Filters








40,558 Hits in 5.5 sec

A law of robustness for two-layers neural networks [article]

Sébastien Bubeck and Yuanzhi Li and Dheeraj Nagaraj
2020 arXiv   pre-print
We make a precise conjecture that, for any Lipschitz activation function and for most datasets, any two-layers neural network with k neurons that perfectly fit the data must have its Lipschitz constant  ...  We initiate the study of the inherent tradeoffs between the size of a neural network and its robustness, as measured by its Lipschitz constant.  ...  Our proposed law of robustness is a first mathematical formalization of the broader phenomenon that "overparametrization in neural networks is necessary for robustness".  ... 
arXiv:2009.14444v2 fatcat:iqcj6heukjbcpmsjfjop7zbipu

On 1/n neural representation and robustness [article]

Josue Nassar, Piotr Aleksander Sokol, SueYeon Chung, Kenneth D. Harris, Il Memming Park
2020 arXiv   pre-print
We empirically investigate the benefits such a neural code confers in neural networks, and illuminate its role in multi-layer architectures.  ...  A pressing question in these areas is understanding how the structure of the representation used by neural networks affects both their generalization, and robustness to perturbations.  ...  Acknowledgments and Disclosure of Funding  ... 
arXiv:2012.04729v1 fatcat:64n46oqxdzg63ouzhkmiscdx2u

Different Spectral Representations in Optimized Artificial Neural Networks and Brains [article]

Richard C. Gerum, Cassidy Pirlot, Alona Fyshe, Joel Zylberberg
2022 arXiv   pre-print
Recent studies suggest that artificial neural networks (ANNs) that match the spectral properties of the mammalian visual cortex – namely, the ∼ 1/n eigenspectrum of the covariance matrix of neural activities  ...  For convolutional networks, the best α values depend on the task complexity and evaluation metric: lower α values optimized validation accuracy and robustness to adversarial attack for networks performing  ...  with a power-law fit (data from Stringer et al. (2019)). e, MNIST digit, presented to a dense artificial neural network f.  ... 
arXiv:2208.10576v1 fatcat:f4akhep4zfeenmhbjhflfspv24

Fuzzy Identification Using Fuzzy Neural Networks With Stable Learning Algorithms

W. Yu, X. Li
2004 IEEE transactions on fuzzy systems  
This paper suggests new learning laws for Mamdani and Takagi-Sugeno-Kang type fuzzy neural networks based on input-to-state stability approach.  ...  Stable learning algorithms for the premise and the consequence parts of fuzzy rules are proposed.  ...  We use the same multilayer neural networks as [18] , it is (two hidden layers with 20 and ten nodes), and a fixed learning rate .  ... 
doi:10.1109/tfuzz.2004.825067 fatcat:3uqhafvucrb5rlubiodihovdi4

Robust Neural Network Tracking Controller Using Simultaneous Perturbation Stochastic Approximation

Qing Song, J.C. Spall, Yeng Chai Soh, Jie Ni
2008 IEEE Transactions on Neural Networks  
We introduce the conic sector theory to establish a robust neural control system, with guaranteed boundedness for both the input/output (I/O) signals and the weights of the neural network.  ...  This paper considers the design of robust neural network tracking controllers for nonlinear systems. The neural network is used in the closed-loop system to estimate the nonlinear system function.  ...  Unlike the robust conic sector analysis for a pretrained neural network [16] , we provide an online scheme for the robustness analysis of the neural control system.  ... 
doi:10.1109/tnn.2007.912315 pmid:18467211 fatcat:6lhrbegek5fmtj52u5q35kg7d4

Neural Network-based Sliding Mode Control for Permanent Magnet Synchronous Motor

Kaiqi Huang, Shilin Zuo
2015 The Open Mechanical Engineering Journal  
In this paper, a scheme of designing neural network-based sliding mode controller is proposed for dealing with the chattering phenomenon existing in conventional sliding mode controller because of its  ...  Firstly, the sliding mode control law is designed using equivalent control technology. Then the neural network and adaptive control are uesd to delete the chattering of sliding mode controller.  ...  SLIDING MODE CONTROLLER DESIGN Fractional sliding mode controller design for a two-step: switching manifold face selection and control law design.  ... 
doi:10.2174/1874155x01509010314 fatcat:ao5b4c6o4rbyjhrmhngmbaphwm

Neural Network-based Sliding Mode Control for Permanent Magnet Synchronous Motor

Kaiqi Huang, Shilin Zuo
2015 Open Electrical & Electronic Engineering Journal  
In this paper, a scheme of designing neural network-based sliding mode controller is proposed for dealing with the chattering phenomenon existing in conventional sliding mode controller because of its  ...  Firstly, the sliding mode control law is designed using equivalent control technology. Then the neural network and adaptive control are uesd to delete the chattering of sliding mode controller.  ...  SLIDING MODE CONTROLLER DESIGN Fractional sliding mode controller design for a two-step: switching manifold face selection and control law design.  ... 
doi:10.2174/1874129001509010314 fatcat:4ap246z37fgejou4gu4mk2slna

Robust Hybrid Control Using Recurrent Wavelet- Neural-Network Sliding-Mode Controller for Two- Axis Motion Control System

