Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “NEURAL NETWORK”

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

A Classification of G -invariant Shallow Neural Networks

When trying to fit a deep neural network (DNN) to a G-invariant target function with G a group, it only makes sense to constrain the DNN to be G-invariant as well. However, there can be many different ways to do this, thus raising the problem of “G-invariant neural architecture design”: What is the optimal Ginvariant architecture for a given problem? Before we can consider the optimization problem itself, we must understand the search space, the architectures in it, and how they relate to one another. In this paper, we take a first step towards this goal; we prove a theorem that gives a classification of all G-invariant single-hidden-layer or “shallow” neural network (G-SNN) architectures with ReLU activation for any finite orthogonal group G, and we prove a second theorem that characterizes the inclusion maps or “network morphisms” between the architectures that can be leveraged during neural architecture search (NAS). The proof is based on a correspondence of every G-SNN to a signed permutation representation of G acting on the hidden neurons; the classification is equivalently given in terms of the first cohomology classes of G, thus admitting a topological interpretation. The G-SNN architectures corresponding to nontrivial cohomology classes have, to our knowledge, never been explicitly identified in the literature previously. Using a code implementation, we enumerate the G-SNN architectures for some example groups G and visualize their structure. Lastly, we prove that architectures corresponding to inequivalent cohomology classes coincide in function space only when their weight matrices are zero, and we discuss the implications of this for NAS.

Agrawal, Devanshu↗

Graph Neural Networks and Applied Linear Algebra v.1.0

SAND2024-01365O The Graph Neural Networks and Applied Linear Algebra is companion software for the educational article with the same title. The software provides illustrative examples of graph neural networks in Matlab and Python. These stand-alone algorithms are for educational purposes. The software also includes graph neural network-based algorithms for a trainable Jacobi iteration as well as diffusion coefficient estimation. The software provides human-interpretable implementations of Graph Neural Networks in Matlab and Python. These implementations are not optimized for performance and instead emphasize readability. Sandia National Laboratories is a multimission laboratory managed and operated by National Technology & Engineering Solutions of Sandia, LLC, a wholly owned subsidiary of Honeywell International Inc., for the U.S. Department of Energy’s National Nuclear Security Administration under contract DE-NA0003525.

SciDAC↗

Polyconvex neural network models of thermoelasticity

Machine-learning function representations such as neural networks have proven to be excellent constructs for constitutive modeling due to their flexibility to represent highly nonlinear data and their ability to incorporate constitutive constraints, which also allows them to generalize well to unseen data. Here, in this work, we extend a polyconvex hyperelastic neural network framework to (isotropic) thermo-hyperelasticity by specifying the thermodynamic and material theoretic requirements for an expansion of the Helmholtz free energy expressed in terms of deformation invariants and temperature. Different formulations which a priori ensure polyconvexity with respect to deformation and concavity with respect to temperature are proposed and discussed. The physics-augmented neural networks are furthermore calibrated with a recently proposed sparsification algorithm that not only aims to fit the training data but also penalizes the number of active parameters, which prevents overfitting in the low data regime and promotes generalization. The performance of the proposed framework is demonstrated on synthetic data, which illustrate the expected thermomechanical phenomena, and existing temperature-dependent uniaxial tension and tension-torsion experimental datasets.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Evolving Ensembles of Spiking Neural Networks for Neuromorphic Systems

Evolutionary algorithms have been proposed as a solution to overcome many of the challenges associated with training spiking neural networks. While evolutionary optimization for spiking neural networks is very flexible, its performance has difficulty scaling to complex tasks and correspondingly complex network structures. Here we propose a method for evolving ensembles of spiking neural networks. By using ensemble learning, the flexibility of evolutionary optimization is fully preserved while scaling to more challenging tasks. We test the performance of the proposed method using handwritten digit classification. We investigate multiple strategies for constructing ensembles of spiking neural networks, and demonstrate that evolving ensembles of SNNs offers significant performance advantages over evolutionary optimization.

