Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Deep neural operator”

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 289 records · Page 16
All sourcesNASA NTRSDOE OSTI
DOE OSTI2024

Bayesian reduced-order deep learning surrogate model for dynamic systems described by partial differential equations

We propose a reduced-order deep-learning surrogate model for dynamic systems described by time-dependent partial differential equations. This method employs space–time Karhunen–Loève expansions (KLEs) of the state variables and space-dependent KLEs of space-varying parameters to identify the reduced (latent) dimensions. Subsequently, a deep neural network (DNN) is used to map the parameter latent space to the state variable latent space. An approximate Bayesian method is developed for uncertainty quantification (UQ) in the proposed KL-DNN surrogate model. The KL-DNN method is tested for the linear advection–diffusion and nonlinear diffusion equations, and the Bayesian approach for UQ is compared with the deep ensembling (DE) approach, commonly used for quantifying uncertainty in DNN models. It was found that the approximate Bayesian method provides a more informative distribution of the PDE solutions in terms of the coverage of the reference PDE solutions (the percentage of nodes where the reference solution is within the confidence interval predicted by the UQ methods) and log predictive probability. The DE method is found to underestimate uncertainty and introduce bias. For the nonlinear diffusion equation, we compare the KL-DNN method with the Fourier Neural Operator (FNO) method and find that KL-DNN is 10% more accurate and needs less training time than the FNO method.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2025

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↗
DOE OSTI2024

Low latency optical-based mode tracking with machine learning deployed on FPGAs on a tokamak

Active feedback control in magnetic confinement fusion devices is desirable to mitigate plasma instabilities and enable robust operation. Optical high-speed cameras provide a powerful, non-invasive diagnostic and can be suitable for these applications. Here, in this study, we process high-speed camera data, at rates exceeding 100 kfps, on in situ field-programmable gate array (FPGA) hardware to track magnetohydrodynamic (MHD) mode evolution and generate control signals in real time. Our system utilizes a convolutional neural network (CNN) model, which predicts the n = 1 MHD mode amplitude and phase using camera images with better accuracy than other tested non-deep-learning-based methods. By implementing this model directly within the standard FPGA readout hardware of the high-speed camera diagnostic, our mode tracking system achieves a total trigger-to-output latency of 17.6 μs and a throughput of up to 120 kfps. This study at the High Beta Tokamak-Extended Pulse (HBT-EP) experiment demonstrates an FPGA-based high-speed camera data acquisition and processing system, enabling application in real-time machine-learning-based tokamak diagnostic and control as well as potential applications in other scientific domains.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2022

Wavelet and Deep-Learning-Based Approach for Generation System Problematic Parameters Identification and Calibration

Accurate models of generation systems are critical for maintaining reliable and secure grid operations. In this paper, a novel and systematic approach is proposed to identify and calibrate the generation system problematic parameters using continuous wavelet transform (CWT) and advanced deep-learning technology. The phasor measurement unit (PMU) data are used through “event playback” to check whether the parameter calibration is required, and if yes, a group of suspicious parameters will be identified as the primary problematic parameter candidates (PPCs). These primary PPCs are randomly perturbed to generate the event playback simulation data, which are used by the CWT and convolutional neural networks (CNNs) to further narrow down the primary PPCs into a smaller set of candidates. Then, the identified candidates are perturbed again to generate massive event playback simulation data for training a parameter calibration neural network. Here, we designed a multi-output neural network structure to find the mappings between the perturbed parameters and the simulation data using both CNN and long short-term memory (LSTM) models. Finally, the well-trained and tested CNN-LSTM model is used to estimate the accurate value of the suspicious parameters with actual PMU measurements. The proposed CNN-LSTM network can accurately and reliably estimate the generation-system problematic parameters, and has better performance when compared to other machine-learning methods, such as the multilayer perceptron network and the conditional variational autoencoder method. The accuracy and effectiveness of the proposed approach have been validated through simulation and real-world data.

