Mathematical Programming Methods of Pattern Classification

Publisher Website
Google Scholar
Abstract
This survey examines eight mathematical programming models of pattern classification. The paper determines the range of applicability and computational merits of each model. The flexibility and large sample properties of the models are also discussed. We find that six of the models produce decision rules that maximize a function we call the quality of a decision rule. The remaining two models minimize a weighted sum of errors.

This publication has 0 references indexed in Scilit: