A NEW BACK-PROPAGATION ALGORITHM WITH COUPLED NEURON

被引：13

作者：

FUKUMI, M

OMATU, S

机构：

[1] Department of Information Science and Intelligent Systems, Faculty of Engineering, University of Tokushima

来源：

IEEE TRANSACTIONS ON NEURAL NETWORKS | 1991年 / 2卷 / 05期

关键词：

D O I：

10.1109/72.134292

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

In this letter a new neuron model and its learning algorithm are presented. They provide a new approach for speeding up convergence in the learning of layered neural networks and for training networks of neurons with a nondifferentiable output function by using the gradient descent method. The neuron is called a saturating linear coupled neuron (sl-CONE). From simulation results, it is shown that the sl-CONE has a high convergence rate in learning compared with the conventional back-propagation algorithm.

引用

页码：535 / 538

页数：4

共 10 条

[1]

ANDES D, 1990, P INT JOINT C NEURAL, V1, P533

[2]

BABA N, 1990, P INT JOINT C NEURAL, V1, P585

[3]

Fukumi M., 1990, Transactions of the Institute of Electronics, Information and Communication Engineers D-II, VJ73D-II, P648

[4]

FUKUMI M, 1989, 28TH P SICE ANN C, P1327

[5]

FUKUMI M, 1989, P INT JOINT C NEURAL, V2, P611

[6] NEOCOGNITRON - A HIERARCHICAL NEURAL NETWORK CAPABLE OF VISUAL-PATTERN RECOGNITION [J].

FUKUSHIMA, K .

NEURAL NETWORKS, 1988, 1 (02) :119-130

[7]

HOSOKAWA M, 1989, P INT JOINT C NEURAL, V1, P183

[8]

Rumelhart DE, 1986, ENCY DATABASE SYST, P45

[9] GENERALIZATION OF BACKPROPAGATION WITH APPLICATION TO A RECURRENT GAS MARKET MODEL [J].

WERBOS, PJ .

NEURAL NETWORKS, 1988, 1 (04) :339-356

[10] LAYERED NEURAL NETS FOR PATTERN-RECOGNITION [J].

WIDROW, B ;

WINTER, RG ;

BAXTER, RA .

IEEE TRANSACTIONS ON ACOUSTICS SPEECH AND SIGNAL PROCESSING, 1988, 36 (07) :1109-1118

← 1 →