Title :
High speed networks that preserve continuity and accuracy
Author :
Armstrong, William W.
Author_Institution :
Dendronic Decisions Ltd., Edmonton, Alta., Canada
Abstract :
Developments in classification and regression like bagging, boosting and support vector machines tend to greatly improve generalization over simpler techniques, but may also result in longer computation times. In this paper we show one way of converting such computations into fast ones, without significant loss of accuracy, using a decision tree with piecewise linear approximants on the blocks
Keywords :
computational complexity; decision trees; learning automata; neural nets; pattern classification; piecewise linear techniques; statistical analysis; SVM; bagging; boosting; classification; continuity; decision tree; generalization; high-speed networks; piecewise linear approximants; regression; support vector machines; Bagging; Boosting; Classification tree analysis; Decision trees; High-speed networks; Piecewise linear approximation; Piecewise linear techniques; Support vector machine classification; Support vector machines; Testing;
Conference_Titel :
Neural Networks, 2001. Proceedings. IJCNN '01. International Joint Conference on
Conference_Location :
Washington, DC
Print_ISBN :
0-7803-7044-9
DOI :
10.1109/IJCNN.2001.938765