24 POWER TRANSMISSION AND DISTRIBUTION↗
DOE OSTI2022

DNN-based policies for stochastic AC OPF

We report a prominent challenge to the safe and optimal operation of the modern power grid arises due to growing uncertainties in loads and renewables. Stochastic optimal power flow (SOPF) formulations provide a mechanism to handle these uncertainties by computing dispatch decisions and control policies that maintain feasibility under uncertainty. Most SOPF formulations consider simple control policies such as affine policies that are mathematically simple and resemble many policies used in current practice. Motivated by the efficacy of machine learning (ML) algorithms and the potential benefits of general control policies for cost and constraint enforcement, we put forth a deep neural network (DNN)-based policy that predicts the generator dispatch decisions in real time in response to uncertainty. The weights of the DNN are learnt using stochastic primal–dual updates that solve the SOPF without the need for prior generation of training labels and can explicitly account for the feasibility constraints in the SOPF. The advantages of the DNN policy over simpler policies and their efficacy in enforcing safety limits and producing near optimal solutions are demonstrated in the context of a chance constrained formulation on a number of test cases.

29 ENERGY PLANNING, POLICY, AND ECONOMY↗
DOE OSTI2025

Towards automated and real-time multi-object detection of anguilliform fishes from sonar data using YOLOv8 deep learning algorithm

Eels (Anguilla spp.), including American eels (Anguilla rostrata), European eels (Anguilla anguilla), and Japanese eels (Anguilla japonica), are species of critical management and regulatory concern due to their vulnerability to various stressors during downstream migrations. Accurate and efficient detection of migrating eels can improve our understanding of fish behaviors and fish-hydraulic structure interactions, thereby facilitating the design, operation, and optimization of more effective downstream passage facilities from both biological and economic perspectives. However, a real-time, automated framework for detecting migrating eels in real-world applications is currently lacking. Leveraging imaging sonar as a reliable technology for fish passage monitoring, field data are acquired using imaging sonar and then converted to single sonar frames/images for subsequent analysis. In this study, a framework based on the You Only Look Once Version 8 (YOLOv8)-based convolutional neural network is proposed for multi-object detection of eels and non-eel fish using the sonar images after image subtraction and additional wavelet denoising. The results from both training and testing phases demonstrate that the framework's ability can successfully detect both eels and non-eel fish in preprocessed sonar images, achieving F1-scores and mAP@0.50 exceeding 0.84. Additionally, the incorporation of wavelet denoising during preprocessing slightly improve detection performance. Furthermore, the transferability of this framework from eel to lamprey detection is demonstrated to be feasible given the similar morphological characteristics of these two species. Overall, the proposed framework achieves accurate and efficient detection of migrating eels, providing reliable and real-time information that can help conserve vulnerable eel and eel-like populations.

Deep learning↗
NASA NTRS2019

Small target detection for search and rescue operations using distributed deep learning and synthetic data generation

It is important to find the target as soon as possible for search and rescue operations. Surveillance camera systems and unmanned aerial vehicles (UAVs) are used to support search and rescue. Automatic object detection is important because a person cannot monitor multiple surveillance screens simultaneously for 24 hours. Also, the object is often too small to be recognized by the human eye on the surveillance screen. This study used UAVs around the Port of Houston and fixed surveillance cameras to build an automatic target detection system that supports the US Coast Guard (USCG) to help find targets (e.g., person overboard). We combined image segmentation, enhancement, and convolution neural networks to reduce detection time to detect small targets. We compared the performance between the auto-detection system and the human eye. Our system detected the target within 8 seconds, but the human eye detected the target within 25 seconds. Our systems also used synthetic data generation and data augmentation techniques to improve target detection accuracy. This solution may help the search and rescue operations of the first responders in a timely manner.

Chow, Edward↗
DOE OSTI2025

Even Higher-Level Synthesis: An Exploration of AI Hardware Accelerators using HLS4ML

With the rise of artificial intelligence, the popularization of deep learning, and a constantly evolving industry, the demand for flexible and efficient tools has never been greater. As algorithms grow more complex, their runtime and energy consumption increase exponentially. Customized hardware accelerators, long used for specific mathematical operations, remain essential for managing modern applications' computational and power demands. Hardware accelerators can speed up complex computations by orders of magnitude, but their manual design and verification processes are often challenging and time-consuming. High-Level Synthesis (HLS) provides a solution by transforming high-level algorithm descriptions, typically written in C++ or SystemC, into synthesizable RTL suitable for hardware implementation. This approach reduces development time for RTL engineers while offering flexibility beyond what traditional handwritten RTL can provide. We extended this capability to the machine-learning domain with the open-source framework hls4ml, which allows neural networks trained in Python frameworks like Tensorflow or PyTorch to be synthesized into efficient hardware representations for the traditional FPGA and ASIC flows. This breakthrough addresses the growing need for reduced design turnaround and easy verification of ML hardware accelerators with low latency and power efficiency constraints. During this tutorial, we will demonstrate how Python complements HLS by simplifying the ML design process, bridging the gap between software and hardware development. Attendees will explore how we translate neural networks modeled in Python into fixed-point C++ models suitable for HLS workflows. We will dive into strategies like Value-Range Analysis and Quantization-Aware Training, which optimize these designs for deployment and evaluate their accuracy, power consumption, and energy efficiency. To exemplify these concepts, experts from Fermilab will share their experiences applying this technology to high-energy physics experiments, where real-time, low-latency processing is critical. Over the years, Fermilab engineers have demonstrated how deep neural networks, optimized for hardware using hls4ml, can meet the stringent requirements of trigger systems at the CERN Large Hadron Collider. These systems rely on rapid decision-making to process immense data volumes while retaining only the most relevant events for further analysis. The application of hls4ml has also been extended to innovative technologies like smart pixel arrays. These smart pixels integrate ML inference capabilities directly into sensor devices, enabling localized data processing at the pixel level. This approach drastically reduces the need to transmit raw data to external processing units, significantly decreasing power consumption and latency. By embedding neural networks within the pixel architecture, the smart pixels can identify and prioritize relevant data in real time, providing a highly efficient solution for edge computing in scenarios such as particle detectors and imaging systems. Fermilab's work highlights the potential of hardware-accelerated ML in scenarios where both speed and power efficiency are mission-critical. Through this tutorial, attendees will gain valuable insights into the challenges and solutions of deploying ML in hardware. Understanding how HLS and hls4ml streamline the development of neural network-based hardware accelerators is fundamental for the industry's future. Participants will learn how these technologies are shaping the future of AI and scientific computing.

Di Guglielmo, Giuseppe [Fermilab]↗
DOE OSTI2024

Deep Multitask Learning Models for Radiation Estimation at High Energy Accelerator Facility

Controlling the dose of radiation exposure in potential radioactive facilities is critical for ensuring the safety of staff and the public. Here, in this paper, we developed machine learning models to estimate radiation exposure efficiently at the Thomas Jefferson National Accelerator Facility (JLab), aiming to enhance safety in both accelerator facilities and public areas. Multiple sensors were deployed around the three experimental halls at JLab. Data on single-beam currents, energy levels, and radiation values at the sensor locations were collected during accelerator operation. We proposed a multi-task learning model for radiation estimation, utilizing either one-dimensional convolutional neural networks (1-D CNNs) or long short-term memory networks (LSTMs) as the backbone. The proposed model was trained to simultaneously estimate radiation levels at the sensor locations. Experimental results demonstrated that the proposed model with LSTM backbone achieved the best estimation performance, with an average R 2 score of 0.7557 for estimation within the same year and 0.7157 for estimation across different years. These results significantly surpassed those of competing models.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2021

Stopping criteria for ending autonomous, single detector radiological source searches

While the localization of radiological sources has traditionally been handled with statistical algorithms, such a task can be augmented with advanced machine learning methodologies. The combination of deep and reinforcement learning has provided learning-based navigation to autonomous, single-detector, mobile systems. However, these approaches lacked the capacity to terminate a surveying/search task without outside influence of an operator or perfect knowledge of source location (defeating the purpose of such a system). Two stopping criteria are investigated in this work for a machine learning navigated system: one based upon Bayesian and maximum likelihood estimation (MLE) strategies commonly used in source localization, and a second providing the navigational machine learning network with a “stop search” action. A convolutional neural network was trained via reinforcement learning in a 10 m × 10 m simulated environment to navigate a randomly placed detector-agent to a randomly placed source of varied strength (stopping with perfect knowledge during training). The network agent could move in one of four directions (up, down, left, right) after taking a 1 s count measurement at the current location. During testing, the stopping criteria for this navigational algorithm was based upon a Bayesian likelihood estimation technique of source presence, updating this likelihood after each step, and terminating once the confidence of the source being in a single location exceeded 0.9. A second network was trained and tested with similar architecture as the previous but which contained a fifth action: for self-stopping. The accuracy and speed of localization with set detector and source initializations were compared over 50 trials of MLE-Bayesian approach and 1000 trials of the CNN with self-stopping. The statistical stopping condition yielded a median localization error of ~1.41 m and median localization speed of 12 steps. The machine learning stopping condition yielded a median localization error of 0 m and median localization speed of 17 steps. This work demonstrated two stopping criteria available to a machine learning guided, source localization system.

38 RADIATION CHEMISTRY, RADIOCHEMISTRY, AND NUCLEA↗
DOE OSTI2020

Machine learning for nonintrusive model order reduction of the parametric inviscid transonic flow past an airfoil

Fluid flow in the transonic regime finds relevance in aerospace engineering, particularly in the design of commercial air transportation vehicles. Computational fluid dynamics models of transonic flow for aerospace applications are computationally expensive to solve because of the high degrees of freedom as well as the coupled nature of the conservation laws. While these issues pose a bottleneck for the use of such models in aerospace design, computational costs can be significantly minimized by constructing special, structure-preserving surrogate models called reduced-order models. In this work, we propose a machine learning method to construct reduced-order models via deep neural networks and we demonstrate its ability to preserve accuracy with a significantly lower computational cost. In addition, our machine learning methodology is physics-informed and constrained through the utilization of an interpretable encoding by way of proper orthogonal decomposition. Application to the inviscid transonic flow past the RAE2822 airfoil under varying freestream Mach numbers and angles of attack, as well as airfoil shape parameters with a deforming mesh, shows that the proposed approach adapts to high-dimensional parameter variation well. Notably, the proposed framework precludes the knowledge of numerical operators utilized in the data generation phase, thereby demonstrating its potential utility in the fast exploration of design space for diverse engineering applications. Comparison against a projection-based nonintrusive model order reduction method demonstrates that the proposed approach produces comparable accuracy and yet is orders of magnitude computationally cheap to evaluate, despite being agnostic to the physics of the problem.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2026

A Benchmark Suite for Evaluating Scientific AI Workloads on GPUs

AI applications have been steadily increasing in the allocation portfolio among leadership computing facilities. These applications depend on deep learning frameworks with hardware acceleration and underlying software systems. With the rapid development of applications, software stacks, and hardware devices, it is essential to evaluate the performance of core operations in AI workloads for direction of optimizations and procurement of next-generation high-performance computing (HPC) infrastructures. Currently, most benchmarks lack scientific AI workloads. So, we present DeepKernelBench and the experimental results of evaluating the benchmark suite for early observations and performance comparisons on datacenter GPUs using representative workloads for scientific AI, including Attentions, General matrix multiplications, Geometrics and Fourier neural operations.

