Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Activation function”

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 37 records · Page 2
All sourcesNASA NTRSDOE OSTI
DOE OSTI2026

Precise Modeling of a Complex Solenoidal Magnetic Field Using a Combination of Analytic Functions and a PINN

We demonstrate an iterative approach to modeling a sparsely measured magnetic field in a large-bore solenoid. This approach uses a hybrid of traditional and machine learning techniques. The traditional technique is a linear least-squares fit using a series solution to Laplace's equation, while the machine learning technique involves the training of a physics-informed neural network (PINN) on the least-squares fit residuals. We use a newly defined activation function "DELTAsnake," a modification to the snake activation function proposed by Ziyin et al. that allows for stronger curvature and non-monotonicity. The combined model approximately obeys Maxwell's equations to a level sufficient for producing high quality physics simulations and analysis. Our approach is applied to a highly realistic calculation of the expected magnetic field in the Mu2e experiment's Detector Solenoid which includes a simple model for the expected statistical measurement uncertainties. Using ten toy measurement simulations, we demonstrate the capabilities of our model in comparison to the least-squares method alone; the least-squares method alone results in a reduced chi-squared statistic of ${2.15 \pm 0.01}$, while our approach improves the reduced chi-square to ${1.034 \pm 0.005}$. Furthermore, for an average toy simulation, we show that the range of the RMS of the three field component residuals reduces from ${0.07-0.37}$ Gauss to ${0.05-0.07}$ Gauss. We find that this novel method is robust against a realistic systematic uncertainty deriving from Hall probe calibration bias and can be used to significantly reduce the number of measurements required to achieve an accurate model.

Kampa, Cole [Caltech] (ORCID:0000000192972920)↗
NASA NTRS1980

Effect of sharply lowered muscular activity on the thyroid gland of the white rat

The effect of hypokinesia on the thyroid gland of 200 white rats was studied. The rats were kept in 16x6x6 cm cages for 90 days. The functional activity of the thyroids increased after 24 hrs of partial immobilization and peaked after 15 days. After 30 days of immobilization, the functional activity returned to normal in one third of the test animals and after 60 days in all animals. After 15 days of immobilization, the test animals began to lose weight (in comparison to the controls) and remained underweight for the rest of the test period (up to 90 days). When returned to normal conditions, they caught up with and even overtook in weight the control animals after about 1 month. All changes produced by hypokinesia were reversible after 1 month.

Bekishev, K.↗
NASA NTRS1985

The human role in space: Technology, economics and optimization

Man-machine interactions in space are explored in detail. The role and the degree of direct involvement of humans that will be required in future space missions are investigated. An attempt is made to establish valid criteria for allocating functional activities between humans and machines and to provide insight into the technological requirements, economics, and benefits of the human presence in space. Six basic categories of man-machine interactions are considered: manual, supported, augmented, teleoperated, supervised, and independent. Appendices are included which provide human capability data, project analyses, activity timeline profiles and data sheets for 37 generic activities, support equipment and human capabilities required in these activities, and cumulative costs as a function of activity for seven man-machine modes.

Hall, S. B.↗
NASA NTRS1994

Preconditioning electromyographic data for an upper extremity model using neural networks

A back propagation neural network has been employed to precondition the electromyographic signal (EMG) that drives a computational model of the human upper extremity. This model is used to determine the complex relationship between EMG and muscle activation, and generates an optimal muscle activation scheme that simulates the actual activation. While the experimental and model predicted results of the ballistic muscle movement are very similar, the activation function between the start and the finish is not. This neural network preconditions the signal in an attempt to more closely model the actual activation function over the entire course of the muscle movement.

