Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Neural Networks (Computer)”

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 235 records · Page 13

Optimizing the optimizer for physics-informed neural networks and Kolmogorov-Arnold networks

Physics-Informed Neural Networks (PINNs) have revolutionized the computation of PDE solutions by integrating partial differential equations (PDEs) into the neural network’s training process as soft constraints, becoming an important component of the scientific machine learning (SciML) ecosystem. More recently, physics-informed Kolmogorv-Arnold networks (PIKANs) have also shown to be effective and comparable in accuracy with PINNs. In their current implementation, both PINNs and PIKANs are mainly optimized using first-order methods like Adam, as well as quasi-Newton methods such as BFGS and its low-memory variant, L-BFGS. However, these optimizers often struggle with highly nonlinear and non-convex loss landscapes, leading to challenges such as slow convergence, local minima entrapment, and (non)degenerate saddle points. In this study, we investigate the performance of Self- Scaled BFGS (SSBFGS), Self-Scaled Broyden (SSBroyden) methods and other advanced quasi-Newton schemes, including BFGS and L-BFGS with different line search strategies. These methods dynamically rescale updates based on historical gradient information, thus enhancing training efficiency and accuracy. We systematically compare these optimizers – using both PINNs and PIKANs – on key challenging PDEs, including the Burgers, Allen-Cahn, Kuramoto-Sivashinsky, Ginzburg-Landau, and Stokes equations. Additionally, we evaluate the performance of SSBFGS and SSBroyden for Deep Operator Network (DeepONet) architectures, demonstrating their effectiveness for data-driven operator learning. Our findings provide state-of-the-art results with orders-of-magnitude accuracy improvements without the use of adaptive weights or any other enhancements typically employed in PINNs. More broadly, our work reveal insights into the effectiveness of quasi-Newton optimization strategies in significantly improving the convergence and accurate generalization of PINNs and PIKANs.

97 MATHEMATICS AND COMPUTING↗

Compute in‐Memory with Non‐Volatile Elements for Neural Networks: A Review from a Co‐Design Perspective

Abstract Deep learning has become ubiquitous, touching daily lives across the globe. Today, traditional computer architectures are stressed to their limits in efficiently executing the growing complexity of data and models. Compute‐in‐memory (CIM) can potentially play an important role in developing efficient hardware solutions that reduce data movement from compute‐unit to memory, known as the von Neumann bottleneck. At its heart is a cross‐bar architecture with nodal non‐volatile‐memory elements that performs an analog multiply‐and‐accumulate operation, enabling the matrix‐vector‐multiplications repeatedly used in all neural network workloads. The memory materials can significantly influence final system‐level characteristics and chip performance, including speed, power, and classification accuracy. With an over‐arching co‐design viewpoint, this review assesses the use of cross‐bar based CIM for neural networks, connecting the material properties and the associated design constraints and demands to application, architecture, and performance. Both digital and analog memory are considered, assessing the status for training and inference, and providing metrics for the collective set of properties non‐volatile memory materials will need to demonstrate for a successful CIM technology.

36 MATERIALS SCIENCE↗

A Decade of Neural Networks: Practical Applications and Prospects

The Jet Propulsion Laboratory Neural Network Workshop, sponsored by NASA and DOD, brings together sponsoring agencies, active researchers, and the user community to formulate a vision for the next decade of neural network research and application prospects. While the speed and computing power of microprocessors continue to grow at an ever-increasing pace, the demand to intelligently and adaptively deal with the complex, fuzzy, and often ill-defined world around us remains to a large extent unaddressed. Powerful, highly parallel computing paradigms such as neural networks promise to have a major impact in addressing these needs. Papers in the workshop proceedings highlight benefits of neural networks in real-world applications compared to conventional computing techniques. Topics include fault diagnosis, pattern recognition, and multiparameter optimization.