Elbrecht, Daniel↗

Estimating Watershed Subsurface Permeability From Stream Discharge Data using Deep Neural Networks, Frontiers in Earth Science: Dataset

This data package contains watershed modeling inputs and outputs as well as deep neural networks training and testing results used in "Estimating Watershed Subsurface Permeability From Stream Discharge Data using Deep Neural Networks" (Cromwell et al., 2021). We train various deep neural network models with different architectures to predict subsurface permeability from stream discharge hydrograph at the watershed outlet. The training data are obtained from ensemble simulations of hydrographs corresponding to an permeability ensemble using a fully-distributed, integrated surface-subsurface hydrologic model. The trained model is then applied to estimate the permeability of the real watershed using its observed hydrograph at the outlet. Our study demonstrates that the permeabilities of the soil and geologic facies that make significant contributions to the outlet discharge can be more accurately estimated from the discharge data. Their estimations are also more robust with observation errors. Compared to the traditional ensemble smoother method, DNNs show stronger performance in capturing the nonlinear relationship between permeability and stream hydrograph to accurately estimate permeability. Our study sheds new light on the value of the emerging deep learning methods in assisting integrated watershed modeling by improving parameter estimation, which will eventually reduce the uncertainty in predictive watershed models. For detailed information regarding watershed model description and DNNs setup, please refer to Cromwell et al., 2021. The deep_learning.zip file contains the inputs and outputs for the different deep neural networks. The watershed_model.zip file contains the inputs and outputs for watershed simulation using ATS. The figures.zip file contains raw figures and their corresponding scripts. For a more detailed description, please see the README.md file within each compressed file.

54 ENVIRONMENTAL SCIENCES↗

Hidden Markov models and neural networks for fault detection in dynamic systems

Neural networks plus hidden Markov models (HMM) can provide excellent detection and false alarm rate performance in fault detection applications, as shown in this viewgraph presentation. Modified models allow for novelty detection. Key contributions of neural network models are: (1) excellent nonparametric discrimination capability; (2) a good estimator of posterior state probabilities, even in high dimensions, and thus can be embedded within overall probabilistic model (HMM); and (3) simple to implement compared to other nonparametric models. Neural network/HMM monitoring model is currently being integrated with the new Deep Space Network (DSN) antenna controller software and will be on-line monitoring a new DSN 34-m antenna (DSS-24) by July, 1994.

Smyth, Padhraic↗

Ultra-low latency recurrent neural network inference on FPGAs for physics applications with hls4ml

Abstract Recurrent neural networks have been shown to be effective architectures for many tasks in high energy physics, and thus have been widely adopted. Their use in low-latency environments has, however, been limited as a result of the difficulties of implementing recurrent architectures on field-programmable gate arrays (FPGAs). In this paper we present an implementation of two types of recurrent neural network layers—long short-term memory and gated recurrent unit—within the hls4ml framework. We demonstrate that our implementation is capable of producing effective designs for both small and large models, and can be customized to meet specific design requirements for inference latencies and FPGA resources. We show the performance and synthesized designs for multiple neural networks, many of which are trained specifically for jet identification tasks at the CERN Large Hadron Collider.

97 MATHEMATICS AND COMPUTING↗

Using Neural Networks for Sensor Validation

This paper presents the results of applying two different types of neural networks in two different approaches to the sensor validation problem. The first approach uses a functional approximation neural network as part of a nonlinear observer in a model-based approach to analytical redundancy. The second approach uses an auto-associative neural network to perform nonlinear principal component analysis on a set of redundant sensors to provide an estimate for a single failed sensor. The approaches are demonstrated using a nonlinear simulation of a turbofan engine. The fault detection and sensor estimation results are presented and the training of the auto-associative neural network to provide sensor estimates is discussed.

Mattern, Duane L.↗

Analysis and mitigation of parasitic resistance effects for analog in-memory neural network acceleration

To support the increasing demands for efficient deep neural network processing, accelerators based on analog in-memory computation of matrix multiplication have recently gained significant attention for reducing the energy of neural network inference. However, analog processing within memory arrays must contend with the issue of parasitic voltage drops across the metal interconnects, which distort the results of the computation and limit the array size. This work analyzes how parasitic resistance affects the end-to-end inference accuracy of state-of-the-art convolutional neural networks, and comprehensively studies how various design decisions at the device, circuit, architecture, and algorithm levels affect the system's sensitivity to parasitic resistance effects. Here, a set of guidelines are provided for how to design analog accelerator hardware that is intrinsically robust to parasitic resistance, without any explicit compensation or re-training of the network parameters.

97 MATHEMATICS AND COMPUTING↗

THE CHOICE OF OPTIMAL STRUCTURE OF ARTIFICIAL NEURAL NETWORK CLASSIFIER INTENDED FOR CLASSIFICATION OF WELDING FLAWS

Nondestructive testing and evaluation are under continuous development. Currently researches are concentrated on three main topics: advancement of existing methods, introduction of novel methods and development of artificial intelligent systems for automatic defect recognition (ADR). Automatic defect classification algorithm comprises of two main tasks: creating a defect database and preparing a defect classifier. Here, the database was built using defect features that describe all geometrical and texture properties of the defect. Almost twenty carefully selected features calculated for flaws extracted from real radiograms were used. The radiograms were obtained from shipbuilding industry and they were verified by qualified operator. Two weld defect's classifiers based on artificial neural networks were proposed and compared. First model consisted of one neural network model, where each output neuron corresponded to different defect group. The second model contained five neural networks. Each neural network had one neuron on output and was responsible for detection of defects from one group. In order to evaluate the effectiveness of the neural networks classifiers, the mean square errors were calculated for test radiograms and compared.

