Neural networks using a logistics sigmoid function: linear classifier bounds and global nonattainability^∗

^∗The research reported here was sponsored by the Office of Naval Research under Contract N00014-89-5-1537

Abstract

Simple examples clearly demonstrate that highly consistent data lead to solution nonattainability, in neural networks utilizing a logistics sigmoid function. Solution attainability requires a high degree of inconsistency. Bounds are obtained on the optimal value of the mean-square error of a one-layer neural network, in terms of the minimum number of misclassifications obtained from three linear classification problems, and conditions are given that imply solution attainability and nonattainability

Keywords:

• Neural Network
• Mean-Square Error
• Solution Attainability
• Optimal Value Bounds

^∗The research reported here was sponsored by the Office of Naval Research under Contract N00014-89-5-1537

^∗The research reported here was sponsored by the Office of Naval Research under Contract N00014-89-5-1537

Notes

^∗The research reported here was sponsored by the Office of Naval Research under Contract N00014-89-5-1537