On the Reliability of Linear Regression and Pattern Recognition Feedforward Artificial Neural Networks in FPGAs

Libano, F.; Rech, Paolo; Tambara, Lucas Antunes; Tonfat, Jorge; Kastensmidt, Fernanda Lima

doi:10.1109/tns.2017.2784367

Cited by 28 publications

(15 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Few authors have explored the radiation effects on machine learning algorithms as it is still a new field. The intrinsic fault tolerance of an FPGA implementation of Artificial Neural Networks (ANN) is evaluated in [7], in which the authors perform a fault injection campaign along with a heavy ion campaign. The work is complemented in [8], where the same architecture along with an FPGA implementation of a Convolutional Neural Network (CNN), a very popular variant of ANN for image processing applications, have been evaluated under the effect of neutrons.…”

Section: Comparison With State-of-the-art Workmentioning

confidence: 99%

“…In [7,8], the authors have made used of the same dataset that we have used for our Multiclass SVM to train an ANN. Also, they have used the same FPGA platform.…”

Section: Comparison With State-of-the-art Workmentioning

confidence: 99%

“…The effects of radiation induced transients have been studied in multiple hardware implementations of machine learning algorithms. Artificial Neural Networks (ANNs) and Convolutional Neural Networks have been analyzed in [7,8,9], while [10] focused on Bayesian Machines and [11] on Binary SVM. To the best of our knowledge, no work has been conducted on the effects of thermal neutrons on hardware implemented SVMs nor the effects of radiation in Multiclass SVMs.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Effects of thermal neutron radiation on a hardware-implemented machine learning algorithm

Trindade

Benevenuti

Létiche

et al. 2021

Microelectronics Reliability

Self Cite

View full text Add to dashboard Cite

Hardware-implemented machine learning algorithms are finding their way in various domains, including safety-critical applications. This has demanded these algorithms to perform correctly even in harsh environmental conditions, such as in avionics altitudes. Support Vector Machine (SVM) is an important Machine Learning that has been target of hardware implementation in recent years. This is the first work to asses both Binary and Multiclass SVMs under thermal neutron radiation, a type of particle noticeably present in high altitudes. A fault injection campaign along with a radiation test with the D50 thermal neutron source, at the Intitut Laue-Languevin, has been performed. The results show a high intrinsic fault tolerance for both varieties of the SVM algorithm, especially for the Multiclass SVM.

show abstract

Section: Comparison With State-of-the-art Workmentioning

confidence: 99%

“…In [7,8], the authors have made used of the same dataset that we have used for our Multiclass SVM to train an ANN. Also, they have used the same FPGA platform.…”

Section: Comparison With State-of-the-art Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Effects of thermal neutron radiation on a hardware-implemented machine learning algorithm

Trindade

Benevenuti

Létiche

et al. 2021

Microelectronics Reliability

Self Cite

View full text Add to dashboard Cite

show abstract

“…From equation (17), one derives that the dimensionless central moments of the series of curves satisfy the following recurrence relation…”

Section: Basic Theorymentioning

confidence: 99%

“…In the last 30 years, ANN algorithms have been rapidly developed for universal function approximations. Libano et al.. 17 used multilayer feedforward networks to obtain the universal approximation of an unknown mapping and its derivatives. Cardaliaguet and Euvrard 18 applied feedforward neural networks to deal with the approximation of both a function and its derivative in control theory.…”

Section: Introductionmentioning

confidence: 99%

Probabilistic sensitivity analysis for the frame structure of missiles

Yang

Yue

Zhao

2018

Proceedings of the Institution of Mechanical Engineers, Part C:

View full text Add to dashboard Cite

In this study, the high-performance and high-reliability of missiles were achieved under the background of the probabilistic analysis for frame structure of missiles. On the basis of deep investigation of the artificial neural network method and mathematical model of multiorder moment method, artificial neural network based higher order moment approach was proposed for the probabilistic analysis of complex structure. As illustrated in frame structure probabilistic analysis with multiphysics fields based on artificial neural network based higher order moment, the reliability and failure probability of frame structure were studied. Furthermore, it was shown that the radial concentrated force P and the outer radius of the frame R2 are the most important factors, and the inside radius of the frame R1 and the ultimate tensile strength of materials σ b also have an important influence on the frame structure. Through comparison with Monte Carlo simulation, second-order fourth-moment approach and the artificial neural network based first-order second-moment approach, it was demonstrated that artificial neural network based higher order moment reshapes the possibility of complex aero-missile probabilistic analysis and improves computing efficiency while preserving the accuracy. Artificial neural network based higher order moment offers a useful insight for missile reliability design and optimization with multi-discipline. The efforts of this study also enrich the theory and method of mechanical reliability design.

show abstract

Dynamic event‐triggered control for delayed switched neural networks: A merging signal scheme

Wang,

Huang

et al. 2023

Intl J Robust & Nonlinear

View full text Add to dashboard Cite

This paper investigates the asynchronous control of delayed switched neural networks via a dynamic event‐triggering mechanism (DETM) and a merging signal scheme. Firstly, an improved DETM is proposed through adding an exponential decaying term into the triggering condition, which is proved to have a larger lower bound by comparison with the static event‐triggering mechanism (SETM) and the existing DETM. Next, a merging signal is constructed by introducing a time‐varying delay to combine the system switching signal with the controller switching signal, which is convenient to deal with the synchronous and asynchronous switching in a unified framework. Then, the closed‐loop system which integrates the proposed DETM with the merging signal is formulated. Moreover, by constructing multiple time‐dependent Lyapunov functionals, sufficient conditions which ensure the asymptotical stability of the resulting closed‐loop system are derived, and a design algorithm for the feedback controller is designed consequently. Finally, the validity and superiority of the proposed results are demonstrated by numerical examples.

show abstract

On the Reliability of Linear Regression and Pattern Recognition Feedforward Artificial Neural Networks in FPGAs

Cited by 28 publications

References 21 publications

Effects of thermal neutron radiation on a hardware-implemented machine learning algorithm

Effects of thermal neutron radiation on a hardware-implemented machine learning algorithm

Probabilistic sensitivity analysis for the frame structure of missiles

Dynamic event‐triggered control for delayed switched neural networks: A merging signal scheme

Contact Info

Product

Resources

About