Scaling properties of neural classifiers

Abstract

One of the problems in evaluating the usefulness of various classification techniques for real-world applications is the lack of information about their scaling properties: how will the complexity of a given method depend on the number of attributes, classes and cases? In this paper a simple classification task is presented as a challenge for different methods, allowing to determine their scaling properties as well as to evaluate the complexity of the solutions obtained by various methods.

3 Figures and Tables

Cite this paper

@inproceedings{Duch1997ScalingPO, title={Scaling properties of neural classifiers}, author={Włodzisław Duch}, year={1997} }