Roberson, D. J.↗
NASA NTRS2001

Expression of extracellular calcium-sensing receptor in human osteoblastic MG-63 cell line

We have previously shown the expression of the extracellular calcium (Ca2+o)-sensing receptor (CaR) in osteoblast-like cell lines, and others have documented its expression in sections of murine, bovine, and rat bone. The existence of the CaR in osteoblasts remains controversial, however, since some studies have failed to document its expression in the same osteoblast-like cell lines. The goals of the present study were twofold. 1) We sought to determine whether the CaR is expressed in the human osteoblast-like cell line, MG-63, which has recently been reported by others not to express this receptor. 2) We investigated whether the CaR, if present in MG-63 cells, is functionally active, since most previous studies have not proven the role of the CaR in mediating known actions of Ca2+o on osteoblast-like cells. We used immunocytochemistry and Western blotting with the specific, affinity-purified anti-CaR antiserum 4637 as well as Northern blot analysis and RT-PCR using a riboprobe and PCR primers specific for the human CaR, respectively, to show readily detectable CaR protein and mRNA expression in MG-63 cells. Finally, we employed the patch-clamp technique to show that an elevation in Ca2+o as well as the specific, allosteric CaR activator NPS R-467 (0.5 microM), but not its less active stereoisomer NPS S-467 (0.5 microM), activate an outward K+ channel in MG-63 cells, strongly suggesting that the CaR in MG-63 cells is not only expressed but is functionally active.

NASA Discipline Regulatory Physiology↗
DOE OSTI2025

Uncertainty propagation in feed-forward neural network models

We develop new uncertainty propagation methods for feed-forward neural network architectures with leaky ReLU activation functions subject to random perturbations in the input vectors. In particular, we derive analytical expressions for the probability density function (PDF) of the neural network output and its statistical moments as a function of the input uncertainty and the parameters of the network, i.e., weights and biases. A key finding is that an appropriate linearization of the leaky ReLU activation function yields accurate statistical results even for large perturbations in the input vectors. This can be attributed to the way information propagates through the network. We also propose new analytically tractable Gaussian copula surrogate models to approximate the full joint PDF of the neural network output. To validate our theoretical results, we conduct Monte Carlo simulations and a thorough error analysis on a multi-layer neural network representing a nonlinear integro-differential operator between two polynomial function spaces. Our findings demonstrate excellent agreement between the theoretical predictions and Monte Carlo simulations.

MLP networks↗
NASA NTRS2016

Comparative Activity and Functional Ecology of Permafrost Soils and Lithic Niches in a Hyper-Arid Polar Desert

This study represents the first metagenomic interrogation of Antarctic permafrost and polar cryptoendolithic microbial communities. The results underlie two different habitability conditions in the same location under extreme cold and dryness: the permafrost habitat where viable microbial life and activity is questionable, and the cryptoendolithic habitat which contains organisms capable of growth under the extreme conditions of the Antarctic Dry Valleys.

Antarctic permafrost↗
DOE OSTI2025

Deep Learning without Global Optimization by Random Fourier Neural Networks

Here we introduce a new training algorithm for deep neural networks that utilize random complex exponential activation functions. Our approach employs a Markov chain Monte Carlo sampling procedure to iteratively train network layers, avoiding global and gradient-based optimization while maintaining error control. It consistently attains the theoretical approximation rate for residual networks with complex exponential activation functions, determined by network complexity. Additionally, it enables efficient learning of multiscale and high-frequency features, producing interpretable parameter distributions. Despite using sinusoidal basis functions, we do not observe Gibbs phenomena in approximating discontinuous target functions.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2025

Functional group activation and coupling agent migration induced by plasma treatment in adhesive for enhanced toughness of metal-composite joints

Commercial adhesive paste was additionally activated using low-power oxygen plasma with different plasma exposure durations to investigate its interfacial bonding contribution in adhesively-bonded Aural-5/CFRP-PA6 double cantilever beam (DCB) joints with plasma-treated adherends. The plasma-treated adhesive showed enhanced functional peak intensities in Fourier transfer infrared spectroscopy (FTIR) as plasma exposure time was increased, and some degree of oxidation was also detected via X-ray photoelectron spectroscopy (XPS). A small molecule, dicyandiamde (coupling agent) in adhesive was migrated to adhesive layer, where double concentration of coupling agent was detected after 10 minutes of plasma treatment. As a result, characterization of the adhesive/CFRP-PA6 interface after DCB fracture showed additional chemical bonds formed through amine-initiated epoxy ring opening polymerization, as well as amide bonds and ester bonds. In conclusion, these additional bonds at the adhesive/CFRP-PA6 interfaces led to significantly increased failure extensions and fracture energies of the joints as the plasma exposure time on the adhesive increased, compared to the joints with plasma-treated adherends but non-plasma-treated adhesive as commonly seen in the literature.