42 ENGINEERING↗

Quantitative analysis of volatile organic compounds using ion mobility spectra and cascade correlation neural networks

Ion Mobility Spectrometry (IMS) is a powerful technique for trace organic analysis in the gas phase. Quantitative measurements are difficult, because IMS has a limited linear range. Factors that may affect the instrument response are pressure, temperature, and humidity. Nonlinear calibration methods, such as neural networks, may be ideally suited for IMS. Neural networks have the capability of modeling complex systems. Many neural networks suffer from long training times and overfitting. Cascade correlation neural networks train at very fast rates. They also build their own topology, that is a number of layers and number of units in each layer. By controlling the decay parameter in training neural networks, reproducible and general models may be obtained.

Harrington, Peter DEB.↗

Neural network based decomposition in optimal structural synthesis

The present paper describes potential applications of neural networks in the multilevel decomposition based optimal design of structural systems. The generic structural optimization problem of interest, if handled as a single problem, results in a large dimensionality problem. Decomposition strategies allow for this problem to be represented by a set of smaller, decoupled problems, for which solutions may either be obtained with greater ease or may be obtained in parallel. Neural network models derived through supervised training, are used in two distinct modes in this work. The first uses neural networks to make available efficient analysis models for use in repetitive function evaluations as required by the optimization algorithm. In the second mode, neural networks are used to represent the coupling that exists between the decomposed subproblems. The approach is illustrated by application to the multilevel decomposition-based synthesis of representative truss and frame structures.

Hajela, P.↗

Skip-Connected Self-Recurrent Spiking Neural Networks with Joint Intrinsic Parameter and Synaptic Weight Training

As an important class of spiking neural networks (SNNs), recurrent spiking neural networks (RSNNs) possess great computational power and have been widely used for processing sequential data like audio and text. However, most RSNNs suffer from two problems. 1. Due to the lack of architectural guidance, random recurrent connectivity is often adopted, which does not guarantee good performance. 2. Training of RSNNs is in general challenging, bottlenecking achievable model accuracy. To address these problems, we propose a new type of RSNNs called Skip-Connected Self-Recurrent SNNs (ScSr-SNNs). Recurrence in ScSr-SNNs is introduced in a stereotyped manner by adding self-recurrent connections to spiking neurons. The SNNs with self-recurrent connections can realize recurrent behaviors similar to those of more complex RSNNs while the error gradients can be more straightforwardly calculated due to the mostly feedforward nature of the network. The network dynamics is enriched by skip connections between nonadjacent layers. Moreover, we propose a new backpropagation (BP) method called backpropagated intrinsic plasticity (BIP) to further boost the performance of ScSr-SNNs by training intrinsic model parameters. Unlike standard intrinsic plasticity rules that adjust the neuron's intrinsic parameters according to neuronal activity, the proposed BIP method optimizes intrinsic parameters based on the backpropagated error gradient of a well-defined global loss function in addition to synaptic weight training. Here, based on challenging speech, neuromorphic speech, and neuromorphic image datasets, the proposed ScSr-SNNs can boost performance by up to 2.85% compared with other types of RSNNs trained by state-of-the-art BP methods.

97 MATHEMATICS AND COMPUTING↗

Consideration for the application of artificial neural networks to criticality code validation - 14270

