Constructing Error Correcting Output Coding Classifiers based on Weighted Least Squares

N. Yamaguchi and N. Ishii (Japan)


ECOC, Error Correcting Codes, Multiple Classifier Systems, Neural Networks


The object of this paper is to develop the Error Correcting Output Coding(ECOC) classifier. The ECOC classifier is one of Multiple Classifier Systems(MCSs), which pre pare a large number of discriminant functions and then construct a classifier by combining these discriminant functions. We focus on a method for combining the dis criminant functions and develop an extension of ECOC. Experiments on artificial and real data demonstrate that a much better performance can be obtained by the ex tended ECOC classifier.

Important Links:

Go Back