Title :
Multistage classification by cascaded classifiers
Author :
Kaynak, Cenk ; Alpaydin, Ethem
Author_Institution :
Dept. of Comput. Eng., Bogazici Univ., Istanbul, Turkey
Abstract :
We propose a new method of classification built as a cascade of a distributed learner and a local learner. The distributed learner generalizes to learn the “rule” and the local learner learns the “exceptions” not covered by the “rule”. We show how such a system can be trained using cross-validation. We use a multilayer perceptron with sigmoidal hidden units as the rule-learner and a k-nearest neighbor classifier as the exception-learner. Cascading is a better approach than voting where multiple learners are used for all cases; the extra computation and memory required for the second learner is unnecessary if we are sufficiently sure that the first one´s response is correct. The cascade algorithm significantly outperforms the individual methods and voting on three optical and pen-based handwritten digit recognition tasks when comparison is based on three criteria; generalization success, learning speed, and number of free parameters
Keywords :
multilayer perceptrons; pattern classification; cascaded classifiers; cross-validation; distributed learner; exception-learner; exceptions; generalization success; handwritten digit recognition tasks; k-nearest neighbor classifier; learning speed; local learner; multilayer perceptron; multistage classification; rule-learner; sigmoidal hidden units; Data acquisition; Distributed computing; Kernel; Multilayer perceptrons; NIST; Neural networks; Optical computing; Optical network units; Pattern recognition; Voting;
Conference_Titel :
Intelligent Control, 1997. Proceedings of the 1997 IEEE International Symposium on
Conference_Location :
Istanbul
Print_ISBN :
0-7803-4116-3
DOI :
10.1109/ISIC.1997.626420