location? Studies were performed to analyze whether machine learning could be done on the compressively sensed data in its raw form. If a machine is performing the task, is it possible to do so without putting the data into a human interpretable form? We show that this is possible for some systems, in particular a compressive sensing snapshot imaging spectrometer. Machine learning tasks were demonstrated to be more effective and more robust to noise when the machine learning algorithm worked on data in its raw form. This system is shown to outperform a traditional spectrometer. Techniques for reducing the complexity of the reconstruction routine were also analyzed. Techniques for such as data regularization, deep neural networks, and matrix completion were studied and shown to have benefits over traditional reconstruction techniques. In this project we showed that compressive sensing sensors are indeed feasible at the edge. As always, sensors and algorithms must be carefully tuned to work in the constrained environment. In this project we developed tools and techniques to enable those analyses.

47 OTHER INSTRUMENTATION↗