Search NASA⌕ Search

SEARCH · Search NASA

Results for “hardware 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 109 records · Page 6
All sourcesNASA NTRSDOE OSTI
DOE OSTI2021

Low-Power Deep Learning Inference using the SpiNNaker Neuromorphic Platform

n this presentation we will discuss recent results on using the SpiNNaker neuromorphic platform (48-chip model) for deep learning neural network inference. We use the Sandia Labs developed Whet stone spiking deep learning library to train deep multi-layer perceptrons and convolutional neural networks suitable for the spiking substrate on the neural hardware architecture. By using the massively parallel nature of SpiNNaker, we are able to achieve, under certain network topologies, substantial network tiling and consequentially impressive inference throughput. Such high-throughput systems may have eventual application in remote sensing applications where large images need to be chipped, scanned, and processed quickly. Additionally, we explore complex topologies that push the limits of the SpiNNaker routing hardware and investigate how that impacts mapping software-implemented networks to on-hardware instantiations.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2021

Freely scalable and reconfigurable optical hardware for deep learning

Abstract As deep neural network (DNN) models grow ever-larger, they can achieve higher accuracy and solve more complex problems. This trend has been enabled by an increase in available compute power; however, efforts to continue to scale electronic processors are impeded by the costs of communication, thermal management, power delivery and clocking. To improve scalability, we propose a digital optical neural network (DONN) with intralayer optical interconnects and reconfigurable input values. The path-length-independence of optical energy consumption enables information locality between a transmitter and a large number of arbitrarily arranged receivers, which allows greater flexibility in architecture design to circumvent scaling limitations. In a proof-of-concept experiment, we demonstrate optical multicast in the classification of 500 MNIST images with a 3-layer, fully-connected network. We also analyze the energy consumption of the DONN and find that digital optical data transfer is beneficial over electronics when the spacing of computational units is on the order of $$>10\,\upmu $$ > 10 μ m.

42 ENGINEERING↗
DOE OSTI2025

AI-Powered Knowledge Graphs for Neuromorphic and Energy-Efficient Computing

The surge in scientific literature obscures breakthroughs and hinders the discovery of new research paths. We propose an artificial intelligence (AI) powered framework using large language models (LLMs) and knowledge graphs (KGs) to automate parts of scientific discovery, focusing on energy-efficient AI circuits. Our hybrid approach combines LLMs, structured data, and ontology-based reasoning to construct a comprehensive knowledge graph that integrates insights across computational neuroscience, spiking neuron models, learning rules, architectural motifs, and neuromorphic device technologies. This multi-domain representation enables the generation of hypotheses that connect biological function with implementable, energy-efficient hardware architectures. Using KG embeddings and graph neural networks, the framework generates hypotheses for novel circuits, validates them through optimization on exascale HPC systems, and with tools like SuperNeuro and Fugu, the most promising designs will be prototyped in hardware. This open-source system aims to accelerate discoveries and bridging neuroscience with hardware innovation, drive collaboration, and unlock new opportunities in low-power AI computing.

Gautam, Ashish [ORNL]↗
NASA NTRS1993

Neural control of magnetic suspension systems

The purpose of this research program is to design, build and test (in cooperation with NASA personnel from the NASA Langley Research Center) neural controllers for two different small air-gap magnetic suspension systems. The general objective of the program is to study neural network architectures for the purpose of control in an experimental setting and to demonstrate the feasibility of the concept. The specific objectives of the research program are: (1) to demonstrate through simulation and experimentation the feasibility of using neural controllers to stabilize a nonlinear magnetic suspension system; (2) to investigate through simulation and experimentation the performance of neural controllers designs under various types of parametric and nonparametric uncertainty; (3) to investigate through simulation and experimentation various types of neural architectures for real-time control with respect to performance and complexity; and (4) to benchmark in an experimental setting the performance of neural controllers against other types of existing linear and nonlinear compensator designs. To date, the first one-dimensional, small air-gap magnetic suspension system has been built, tested and delivered to the NASA Langley Research Center. The device is currently being stabilized with a digital linear phase-lead controller. The neural controller hardware is under construction. Two different neural network paradigms are under consideration, one based on hidden layer feedforward networks trained via back propagation and one based on using Gaussian radial basis functions trained by analytical methods related to stability conditions. Some advanced nonlinear control algorithms using feedback linearization and sliding mode control are in simulation studies.

