Institut für Informatik
 
Abteilung V

 
Universität Bonn -> Institut für Informatik -> Abteilung V
CS-APX-Reports 1994 Copyright 1994 Universität Bonn, Institut für Informatik, Abt. V
8920

Polynomial Bounds for VC Dimension of Sigmoidal Neural Networks
Marek Karpinski, Angus Macintyre
[Download PostScript] [Download PDF]

We introduce a new method for proving explicit upper bounds on the VC Dimension of general functional basis networks, and prove as an application, for the first time, the VC Dimension of analog neural networks with the sigmoid activation function $\sigma(y)=1/1+e^{-y}$ to be bounded by a quadratic polynomial in the number of programmable parameters.

Last Change: 11/05/14 at 09:51:10
 English
Universität Bonn -> Institut für Informatik -> Abteilung V