We propose a novel boosting algorithm which improves on current algorithms for weighted voting classification by striking a better balance between classification accuracy and the ...
Semi-supervised learning (SSL), is classification where additional unlabeled data can be used to improve accuracy. Generative approaches are appealing in this situation, as a mode...
Information available in the Internet is frequently supplied simply as plain ascii text, structured according to orthographic and semantic conventions. Traditional document classi...
A novel training method has been proposed for increasing efficiency and generalization of support vector machine (SVM). The efficiency of SVM in classification is directly determi...
This work describes an approach for the interpretation and explanation of human behavior in image sequences, within the context of a Cognitive Vision System. The information source...