柯西不等式.pdf

Journal of Intelligent Learning Systems and Applications, 2011, 3, 242-248
doi:.34027 Published Online November 2011 (rnal/jilsa)
A New Weight Initialization Method Using
Cauchy’s Inequality Based on Sensitivity Analysis
Thangairulappan Kathirvalavakumar1, Subramanian Jeyaseeli Subavathi2

1Department puter Science, . College, Virudhunagar, India; 2Department of Information Technology, Sri Kaliswari
College, Sivakasi, India.
Email: kathirvalavakumar@, ******@

Received June 7th, 2011; revised July 7th, 2011; accepted July 30th, 2011.

ABSTRACT
In this paper, an efficient weight initialization method is proposed using Cauchy’s inequality based on sensitivity analy-
sis to improve the convergence speed in single hidden layer feedforward works. The proposed method ensures
that the outputs of hidden neurons are in the active region which increases the rate of convergence. Also the weights
are learned by minimizing the sum of squared errors and obtained by solving linear system of equations. The proposed
method is simulated on various problems. In all the problems the number of epochs and time required for the proposed
method is found to be pared with other weight initialization methods.

Keywords: Weight Initialization, Backpropagation, Feedforward work, Cauchy’s Inequality, Linear System
of Equati