A Fault Aware Broad Learning System for Concurrent Network Failure Situations

被引:7
作者
Adegoke, Muideen [1 ]
Wong, Hiu Tung [1 ]
Leung, Chi Sing [1 ]
机构
[1] City Univ Hong Kong, Dept Elect Engn, Hong Kong, Peoples R China
关键词
Training; Fault tolerant systems; Linear programming; Fault tolerance; Training data; Learning systems; Simulation; broad learning system; regression; multiplicative noise; open fault; incremental learning; NEURAL-NETWORKS; MULTILAYER PERCEPTRONS; NOISE-IMMUNITY; TOLERANCE; WEIGHT;
D O I
10.1109/ACCESS.2021.3066217
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
The broad learning system (BLS) framework gives an efficient solution for training flat-structured feedforward networks and flat structured deep neural networks. However, the classical BLS model and other variants focus on the faultless situation only, where enhancement nodes, feature mapped nodes, and output weights of a BLS network are assumed to be realized in a perfect condition. When a trained BLS network suffers from coexistence of weight/node failures, the trained network has a greatly degradation in its performance if a countermeasure is not taken. In order to reduce the effect of weight/node failures on the BLS network's performance, this paper proposes an objective function for enhancing the fault aware performance of BLS networks. The objective function contains a fault aware regularizer term which handles the weight/node failures. A learning algorithm is then derived based on the objective function. The simulation results show that the performance of the proposed fault aware BLS (FABLS) algorithm is superior to the classical BLS and two state-of-the-arts BLS algorithms, namely correntropy criterion BLS (CBLS) and weighted BLS (WBLS).
引用
收藏
页码:46129 / 46142
页数:14
相关论文
共 66 条
[1]  
Adegoke Muideen, 2019, J AMB INTEL HUM COMP, P1
[2]   A Low-Cost Fault-Tolerant Approach for Hardware Implementation of Artificial Neural Networks [J].
Ahmadi, A. ;
Sargolzaie, M. H. ;
Fakhraie, S. M. ;
Lucas, C. ;
Vakili, Sh. .
2009 INTERNATIONAL CONFERENCE ON COMPUTER ENGINEERING AND TECHNOLOGY, VOL II, PROCEEDINGS, 2009, :93-97
[3]  
[Anonymous], 2006, Code Design for Dependable Systems: Theory and Practical Applications
[4]   Hardware implementation of a fault-tolerant Hopfield Neural Network on FPGAs [J].
Antonio Clemente, Juan ;
Mansour, Wassim ;
Ayoubi, Rafic ;
Serrano, Felipe ;
Mecha, Hortensia ;
Ziade, Haissam ;
El Falou, Wassim ;
Velazco, Raoul .
NEUROCOMPUTING, 2016, 171 :1606-1609
[5]  
Avizienis A., 1983, 10th Annual International Conference on Computer Architecture Conference Proceedings, P16, DOI 10.1145/800046.801633
[6]  
Bautista-Gomez L, 2016, SC '16: PROCEEDINGS OF THE INTERNATIONAL CONFERENCE FOR HIGH PERFORMANCE COMPUTING, NETWORKING, STORAGE AND ANALYSIS, P645, DOI 10.1109/SC.2016.54
[7]   Improving the tolerance of multilayer perceptrons by minimizing the statistical sensitivity to weight deviations [J].
Bernier, JL ;
Ortega, J ;
Rojas, I ;
Prieto, A .
NEUROCOMPUTING, 2000, 31 (1-4) :87-103
[8]   A quantitative study of fault tolerance, noise immunity, and generalization ability of MLPs [J].
Bernier, JL ;
Ortega, J ;
Ros, E ;
Rojas, I ;
Prieto, A .
NEURAL COMPUTATION, 2000, 12 (12) :2941-2964
[9]   Assessing the noise immunity and generalization of radial basis function networks [J].
Bernier, JL ;
Díaz, AF ;
Fernández, FJ ;
Cañas, A ;
González, J ;
Martín-Smith, P ;
Ortega, J .
NEURAL PROCESSING LETTERS, 2003, 18 (01) :35-48
[10]   Obtaining fault tolerant multilayer perceptrons using an explicit regularization [J].
Bernier, JL ;
Ortega, J ;
Rojas, I ;
Ros, E ;
Prieto, A .
NEURAL PROCESSING LETTERS, 2000, 12 (02) :107-113