Kemeny, Sabrina E.↗

A historical survey of algorithms and hardware architectures for neural-inspired and neuromorphic computing applications

In this study, biological neural networks continue to inspire new developments in algorithms and microelectronic hardware to solve challenging data processing and classification problems. Here in this research, we survey the history of neural-inspired and neuromorphic computing in order to examine the complex and intertwined trajectories of the mathematical theory and hardware developed in this field. Early research focused on adapting existing hardware to emulate the pattern recognition capabilities of living organisms. Contributions from psychologists, mathematicians, engineers, neuroscientists, and other professions were crucial to maturing the field from narrowly-tailored demonstrations to more generalizable systems capable of addressing difficult problem classes such as object detection and speech recognition. Algorithms that leverage fundamental principles found in neuroscience such as hierarchical structure, temporal integration, and robustness to error have been developed, and some of these approaches are achieving world-leading performance on particular data classification tasks. Additionally, novel microelectronic hardware is being developed to perform logic and to serve as memory in neuromorphic computing systems with optimized system integration and improved energy efficiency. Key to such advancements was the incorporation of new discoveries in neuroscience research, the transition away from strict structural replication and towards the functional replication of neural systems, and the use of mathematical theory frameworks to guide algorithm and hardware developments.

97 MATHEMATICS AND COMPUTING↗

Reconfigurable plasma-dielectric hybrid photonic crystal as a platform for electromagnetic wave manipulation and computing

The introduction of 3D printing has enabled fabrication of photonic crystal devices with complex crystal structures that would be challenging to construct using other fabrication methods. In this paper, we construct and characterize a photonic crystal consisting of two 4 x 8 x 8 cubic lattices composed of spherical silicon nitride elements straddling a layer of 8 plasma discharge tubes, creating a 3D hybrid plasma photonic crystal device. Integrating under-dense gaseous plasma elements provides a unique coupling dynamic between the dielectric spheres and the cylindrical plasma discharges, creating a monolithic hybrid photonic crystal with solid state and reconfigurable elements. The device has resonant modes that have attenuation peaks that are either switchable, tunable in amplitude, or tunable in frequency with variations in plasma density. The response of these bands seen with varying plasma density is confirmed through simulations when effects due to the heating of the photonic crystal from the gaseous plasma elements are accounted for in the experiments. We discuss how this reconfigurable device may be used and expanded upon for applications in photonic artificial neural networks and optical computing systems.

3D printing↗

Neural network technologies

A whole new arena of computer technologies is now beginning to form. Still in its infancy, neural network technology is a biologically inspired methodology which draws on nature's own cognitive processes. The Software Technology Branch has provided a software tool, Neural Execution and Training System (NETS), to industry, government, and academia to facilitate and expedite the use of this technology. NETS is written in the C programming language and can be executed on a variety of machines. Once a network has been debugged, NETS can produce a C source code which implements the network. This code can then be incorporated into other software systems. Described here are various software projects currently under development with NETS and the anticipated future enhancements to NETS and the technology.

Villarreal, James A.↗

Neural networks: Alternatives to conventional techniques for automatic docking

Automatic docking of orbiting spacecraft is a crucial operation involving the identification of vehicle orientation as well as complex approach dynamics. The chaser spacecraft must be able to recognize the target spacecraft within a scene and achieve accurate closing maneuvers. In a video-based system, a target scene must be captured and transformed into a pattern of pixels. Successful recognition lies in the interpretation of this pattern. Due to their powerful pattern recognition capabilities, artificial neural networks offer a potential role in interpretation and automatic docking processes. Neural networks can reduce the computational time required by existing image processing and control software. In addition, neural networks are capable of recognizing and adapting to changes in their dynamic environment, enabling enhanced performance, redundancy, and fault tolerance. Most neural networks are robust to failure, capable of continued operation with a slight degradation in performance after minor failures. This paper discusses the particular automatic docking tasks neural networks can perform as viable alternatives to conventional techniques.

