Summary
International Symposium on Nonlinear Theory and its Applications
2010
Session Number:A2L-D
Session:
Number:A2L-D4
Effect of Piecewise Linear Function on Maximum-Flow Neural Network
Masatoshi Sato, HisashiAomori, Mamoru Tanaka,
pp.131-134
Publication Date:2010/9/5
Online ISSN:2188-5079
DOI:10.34385/proc.44.A2L-D4
PDF download (114.2KB)
Summary:
In our previous research, the Maximum- Flow Neural Network (MF-NN) was proposed, and we showed that theMF-NN is possible to solve any maximumflow problems. Each neuron of theMF-NN is connected by nonlinear resistances with the saturation saturation property. However, the conventional MF-NN using the sigmoidal function has problems where the sigmoidal function dose not converge to 0, 1 that is saturation value. In this research, the saturation property is improved by using the picewise linear function. Moreover, this novel method is possible to considerably reduce a calculation cost.