Fayez F. M. El-Sousy
2021 International journal of neural networks and advanced applications  
In this paper, a robust hybrid control system (RHCS) for achieving high precision motion tracking performance of a two-axis motion control system is proposed.  ...  obtained as well for the two-axis motion control system.  ...  ACKNOWLEDGMENT The author would like to acknowledge the financial support of the Deanship of Scientific Research at Salman bin Abdulaziz University, Saudi Arabia through its grant no. 29/T/33.  ... 
doi:10.46300/91016.2021.8.6 fatcat:yhviu75f6zb5zgfx2ojvixvbve

A Universal Law of Robustness via Isoperimetry [article]

Sébastien Bubeck, Mark Sellke
2021 arXiv   pre-print
In the case of two-layers neural networks and Gaussian covariates, this law was conjectured in prior work by Bubeck, Li and Nagaraj.  ...  We prove this universal law of robustness for any smoothly parametrized function class with polynomial size weights, and any covariate distribution verifying isoperimetry.  ...  A finite approach to the law of robustness For the function class of two-layer neural networks, [BLN21] investigated several approaches to prove the law of robustness.  ... 
arXiv:2105.12806v3 fatcat:jmeikybuung7tpnqzukaufisdu

Intelligent second-order sliding-mode control for chaotic tracking problem

Chun-Fei Hsu, Tsu-Tian Lee, Chun-Wei Chang
2014 2014 Proceedings of the SICE Annual Conference (SICE)  
A neural controller and a robust compensator are designed in the proposed ISSMC system.  ...  In this paper, a recurrent fuzzy neural network (RFNN) is used to online approximate the unknown nonlinear term of chaotic system dynamics with a good accuracy.  ...  ACKNOWLEDGMENT The authors appreciate the partial financial support from the National Science Council of Republic of China under grant NSC 102-2221-E-032-052.  ... 
doi:10.1109/sice.2014.6935181 fatcat:kvafvnvrxbeg7g2zq2ymwvqzny

Multiple Evaluation Models for Education Based on Artificial Neural Networks

Qianyin Xiao, Bo Liu
2015 International Journal of Hybrid Information Technology  
Using artificial neural networks (ANNs) can ensure the accuracy and fairness of the evaluation works.  ...  To address this problem, we used machine learning techniques to develop two groups of models for evaluating teachers and students' performances respectively.  ...  It is a layer existing between the input and output layer. The more the number of the neurons in the hidden layer has, the greater robustness of the artificial neural network is.  ... 
doi:10.14257/ijhit.2015.8.9.01 fatcat:5lv75y3jozewnckdxgjlm6ijxq

Adaptive Fuzzy Control Based on Fuzzy Neural Network for Uncertain Nonlinear Systems

Y.J. Huang, T.C. Kuo
2008 IFAC Proceedings Volumes  
A fuzzy neural network is applied to on-line identify the control system and provide sufficient information of the adaptive laws for the proposed fuzzy controller.  ...  In this paper, an adaptive fuzzy controller based on fuzzy neural network is proposed for uncertain nonlinear systems.  ...  The authors would like to thank the National Science Council of Taiwan, for financially supporting this work under Grants NSC95-2221-E-231-013 and NSC95-2622-E-155-008-CC3.  ... 
doi:10.3182/20080706-5-kr-1001.01339 fatcat:2lnc36dtozcs3m72bmvdrdyane

Robust Adaptive Tracking Control of Manipulator Arms with Fuzzy Neural Networks

M. Fouzia, N. Khenfer, N. E. Boukezzoula
2020 Zenodo  
The objective of this paper is to design a nonlinear system based on the fuzzy neural network control using supervised training, into executing reference trajectories by a flexible joint manipulator.  ...  A comparative study has been carried out between these two methods in order to prove the effectiveness of the later.  ...  In this work, two network controllers based on fuzzy neural network control using supervised training are proposed, where each network will control one joint of the arm manipulator with two DoF.  ... 
doi:10.5281/zenodo.4016285 fatcat:frwikl4skfcmjgncdl436vwuwm

Adaptive Neural Network Sliding Mode Control for Quad Tilt Rotor Aircraft

Yanchao Yin, Hongwei Niu, Xiaobao Liu
2017 Complexity  
A novel neural network sliding mode control based on multicommunity bidirectional drive collaborative search algorithm (M-CBDCS) is proposed to design a flight controller for performing the attitude tracking  ...  law, and the novel M-CBDCS algorithm is developed to uniformly update the unknown neural network weights and essential model parameters adaptively.  ...  Acknowledgments This paper was supported by the National Natural Science Foundation of China (51365022) and the Science Research Foundation Project of Education Department of Yunnan Province (2016YJS022  ... 
doi:10.1155/2017/7104708 fatcat:sa6ynxhe6bdcplqw7gshzbv73i

Fractional-Order Iterative Sliding Mode Control Based on the Neural Network for Manipulator

Xin Zhang, Wenbo Xu, Wenru Lu, Adnan Maqsood
2021 Mathematical Problems in Engineering  
The study is based on the theory of fractional calculus, radial basis function (RBF) neural network control, and iterative sliding mode control, and the RBF neural network fractional-order iterative sliding  ...  law control strategy, and fractional-order iterative sliding mode surface control strategy.  ...  For a feedforward neural network, its operation is relatively simple. e RBF neural network is a three-layer network structure, including an input layer, a hidden layer, and an output layer. e input signal  ... 
doi:10.1155/2021/9996719 fatcat:2iqmo46d3zaj3iu3aqp57tk25u
« Previous Showing results 1 — 15 out of 40,558 results