Jin, Zheming [Advanced Micro Devices (AMD)]↗
DOE OSTI2023

Optical information transfer through random unknown diffusers using electronic encoding and diffractive decoding

Free-space optical information transfer through diffusive media is critical in many applications, such as biomedical devices and optical communication, but remains challenging due to random, unknown perturbations in the optical path. We demonstrate an optical diffractive decoder with electronic encoding to accurately transfer the optical information of interest, corresponding to, e.g., any arbitrary input object or message, through unknown random phase diffusers along the optical path. This hybrid electronic-optical model, trained using supervised learning, comprises a convolutional neural network-based electronic encoder and successive passive diffractive layers that are jointly optimized. After their joint training using deep learning, our hybrid model can transfer optical information through unknown phase diffusers, demonstrating generalization to new random diffusers never seen before. The resulting electronic-encoder and optical-decoder model was experimentally validated using a 3D-printed diffractive network that axially spans <70λ, where λ = 0.75 mm is the illumination wavelength in the terahertz spectrum, carrying the desired optical information through random unknown diffusers. The presented framework can be physically scaled to operate at different parts of the electromagnetic spectrum, without retraining its components, and would offer low-power and compact solutions for optical information transfer in free space through unknown random diffusive media.

36 MATERIALS SCIENCE↗
NASA NTRSDate not supplied

Toward The Development of Hailstorm Climatologies Derived From Reanalyses and Infared/Passive Microwave Satellite Imagers

Geostationary satellite imagers, such as those of the Geostationary Operational Environmental Satellite (GOES) and Meteosat series, provide both historical and near-real-time observations of cloud top patterns that are commonly associated with severe convection. Environmental conditions favorable for severe weather are thought to be represented well by reanalyses. Predicting exactly where convection and costly storm hazards like hail will occur using models or satellite imagery alone, however, is extremely challenging. The multivariate combination of satellite-observed cloud patterns with reanalysis environmental parameters, linked to United States Next Generation Weather Radar- (NEXRAD-) estimated Maximum Expected Size of Hail (MESH) using a deep neural network (DNN), enables estimation of potentially severe hail likelihood for any observed storm cell. These estimates are specifically designed to make hail likelihood distinctions based on satellite-indicated points of deep convection within environments favorable for storm development. We seek an approach that can be used to estimate climatological hailstorm frequency and risk throughout the historical satellite data record. This presentation demonstrates that statistical distributions of convective parameters from satellite and reanalysis show separation between non-severe/severe hailstorm classes for predictors including overshooting cloud top temperature and area characteristics, convective available potential energy, vertical wind shear, 500 hPa temperature, mid-level lapse rate, precipitable water, and convective inhibition. These complex, multivariate predictor relationships are exploited within a DNN to produce a hail likelihood metric with a critical success index of 0.504 and Heidke skill score of 0.403, which is exceptional among recent analogous hail studies. Furthermore, applications of the DNN to select case studies demonstrate good qualitative agreement between hail likelihood and MESH. These hail classifications are aggregated across an 11-year GOES-12/13 image database to derive a hail frequency and severity climatology, which denotes the Central Plains, the Midwest, and northwestern Mexico as being the most hail-prone regions within the domain studied. Opportunities for training and applying DNN-based hailstorm predictions to recently developed GOES-8/10/12/13/16 and Meteosat Second Generation convective storm detection and characterization climatologies over South America and South Africa, respectively, will also be presented.

Kristopher Bedka↗
DOE OSTI2021

Neoclassical toroidal viscosity torque prediction via deep learning

GPECnet is a densely connected neural network that has been trained on GPEC data, to predict the plasma stability, neoclassical toroidal viscosity (NTV) torque, and optimized 3D coil current distributions for desired NTV torque profiles. Using NTV torque, driven by non-axisymmetric field perturbations in a tokamak, can be vital in optimizing pedestal performance by controlling the rotation profile in both the core, to ensure tearing stability, and the edge, to avoid edge localized modes (ELMs). The generalized perturbed equilibrium code (GPEC) software package can be used to calculate the plasma stability to 3D perturbations and the NTV torque profile generated by applied 3D magnetic fields. These calculations, however, involve complex integrations over space and energy distributions, which takes time to compute. Initially, GPECnet has been trained solely on data representative of the quiescent H-mode (QH) scenario, in which neutral beams are often balanced and toroidal rotation is low across the plasma profile. Lastly, this work provides the foundation for active control of the rotation shear using a combination of beams and 3D fields for robust and high performance QH mode operation.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2026