36 MATERIALS SCIENCE↗
DOE OSTI2024

Physics-informed neural networks for heterogeneous poroelastic media

This study presents a novel physics-informed neural network (PINN) framework for modeling poroelasticity in heterogeneous media with material interfaces. The approach introduces a composite neural network (CoNN) where separate neural networks predict displacement and pressure variables for each material. While sharing identical activation functions, these networks are independently trained for all other parameters. To address challenges posed by heterogeneous material interfaces, the CoNN is integrated with the Interface-PINNs (I-PINNs) framework (Sarma et al., Comput. Methods Appl. Mech. Eng. 429: 117135, 2024), allowing different activation functions across material interfaces. Further, this ensures accurate approximation of discontinuous solution fields and gradients. Performance and accuracy of this combined architecture were evaluated against the conventional PINNs approach, a single neural network (SNN) architecture, and the eXtended PINNs (XPINNs) framework through two one-dimensional benchmark examples with discontinuous material properties. The results show that the proposed CoNN with I-PINNs architecture achieves an RMSE that is two orders of magnitude better than the conventional PINNs approach and is at least 40 times faster than the SNN framework. Compared to XPINNs, the proposed method achieves an RMSE at least one order of magnitude better and is 40% faster.

42 ENGINEERING↗
NASA NTRS1990

Discrete event simulation tool for analysis of qualitative models of continuous processing systems

An artificial intelligence design and qualitative modeling tool is disclosed for creating computer models and simulating continuous activities, functions, and/or behavior using developed discrete event techniques. Conveniently, the tool is organized in four modules: library design module, model construction module, simulation module, and experimentation and analysis. The library design module supports the building of library knowledge including component classes and elements pertinent to a particular domain of continuous activities, functions, and behavior being modeled. The continuous behavior is defined discretely with respect to invocation statements, effect statements, and time delays. The functionality of the components is defined in terms of variable cluster instances, independent processes, and modes, further defined in terms of mode transition processes and mode dependent processes. Model construction utilizes the hierarchy of libraries and connects them with appropriate relations. The simulation executes a specialized initialization routine and executes events in a manner that includes selective inherency of characteristics through a time and event schema until the event queue in the simulator is emptied. The experimentation and analysis module supports analysis through the generation of appropriate log files and graphics developments and includes the ability of log file comparisons.

Malin, Jane T.↗
NASA NTRS1997

Linear and nonlinear ARMA model parameter estimation using an artificial neural network

This paper addresses parametric system identification of linear and nonlinear dynamic systems by analysis of the input and output signals. Specifically, we investigate the relationship between estimation of the system using a feedforward neural network model and estimation of the system by use of linear and nonlinear autoregressive moving-average (ARMA) models. By utilizing a neural network model incorporating a polynomial activation function, we show the equivalence of the artificial neural network to the linear and nonlinear ARMA models. We compare the parameterization of the estimated system using the neural network and ARMA approaches by utilizing data generated by means of computer simulations. Specifically, we show that the parameters of a simulated ARMA system can be obtained from the neural network analysis of the simulated data or by conventional least squares ARMA analysis. The feasibility of applying neural networks with polynomial activation functions to the analysis of experimental data is explored by application to measurements of heart rate (HR) and instantaneous lung volume (ILV) fluctuations.

NASA Discipline Regulatory Physiology↗
NASA NTRS1991

The effect of anisotropic emission from thick accretion disks on the luminosity functions of active galactic nuclei

High-luminosity active galactic nuclei (AGNs) powered by accretion onto a massive black hole (or other compact object) may have bolometric luminosities dominated by thermal emission from a geometrically thick accretion disk. Radiation from these disks is strongly anisotropic, which has important consequences for the observed luminosity distribution, and therefore for systematic biases in flux-limited samples. The effect of anisotropic emission from an ensemble of AGNs with random oriented thick disks radiating at or near the Eddington limit is calculated. Because of their higher luminosities, it is predicted face-on disks should constitute an increasing fraction of observed high-redshift, high-luminosity AGNs. Comparison of the results with observed quasar luminosity functions suggests a narrow mass distribution with an upper limit of about a billion solar masses for high-redshift quasars.