Gray, W. Steven↗
DOE OSTI2020

Real-time biomass feedstock particle quality detection using image analysis and machine vision

Abstract A common and costly challenge in the nascent biorefinery industry is the consistent handling and conveyance of biomass feedstock materials, which can vary widely in their chemical, physical, and mechanical properties. Solutions to cope with varying feedstock qualities will be required, including advanced process controls to adjust equipment and reject feedstocks that do not meet a quality standard. In this work, we present and evaluate methods to autonomously assess corn stover feedstock quality in real time and provide data to process controls with low-cost camera hardware. We explore the use of neural networks to classify feedstocks based on actual processing behavior and pixel matrix feature parameterization to further assess particle attributes that may explain the variable processing behavior. We used the pretrained ResNet neural network coupled with a gated recurrent unit (GRU) time-series classifier trained on our image data, resulting in binary classification of feedstock anomalies with favorable performance. The textural aspects of the image data were statistically analyzed to determine if the textural features were predictive of operational disruptions. The significant textural features were angular second moment, prominence, mean height of surface profile, mean resultant vector, shade, skewness, variation of the polar facet orientation, and direction of azimuthal facets. Expansion of these models is recommended across a wider variety of labeled feedstock images of different qualities and species to develop a more robust tool that may be deployed using low-cost cameras within biorefineries.

09 BIOMASS FUELS↗
DOE OSTI2021

Quantum-Based Molecular Dynamics Simulations Using Tensor Cores

Tensor cores, along with tensor processing units, represent a new form of hardware acceleration specifically designed for deep neural network calculations in artificial intelligence applications. Tensor cores provide extraordinary computational speed and energy efficiency but with the caveat that they were designed for tensor contractions (matrix–matrix multiplications) using only low-precision floating-point operations. Despite this perceived limitation, we demonstrate how tensor cores can be applied with high efficiency to the challenging and numerically sensitive problem of quantum-based Born–Oppenheimer molecular dynamics, which requires highly accurate electronic structure optimizations and conservative force evaluations. The interatomic forces are calculated on-the-fly from an electronic structure that is obtained from a generalized deep neural network, where the computational structure naturally takes advantage of the exceptional processing power of the tensor cores and allows for high performance in excess of 100 Tflops on a single Nvidia A100 GPU. Stable molecular dynamics trajectories are generated using the framework of extended Lagrangian Born–Oppenheimer molecular dynamics, which combines computational efficiency with long-term stability, even when using approximate charge relaxations and force evaluations that are limited in accuracy by the numerically noisy conditions caused by the low-precision tensor core floating-point operations. A canonical ensemble simulation scheme is also presented, where the additional numerical noise in the calculated forces is absorbed into a Langevin-like dynamics.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2024

Generic Multi-Layer Perceptron Inference Accelerator on FPGA (vneuron) v1.0

We have designed and implemented a neural network inference compute engine (vneuron) that can be deployed in the fabric of any FPGA without using special hardware accelerator primitive. The "vneuron" is purely written in verilog, and supports scalable neural network structure with fully connected layers and ReLU activation ( Multi-Layer Perceptron architecture) with 16 bits of precision. We have demonstrated it on an Xilinx Artix 7 FPGA for a 16-input, 8-output MLP with 3 layer, 1600 parameters. It takes 40 DSP48E and 40 BRAM18, and takes 131 clock cycles for computing (1048 ns when clocked at 125MHz). We include PyTorch quantization from a given floating point model, and provide behavioral verification simulation in the disclosed software package.

Du, Qiang↗
NASA NTRS2019

Testing a Neural Network Accelerator on a High-Altitude Balloon

The cognitive communications project has been working to re d machine learning approaches to support their deployment and sustained use in space environments. It has historically been difficult to implement such techniques on space platforms, however, due to the computational requirements they levy onto general-purpose avionics hardware. While technologies exist to accelerate the computation of aspects of neural networks, such platforms have not historically been deployed in space environments. Given that testing payloads in such environments can be both cost- and time-prohibitive, high-altitude balloons can be used as a way to approximate a space environment at a much lower cost, thus providing a cost-effective way in which to test newer approaches to hardware acceleration for artificial intelligence which may be deployed onto spacecraft more directly. This paper describes a successful test of a commercial off- the-shelf neural network accelerator on a high-altitude balloon. It begins by explaining our selection criteria when evaluating different commercial neural network acceleration techniques: primary considerations include size, weight, and power (SWaP) as well as ease of integration. Next, the paper describes the development and implementation of an experimental flight test platform: flight and ground components are discussed. Afterward, the paper discusses the experimental payload itself: this includes the experimental procedure as well as the specific image and method used for testing. Finally, the paper concludes with an evaluation of both the experimental device tested at altitude as well as the flight test framework itself, identifying how the existing platform can be used to continue tes g commercial off-the-shelf (COTS) solutions for acceleration.

