Seminar on Computational Learning and Adaptation


  Decision Tree Grafting

Geoff Webb
School of Computing and Mathematics
Deakin University
Geelong, Victoria 3217, Australia
http://www.cm.deakin.edu.au/webb/

Decision tree grafting is a machine learning technique that adds tests and nodes to existing decision trees, resulting in bagging-like variance reduction while providing a single directly interpretable decision tree. Experiments demonstrate that decision tree grafting consistently reduces prediction error over a wide range of learning tasks. This talk presents the motivations for decision tree grafting, two algorithms, an analysis of the relationship between pruning and grafting, and experimental results.



Date: Thurs., Apr 26

Time: 4:15-5:30PM

Place: Ventura 17


Return to the seminar schedule