The artificial neural network (ANN) is a consistent example for non-parametric parameter estimation and regression, with powerful universal approximation properties, that is distribution independent. ANN application to criticality code validation centers on aligning mathematical relationships between code validation parameters and neural network prediction for those parameter values. Two specific goals are: (1) an appropriate ANN architecture to determine validation parameter values, and (2) bootstrapping individual ANNs as a 'Prediction Interval' (PI) technique to estimate the Calculation Margin and Upper Subcritical Limit. The individual ANN workflow is segregated into Training, Validation (network), and Testing Sets, where the network predictive properties and performance are expressed in the latter set. The Bootstrap ANN replicates a large set of individual ANNs (taken from the original data distribution) where (1-α/2)100% PI estimates can be determined. Results from the ANN method are compared to a variety of validation examples: a 'toy example' from ANSI/ANS-8.24, Appendix D, a variety of methods noted in a Nuclear Criticality Safety Division 2009 paper, and several validation campaigns for highly enriched uranium and plutonium systems. The ANN method is robust and can be applied to single or multivariate inputs. A trained ANN can provide both an average estimate (e.g., single value) and an estimate over a range of independent variable values where the kUSL can be formed in a regression relationship suitable for extrapolation. The input dependence has been tested, ranging from 15 to over 400 sample size distributions. An expanded Shapiro-Wilk test (for sample sizes up to 5000) was implemented to confirm that the ANN method is transparent to distribution function. The Principle Component Analysis method was applied as an input preprocessing step embedded within the ANN in order to reduce dimensionality and correlation for multivariable inputs. This can potentially aid in determining accurate margins over applying arbitrary margins. The automatic optimization of the Bootstrap ANN PI is left for future consideration. (authors)

42 ENGINEERING↗

Satellite image analysis using neural networks

The tremendous backlog of unanalyzed satellite data necessitates the development of improved methods for data cataloging and analysis. Ford Aerospace has developed an image analysis system, SIANN (Satellite Image Analysis using Neural Networks) that integrates the technologies necessary to satisfy NASA's science data analysis requirements for the next generation of satellites. SIANN will enable scientists to train a neural network to recognize image data containing scenes of interest and then rapidly search data archives for all such images. The approach combines conventional image processing technology with recent advances in neural networks to provide improved classification capabilities. SIANN allows users to proceed through a four step process of image classification: filtering and enhancement, creation of neural network training data via application of feature extraction algorithms, configuring and training a neural network model, and classification of images by application of the trained neural network. A prototype experimentation testbed was completed and applied to climatological data.

Sheldon, Roger A.↗

Understanding Selection And Diversity For Evolution Of Spiking Recurrent Neural Networks

Evolutionary optimization or genetic algorithms have been used to optimize a variety of neural network types, including spiking recurrent neural networks, and are attractive for many reasons. However, a key impediment to their widespread use is the potential for slow training times and failure to converge to a good fitness value in a reasonable amount of time. In this work, we evaluate the effect of different selection algorithms on the performance of an evolutionary optimization method for designing spiking recurrent neural networks, including those that are meant to be deployed in a neuromorphic system. We propose a selection approach that utilizes a richer understanding of the fitness of an individual network to inform the selection process and to promote diversity in the population. We show that including this feature can provide a significant increase in performance over utilizing a standard selection approach.

Schuman, Catherine↗

Neural network based automatic limit prediction and avoidance system and method

A method for performance envelope boundary cueing for a vehicle control system comprises the steps of formulating a prediction system for a neural network and training the neural network to predict values of limited parameters as a function of current control positions and current vehicle operating conditions. The method further comprises the steps of applying the neural network to the control system of the vehicle, where the vehicle has capability for measuring current control positions and current vehicle operating conditions. The neural network generates a map of current control positions and vehicle operating conditions versus the limited parameters in a pre-determined vehicle operating condition. The method estimates critical control deflections from the current control positions required to drive the vehicle to a performance envelope boundary. Finally, the method comprises the steps of communicating the critical control deflection to the vehicle control system; and driving the vehicle control system to provide a tactile cue to an operator of the vehicle as the control positions approach the critical control deflections.

Calise, Anthony J.↗

Detection of Forced Change Within Combined Climate Fields Using Explainable Neural Networks

Abstract Assessing forced climate change requires the extraction of the forced signal from the background of climate noise. Traditionally, tools for extracting forced climate change signals have focused on one atmospheric variable at a time, however, using multiple variables can reduce noise and allow for easier detection of the forced response. Following previous work, we train artificial neural networks to predict the year of single‐ and multi‐variable maps from forced climate model simulations. To perform this task, the neural networks learn patterns that allow them to discriminate between maps from different years—that is, the neural networks learn the patterns of the forced signal amidst the shroud of internal variability and climate model disagreement. When presented with combined input fields (multiple seasons, variables, or both), the neural networks are able to detect the signal of forced change earlier than when given single fields alone by utilizing complex, nonlinear relationships between multiple variables and seasons. We use layer‐wise relevance propagation, a neural network explainability tool, to identify the multivariate patterns learned by the neural networks that serve as reliable indicators of the forced response. These “indicator patterns” vary in time and between climate models, providing a template for investigating inter‐model differences in the time evolution of the forced response. This work demonstrates how neural networks and their explainability tools can be harnessed to identify patterns of the forced signal within combined fields.

54 ENVIRONMENTAL SCIENCES↗