Clark, Gilbert↗
DOE OSTI2020

Resilience and Robustness of Spiking Neural Networks for Neuromorphic Systems

Though robustness and resilience are commonly quoted as features of neuromorphic computing systems, the expected performance of neuromorphic systems in the face of hardware failures is not clear. In this work, we study the effect of failures on the performance of four different training algo-rithms for spiking neural networks on neuromorphic systems: two back-propagation-based training approaches (Whetstone and SLAYER), a liquid state machine or reservoir computing approach, and an evolutionary optimization-based approach (EONS). We show that these four different approaches have very different resilience characteristics with respect to simulated hardware failures. We then analyze an approach for training more resilient spiking neural networks using the evolutionary optimization approach. We show how this approach produces more resilient networks and discuss how it can be extended to other spiking neural network training approaches as well.

Schuman, Catherine↗
NASA NTRS1999

ELIPS: Toward a Sensor Fusion Processor on a Chip

The paper presents the concept and initial tests from the hardware implementation of a low-power, high-speed reconfigurable sensor fusion processor. The Extended Logic Intelligent Processing System (ELIPS) processor is developed to seamlessly combine rule-based systems, fuzzy logic, and neural networks to achieve parallel fusion of sensor in compact low power VLSI.

sensor↗
DOE OSTI2021

Bayesian tensorized neural networks with automatic rank selection

Tensor decomposition is an effective approach to compress over-parameterized neural networks and to enable their deployment on resource-constrained hardware platforms. However, directly applying tensor compression in the training process is a challenging task due to the difficulty of choosing a proper tensor rank. In order to address this challenge, this paper proposes a low-rank Bayesian tensorized neural network. Our Bayesian method performs automatic model compression via an adaptive tensor rank determination. We also present approaches for posterior density calculation and maximum a posteriori (MAP) estimation for the end-to-end training of our tensorized neural network. Here, we provide experimental validation on a two-layer fully connected neural network, a 6-layer CNN and a 110-layer residual neural network where our work produces 7.4x to 137x more compact neural networks directly from the training while achieving high prediction accuracy.

Low-rank tensor↗
DOE OSTI2024

Effects of noise on the overparametrization of quantum neural networks

Overparametrization is one of the most surprising and notorious phenomena in machine learning. Recently, there have been several efforts to study if, and how, quantum neural networks (QNNs) acting in the absence of hardware noise can be overparametrized. In particular, it has been proposed that a QNN can be defined as overparametrized if it has enough parameters to explore all available directions in state space. That is, if the rank of the quantum Fisher information matrix (QFIM) for the QNN's output state is saturated. Here, we explore how the presence of noise affects the overparametrization phenomenon. Our results show that noise can “turn on” previously zero eigenvalues of the QFIM. This enables the parametrized state to explore directions that were otherwise inaccessible, thus potentially turning an overparametrized QNN into an underparametrized one. For small noise levels, the QNN is quasioverparametrized, as large eigenvalues coexists with small ones. Then, we prove that as the magnitude of noise increases all the eigenvalues of the QFIM become exponentially suppressed, indicating that the state becomes insensitive to any change in the parameters. As such, there is a pull-and-tug effect where noise can enable new directions but also suppress the sensitivity to parameter updates. Finally, our results imply that current QNN capacity measures are ill-defined when hardware noise is present. Published by the American Physical Society 2024

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

Fast b -tagging at the high-level trigger of the ATLAS experiment in LHC Run 3

The ATLAS experiment relies on real-time hadronic jet reconstruction and b-tagging to record fully hadronic events containing b-jets. These algorithms require track reconstruction, which is computationally expensive and could overwhelm the high-level-trigger farm, even at the reduced event rate that passes the ATLAS first stage hardware-based trigger. In LHC Run 3, ATLAS has mitigated these computational demands by introducing a fast neural-network-based b-tagger, which acts as a low-precision filter using input from hadronic jets and tracks. It runs after a hardware trigger and before the remaining high-level-trigger reconstruction. This design relies on the negligible cost of neural-network inference as compared to track reconstruction, and the cost reduction from limiting tracking to specific regions of the detector. In the case of Standard Model HH → bb̅bb̅, a key signature relying on b-jet triggers, the filter lowers the input rate to the remaining high-level trigger by a factor of five at the small cost of reducing the overall signal efficiency by roughly 2%.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