Urry, C. M.↗
NASA NTRS1972

Functions and activities of the Arizona Regional Ecological Test Site

The Arizona Regional Ecological Test Site (ARETS) project and its efforts to coordinate and transfer satellite remote sensing techniques to solving earth resources and environmental problems in the Arizona area are discussed. Project objectives include: (1) establishing a multidiscipline resources and environment team in Arizona capable of applying satellite and supporting aircraft data to expedite remote sensing technology transfer to user agencies; (2) determine future resources and environment information needs for Arizona; (3) conduct coordinated multidiscipline experiments in the use of ERTS data; and (4) provide and recommend facilities, instruments, and training necessary to support the practical uses of remote sensing data.

Lepley, L. K.↗
DOE OSTI2025

Simplifying activations with linear approximations in neural networks

A key step in Neural Networks is activation. Among the different types of activation functions, sigmoid, tanh, and others involve the usage of exponents for calculation. From a hardware perspective, exponential implementation implies the usage of Taylor series or repeated methods involving many addition, multiplication, and division steps, and as a result are power-hungry and consume many clock cycles. We implement a piecewise linear approximation of the sigmoid function as a replacement for standard sigmoid activation libraries. This approach provides a practical alternative by leveraging piecewise segmentation, which simplifies hardware implementation and improves computational efficiency. In this paper, we detail piecewise functions that can be implemented using linear approximations and their implications for overall model accuracy and performance gain. Our results show that for the DenseNet, ResNet, and GoogLeNet architectures, the piecewise linear approximation of the sigmoid function provides faster execution times compared to the standard TensorFlow sigmoid implementation while maintaining comparable accuracy. Specifically, for MNIST with DenseNet, accuracy reaches 99.91% (Piecewise) vs. 99.97% (Base) with up to 1.31x speedup in execution time. For CIFAR-10 with DenseNet, accuracy improves to 98.97% (Piecewise) vs. 99.40% (Base) while achieving 1.24x faster execution. Similarly, for CIFAR-100 with DenseNet, the accuracy is 97.93% (Piecewise) vs. 98.39% (Base), with a 1.18x execution time reduction. These results confirm the proposed method’s capability to efficiently process large-scale datasets and computationally demanding tasks, offering a practical means to accelerate deep learning models, including LSTMs, without compromising accuracy.

Activation function↗
NASA NTRS2010

Multifunctional and biologically active matrices from multicomponent polymeric solutions

The present invention relates to a biologically active functionalized electrospun matrix to permit immobilization and long-term delivery of biologically active agents. In particular the invention relates to a functionalized polymer matrix comprising a matrix polymer, a compatibilizing polymer and a biomolecule or other small functioning molecule. In certain aspects the electrospun polymer fibers comprise at least one biologically active molecule functionalized with low molecular weight heparin. Examples of active molecules that may be used with the multicomponent polymer of the invention include, for example, a drug, a biopolymer, for example a growth factor, a protein, a peptide, a nucleotide, a polysaccharide, a biological macromolecule or the like. The invention is further directed to the formation of functionalized crosslinked matrices, such as hydrogels, that include at least one functionalized compatibilizing polymer capable of assembly.

Kiick, Kristi L.↗
NASA NTRS1984

A new minimax algorithm

The representation min t s.t. F(I)(x). - t less than or equal to 0 for all i is examined. An active set strategy is designed of functions: active, semi-active, and non-active. This technique will help in preventing zigzagging which often occurs when an active set strategy is used. Some of the inequality constraints are handled with slack variables. Also a trust region strategy is used in which at each iteration there is a sphere around the current point in which the local approximation of the function is trusted. The algorithm is implemented into a successful computer program. Numerical results are provided.

Vardi, A.↗