Recent Developnent of Globally Trained Tree Classifier

Globally trained tree (GTT) classifier was proposed to overcome major difficulties in large set, pattern recogniton problem. In the present work, we have developed multiple GTT classifier, which consists of a couple of sub GTTs. Each sub GTT recognizes patterns from a subset of the given classess, and refuses patterns from other classes. The multiple GTT makes use of a priori statistical knowledge and achieves a better performance than the single GTT classifier.