Weight estimation for the learning of modular perceptron networks

Author

Lee, Y.P. ; Fu, Hsin-Chia

Author_Institution

Dept. of Comput. Sci. & Inf. Eng., Nat. Chiao Tung Univ., Hsinchu, Taiwan

fYear

1999

fDate

36373

Firstpage

103

Lastpage

111

Abstract

We propose a weight estimation method for feedforward neural networks. The proposed method includes two steps: (1) weight vector orientation estimation and (2) weight vector length estimation, such that the iteration learning of modular perceptron networks (MPN) can be reduced. We have applied the proposed method to the divide-and-conquer learning (DCL) for two-spiral problems (TPS) on an MPN. The experimental results show that the number of pattern presentations can be reduced 79.05% (from 190,980 to 40,012), and the number of subnets in MPN can also be reduced 20.38% (from 26.5 to 21.1)

Keywords

Bayes methods; Gaussian distribution; decision theory; feedforward neural nets; generalisation (artificial intelligence); learning (artificial intelligence); pattern recognition; divide-and-conquer learning; iteration learning; modular perceptron networks; pattern presentations; two-spiral problems; weight estimation; weight vector length estimation; weight vector orientation estimation; Artificial neural networks; Computer science; Convergence; Feedforward neural networks; Feedforward systems; Multi-layer neural network; Neural networks; Pattern recognition; Supervised learning; Unsupervised learning;

fLanguage

English

Publisher

ieee

Conference_Titel

Neural Networks for Signal Processing IX, 1999. Proceedings of the 1999 IEEE Signal Processing Society Workshop.

Conference_Location

Madison, WI

Print_ISBN

0-7803-5673-X

Type

conf

DOI

10.1109/NNSP.1999.788128

Filename

788128