Vinz, Bradley L.↗

SwitchX : Gmin-Gmax Switching for Energy-efficient and Robust Implementation of Binarized Neural Networks on ReRAM Xbars

Memristive crossbars can efficiently implement Binarized Neural Networks (BNNs) wherein the weights are stored in high-resistance states (HRS) and low-resistance states (LRS) of the synapses. We propose SwitchX mapping of BNN weights onto ReRAM crossbars such that the impact of crossbar non-idealities, that lead to degradation in computational accuracy, are minimized. Essentially, SwitchX maps the binary weights in such a manner that a crossbar instance comprises of more HRS than LRS synapses. We find BNNs mapped onto crossbars with SwitchX to exhibit better robustness against adversarial attacks than the standard crossbar mapped BNNs, the baseline. Finally, we combine SwitchX with state-aware training (that further increases the feasibility of HRS states during weight mapping) to boost the robustness of a BNN on hardware. We find that this approach yields stronger defense against adversarial attacks than adversarial training, a state-of the-art software defense. We perform experiments on a VGG16 BNN with benchmark datasets (CIFAR-10, CIFAR-100 and TinyImagenet) and use Fast Gradient Sign Method (ϵ = 0.05 to 0.3) and Projected Gradient Descent (ϵ = $\frac{2}{255}$ to $\frac{32}{255}$, α = $\frac{2}{255}$) adversarial attacks. We show that SwitchX combined with state-aware training can yield upto ~35% improvements in clean accuracy and ~6–16% in adversarial accuracies against conventional BNNs. Furthermore, an important by-product of SwitchX mapping is increased crossbar power savings, owing to an increased proportion of HRS synapses, which is furthered with state-aware training. We obtain upto ~21–22% savings in crossbar power consumption for state-aware trained BNN mapped via SwitchX on 16 × 16 and 32 × 32 crossbars using the CIFAR-10 and CIFAR-100 datasets.

97 MATHEMATICS AND COMPUTING↗

Delayed switching applied to memristor neural networks

Magnetic flux and electric charge are linked in a memristor. We reported recently that a memristor has a peculiar effect in which the switching takes place with a time delay because a memristor possesses a certain inertia. This effect was named the ''delayed switching effect.'' In this work, we elaborate on the importance of delayed switching in a brain-like computer using memristor neural networks. The effect is used to control the switching of a memristor synapse between two neurons that fire together (the Hebbian rule). A theoretical formula is found, and the design is verified by a simulation. We have also built an experimental setup consisting of electronic memristive synapses and electronic neurons.

97 MATHEMATICS AND COMPUTING↗

Sharing leaky-integrate-and-fire neurons for memory-efficient spiking neural networks

Spiking Neural Networks (SNNs) have gained increasing attention as energy-efficient neural networks owing to their binary and asynchronous computation. However, their non-linear activation, that is Leaky-Integrate-and-Fire (LIF) neuron, requires additional memory to store a membrane voltage to capture the temporal dynamics of spikes. Although the required memory cost for LIF neurons significantly increases as the input dimension goes larger, a technique to reduce memory for LIF neurons has not been explored so far. To address this, we propose a simple and effective solution, EfficientLIF-Net, which shares the LIF neurons across different layers and channels. Our EfficientLIF-Net achieves comparable accuracy with the standard SNNs while bringing up to ~4.3× forward memory efficiency and ~21.9× backward memory efficiency for LIF neurons. We conduct experiments on various datasets including CIFAR10, CIFAR100, TinyImageNet, ImageNet-100, and N-Caltech101. Furthermore, we show that our approach also offers advantages on Human Activity Recognition (HAR) datasets, which heavily rely on temporal information. The code has been released at https://github.com/Intelligent-Computing-Lab-Yale/EfficientLIF-Net.

