Learning Algorithms for Nonparametric Solution to the Minimum Error Classification Problem

1 July 1978

journal article
Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computers

Vol. C-27 (7) , 648-659
https://doi.org/10.1109/tc.1978.1675165

Abstract

This paper discusses the two class classification problem using discriminant function solution that minimizes the probability of classification error. Learning algorithms using window function techniques are presented. The convergence rates are estimated and a particular strategy is proposed. Within this strategy it is recommended to use a triangular window function. The proposed algorithms are tested on several artificial pattern classification problems and their efficiency is proven. A comparison with the mean-square-error algorithm is also presented.

Keywords

This publication has 6 references indexed in Scilit:

Discriminant function implementation of a minimum risk classifier
Biological Cybernetics, 1975
Training a One-Dimensional Classifier to Minimize the Probability of Error
IEEE Transactions on Systems, Man, and Cybernetics, 1972
Design of Pattern Classifiers with the Updating Property Using Stochastic Approximation Techniques
IEEE Transactions on Computers, 1968
A Theory of Adaptive Pattern Classifiers
IEEE Transactions on Electronic Computers, 1967
Generation of Polynomial Discriminant Functions for Pattern Recognition
IEEE Transactions on Electronic Computers, 1967
Linear Decision Functions, with Application to Pattern Recognition
Proceedings of the IRE, 1962