Identification of Common Types of Plastics by Vibrational Spectroscopic Techniques

Polyethylene Terephthalate (PET), High-Density Polyethylene (HDPE), Polyvinyl Chloride (PVC), Low-Density Polyethylene (LDPE), Polypropylene (PP), and Polystyrene (PS) account for most plastic use worldwide, with production nearing 380 million tons annually. A considerable portion enters municipal solid waste and landfills, creating long-term environmental concerns. Scaling recycling operations requires automated sorting technologies, with spectroscopy and machine learning offering promising solutions. In this study, a six-class convolutional neural network (CNN) was developed for plastic identification using vibrational spectroscopies. Raman Scattering (RS) spectra collected from recycling samples enabled accurate chemical differentiation while assessing the influence of visible features such as color. A CNN trained on RS data achieved 100% classification accuracy. To strengthen field applicability, Attenuated Total Reflectance–Fourier Transform Infrared (ATR-FTIR) spectroscopy was incorporated, achieving 95% accuracy with a similar CNN model. These findings demonstrate the potential of integrating spectroscopy with deep learning for reliable plastic classification, advancing development of scalable, field-ready recycling technologies.

Garcia Tovar, Maria P.↗
DOE OSTI2024

Visibility-enhanced model-free deep reinforcement learning algorithm for voltage control in realistic distribution systems using smart inverters

Increasing integration of distributed solar photovoltaic (PV) into distribution networks could result in adverse effects on grid operation. Traditional model-based control algorithms require accurate model information that is difficult to acquire and thus are challenging to implement in practice. Here, this paper proposes a surrogate model-enabled grid visibility scheme to empower deep reinforcement learning (DRL) approach for distribution network voltage regulation using PV inverters with minimal system knowledge. In contrast to existing DRL methods, this paper presents and corroborates the adverse impact of missing load information on DRL performance and, based on this finding, proposes a surrogate model methodology to impute load information utilizing observable data. Additionally, a multi-fidelity neural network is utilized to construct the DRL training environment, chosen for its efficient data utilization and enhanced robustness to data uncertainty. The feasibility and effectiveness of the proposed algorithm are assessed by considering DRL testing across varying degrees of observable load information and diverse training environments on a realistic power system.

14 SOLAR ENERGY↗
NASA NTRS2022

Radiation Tolerance and Mitigation for Neuromorphic Processors

Neuromorphic processors are designed to execute Deep Neural Networks (DNNs) at very high speed using only a fraction of the electrical power needed to run a DNN on a traditional CPU or GPU. This unique capability makes Neuromorphic processors a prime candidate for space systems, where advanced computational tasks like image analysis, depth map reconstruction, or rover control need to be executed in a power-starved environment. In contrast to the growing number of applications of Neuromorphic processors in smart phones, the automotive and robotics domain, the space environment is unforgiving because of extreme temperatures and high levels of radiation. Any space system, operating beyond LEO requires computing hardware that is resilient against radiation effects. However, Neuromorphic processors have not yet been designed or tested for their radiation tolerance. In this report, we consider traditional methods of detection of radiation events and mitigation via redundancy and gauge their effectiveness on DNNs. In contrast to traditional flight software, however, neural networks represent a statistical algorithm, which might affect its resilience against radiation events. We will focus on the analysis of the tolerance of DNNs with respect to radiation events and discuss techniques to detect radiation hits using on-chip triple modular redundancy (TMR) on an Intel Loihi neuromorphic processor and to mitigate radiation damage. We describe an architecture for on-chip TMR for the Intel Loihi and present results of initial experiments.

Neural Networks↗