Abstract—An important issue in the design of a neural network is the sensitivity of its output to input, and node fault. In this paper, new sensitive measures are proposed for node fault, specifically node stuck-at-zero fault. Correlation coefficient between empirical mean squared error and error due to proposed metric shows that the proposed metrics are significant metrics due to their statistical significance at 95% confidence level for node stuck-at-zero fault.
Index Terms—Sigmoidal feedforward neural network, node fault, fault metric, fault tolerance.
The authors are with the University School of Information & Communication Technology, Guru Gobind Singh Indraprastha University, Dwarka, Delhi, India (e-mail: aps.ipu@gmail.com; csrai_ipu@yahoo.com; chandra.pravin@gmail.com).
Cite: Amit P. Singh, C. S. Rai, and Pravin Chandra, "Metrics for Measurement of Node Fault in Sigmoidal FFANNs," International Journal of Modeling and Optimization vol. 3, no. 1, pp. 87-91, 2013.
Copyright © 2008-2025. International Journal of Modeling and Optimization. All rights reserved.
E-mail: ijmo@iacsitp.com