作者: Gao Daqi , Zhu Shangming
DOI: 10.1109/FUZZY.2005.1452431
关键词:
摘要: In order to use combinative classifiers effectively solve large-scale learning problems, this paper focuses on the following aspects. (A) Decomposition of problems. (B) Selection units classifiers. (C) Transformation outputs single into grades membership. We select improved kernel Fisher, Mahalanobis distance, and 10-nearest-neighbor classifier, as units, only let most relative part original datasets take in training a then transform each classifier same The experiment for letter recognition shows that proposed method is effective