60 APPLIED LIFE SCIENCES↗

VLSI neuroprocessors

Electronic and optoelectronic hardware implementations of highly parallel computing architectures address several ill-defined and/or computation-intensive problems not easily solved by conventional computing techniques. The concurrent processing architectures developed are derived from a variety of advanced computing paradigms including neural network models, fuzzy logic, and cellular automata. Hardware implementation technologies range from state-of-the-art digital/analog custom-VLSI to advanced optoelectronic devices such as computer-generated holograms and e-beam fabricated Dammann gratings. JPL's concurrent processing devices group has developed a broad technology base in hardware implementable parallel algorithms, low-power and high-speed VLSI designs and building block VLSI chips, leading to application-specific high-performance embeddable processors. Application areas include high throughput map-data classification using feedforward neural networks, terrain based tactical movement planner using cellular automata, resource optimization (weapon-target assignment) using a multidimensional feedback network with lateral inhibition, and classification of rocks using an inner-product scheme on thematic mapper data. In addition to addressing specific functional needs of DOD and NASA, the JPL-developed concurrent processing device technology is also being customized for a variety of commercial applications (in collaboration with industrial partners), and is being transferred to U.S. industries. This viewgraph p resentation focuses on two application-specific processors which solve the computation intensive tasks of resource allocation (weapon-target assignment) and terrain based tactical movement planning using two extremely different topologies. Resource allocation is implemented as an asynchronous analog competitive assignment architecture inspired by the Hopfield network. Hardware realization leads to a two to four order of magnitude speed-up over conventional techniques and enables multiple assignments, (many to many), not achievable with standard statistical approaches. Tactical movement planning (finding the best path from A to B) is accomplished with a digital two-dimensional concurrent processor array. By exploiting the natural parallel decomposition of the problem in silicon, a four order of magnitude speed-up over optimized software approaches has been demonstrated.

Kemeny, Sabrina E.↗

Benchmarking the Performance of Neuromorphic and Spiking Neural Network Simulators

Software simulators play a critical role in the development of new algorithms and system architectures in any field of engineering. Neuromorphic computing, which has shown potential in building brain-inspired energy-efficient hardware, suffers a slow-down in the development cycle due to a lack of flexible and easy-to-use simulators of either neuromorphic hardware itself or of spiking neural networks (SNNs), the type of neural network computation executed on most neuromorphic systems. While there are several openly available neuromorphic or SNN simulation packages developed by a variety of research groups, they have mostly targeted computational neuroscience simulations, and only a few have targeted small-scale machine learning tasks with SNNs. Evaluations or comparisons of these simulators have often targeted computational neuroscience-style workloads. In this work, we seek to evaluate the performance of several publicly available SNN simulators with respect to non-computational neuroscience workloads, in terms of speed, flexibility, and scalability. We evaluate the performance of the NEST, Brian2, Brian2GeNN, BindsNET and Nengo packages under a common front-end neuromorphic framework. Our evaluation tasks include a variety of different network architectures and workload types to mimic the computation common in different algorithms, including feed-forward network inference, genetic algorithms, and reservoir computing. We also study the scalability of each of these simulators when running on different computing hardware, from single core CPU workstations to multi-node supercomputers. Our results show that the BindsNET simulator has the best speed and scalability for most of the SNN workloads (sparse, dense, and layered SNN architectures) on a single core CPU. However, when comparing the simulators leveraging the GPU capabilities, Brian2GeNN outperforms the others for these workloads in terms of scalability. NEST performs the best for small sparse networks and is also the most flexible simulator in terms of reconfiguration capability NEST shows a speedup of at least 2x compared to the other packages when running evolutionary algorithms for SNNs. The multi-node and multi-thread capabilities of NEST show at least 2x speedup compared to the rest of the simulators (single core CPU or GPU based simulators) for large and sparse networks. We conclude our work by providing a set of recommendations on the suitability of employing these simulators for different tasks and scales of operations. We also present the characteristics for a future generic ideal SNN simulator for different neuromorphic computing workloads.

97 MATHEMATICS AND COMPUTING↗

Quantum Perturbation Theory Using Tensor Cores and a Deep Neural Network

In this work, time-independent quantum response calculations are performed using Tensor cores. This is achieved by mapping density matrix perturbation theory onto the computational structure of a deep neural network. The main computational cost of each deep layer is dominated by tensor contractions, i.e., dense matrix–matrix multiplications, in mixed-precision arithmetics, which achieves close to peak performance. Quantum response calculations are demonstrated and analyzed using self-consistent charge density-functional tight-binding theory as well as coupled-perturbed Hartree–Fock theory. For linear response calculations, a novel parameter-free convergence criterion is presented that is well-suited for numerically noisy low-precision floating point operations and we demonstrate a peak performance of almost 200 Tflops using the Tensor cores of two Nvidia A100 GPUs.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Neural Architecture Search Benchmarks: Insights and Survey

Neural Architecture Search (NAS), a promising and fast-moving research field, aims to automate the architectural design of Deep Neural Networks (DNNs) to achieve better performance on the given task and dataset. NAS methods have been very successful in discovering efficient models for various Computer Vision, Natural Language Processing, etc. The major obstacles to the advancement of NAS techniques are the demand for large computation resources and fair evaluation of various search methods. The differences in training pipeline and setting make it challenging to compare the efficiency of two NAS algorithms. A large number of NAS Benchmarks to simulate the architecture evaluation in seconds have been released over the last few years to ease the computation burden of training neural networks and can aid in the unbiased assessment of different search methods. This paper provides an extensive review of several publicly available NAS Benchmarks in the literature. We provide technical details and a deeper understanding of each benchmark and point out future directions.

97 MATHEMATICS AND COMPUTING↗

Improved Adjoint-Operator Learning For A Neural Network

Improved method of adjoint-operator learning reduces amount of computation and associated computational memory needed to make electronic neural network learn temporally varying pattern (e.g., to recognize moving object in image) in real time. Method extension of method described in "Adjoint-Operator Learning for a Neural Network" (NPO-18352).

Toomarian, Nikzad↗

Convolutional Neural Network for Segmenting Micro-X-ray Computed Tomography Images of Wood Cellular Structures

To further enhance the performance of wood products, improved tools are needed to study in situ cellular scale phenomena like mechanical deformations and moisture swelling. Micro-X-ray computed tomography (μXCT) using brilliant synchrotron light sources now has the spatial and temporal resolution for real-time visualization of phenomena in three-dimensional cellular structures. However, the tradeoff for speed includes the loss of intensity contrast between different types of materials within the imaged structure, such as cell wall and air in wood. This loss of contrast prevents traditional histogram-based segmentation methods from being used effectively. A new convolutional neural network (CNN) approach was therefore developed to segment fast μXCT images of wood into cell wall and air volumes. The fast μXCT and segmentation were demonstrated in the study of moisture swelling in loblolly pine (Pinus taeda) earlywood and latewood cellular structures conditioned at 0%, 33%, 75%, and 95% relative humidity (RH). The CNN segmentation results had a mean intersection over union (IoU) metric accuracy of 96%. Initial analysis of the swelling in the latewood revealed cell walls swelled about 25% when conditioned from 0% to 95% RH. Additionally, the widths of ray cell lumina in the transverse plane of latewood could be observed to increase at higher RH. The segmentation method presented here will facilitate future quantitative analyses in in situ μXCT studies of wood and other similar cellular materials.

Arzola-Villegas, Xavier (ORCID:0000000305369766)↗

Transfer learning of neural surrogates on multifidelity groundwater simulations

Multifidelity data used in the paper published in Advances in Water Resources 206 (2025) 105140, https://doi.org/10.1016/j.advwatres.2025.105140 The code used to process the data is openly available on GitHub at https://github.com/Model-Reduction-and-UQ-Group/Transfer_Learning_K_reconstruction Computationally inexpensive surrogates of process-based models, such as deep neural networks, enable ensemble-based computations used in risk assessment, data assimilation, etc. However, generation of large datasets required to train a neural network can be as expensive as the ensemble simulations themselves. We ameliorate this challenge by using data from multifidelity (MF) groundwater simulations and transfer learning (TL) to reduce data generation costs while maintaining model accuracy. As a computational example, we train a deep convolutional neural network (CNN) to reconstruct permeability fields from saturation maps derived from a multiphase flow model. Starting with very low- and low-fidelity data generated on increasingly coarse meshes, we pretrain the CNN, followed by output-layer training and fine-tuning using only a limited number of high-fidelity samples. We demonstrate the surrogate’s robustness when interpreting low-quality inputs—such as interpolated maps or data affected by noise—which has strong implications for the applicability in practical hydrogeological scenarios. This multilevel MF-TL strategy achieves a favorable trade-off between computational efficiency and predictive accuracy, significantly outperforming high-fidelity-only approaches under the same computational budget.

Chiofalo, Alessia [University of Bologna] (ORCID:0↗