Recent Research Projects
This page lists some of the AI research projects that I supervised in
the recent past. Information on my current research can be found on the web
page listing the current AI projects.
Massive Search with Bagging
Siu K Lee
Bagging is a method of a voting classification algorithm, which has been successfully proved by many emppirical tests to significantly improve the classification error over artificial and real world datasets. However Breiman suggested the effectiveness of bagging depends on the instability of the learning algorithm. Decision lists and neural networks have been shown to be unstable algorithms. Many researchers have shown that bagged C4.5 has a dramatic reduction on classification error.
A rule learning system can generate all possible solutions represented by a set of rules to an existing database. With the power of massive search we could extract the best possible rules from the search space that maximise the evaluation function. An alternative approach to generating a classification model using a rule learning system is to assemble the rules into a classification algorithm.