Publication View

Polynomial Bounds for VC Dimension of Sigmoidal Neural Networks (1995)

Abstract
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 oe(y) = 1=1 + e \Gammay to be bounded by a quadratic polynomial in the number of programmable parameters. 1 Dept. of Computer Science, University of Bonn, 53117 Bonn, and the International Computer Science Institute, Berkeley, California. Research partially supported by the International Computer Science Institute, Berkeley, by the DFG Grant KA 673/4-1, and by the ESPRIT BR Grants 7097 and ECUS 030. Email: marek@cs.uni-bonn.de 2 Mathematical Institute, University of Oxford, Oxford OX1 3LB. Research supported in part by a Senior Research Fellowship of the SERC. Email: ajm@maths.ox.ac.uk Introduction The most commonly used activation function in various neural networks applications is the sigmoid oe(y) = 1=1+ e \Gammay (cf. [HKP91]). I...

Publication details
Download http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.56.7756
Source ftp://ftp.icsi.berkeley.edu/pub/techreports/1995/tr-95-001.ps.gz
Contributors CiteSeerX
Repository CiteSeerX - Scientific Literature Digital Library and Search Engine (United States)
Type text
Language English
Relation 10.1.1.49.2780, 10.1.1.29.3121, 10.1.1.63.4487, 10.1.1.12.4910, 10.1.1.27.2560, 10.1.1.47.6086, 10.1.1.56.8830, 10.1.1.16.8020, 10.1.1.53.126, 10.1.1.53.5221, 10.1.1.43.9916, 10.1.1.51.1761, 10.1.1.20.3252, 10.1.1.39.9814, 10.1.1.132.2764, 10.1.1.53.8861, 10.1.1.103.7655, 10.1.1.48.1517, 10.1.1.50.3917, 10.1.1.64.5147, 10.1.1.66.4531, 10.1.1.84.1968, 10.1.1.117.1639, 10.1.1.133.7790, 10.1.1.135.3099, 10.1.1.45.6291, 10.1.1.12.6378, 10.1.1.12.7859, 10.1.1.8.562