Clustering data manipulation method for ensembles

Authors:
Matthew Spencer;John McCullagh;Tim Whitfort
Affiliations:
Department of Computer Science & Computer Engineering, La Trobe University, Bendigo, Victoria, Australia;Department of Computer Science & Computer Engineering, La Trobe University, Bendigo, Victoria, Australia;Department of Computer Science & Computer Engineering, La Trobe University, Bendigo, Victoria, Australia
Venue:
AI'06 Proceedings of the 19th Australian joint conference on Artificial Intelligence: advances in Artificial Intelligence
Year:
2006

Citing 3
Cited 1

Bagging predictors

Machine Learning
Dynamic ensemble approach for estimating organic carbon using computational intelligence

ACST'06 Proceedings of the 2nd IASTED international conference on Advances in computer science and technology
Adaptive mixtures of local experts

Neural Computation

Clustering data manipulation methods for the development of local specialists

AIAP'07 Proceedings of the 25th conference on Proceedings of the 25th IASTED International Multi-Conference: artificial intelligence and applications

Quantified Score

Hi-index	0.00

Visualization

Abstract

Traditional data manipulation models such as Bagging and Boosting select training cases from throughout the problem space to generate diversity and improve performance. A new data manipulation model is proposed that dynamically assigns specialists to train on difficult clusters of training data. The model allows the expertise of specialists to overlap for difficult regions of the problem. It has been coupled with a dynamic combination model to exploit the diversity of specialist members. The model has been applied to an environmental problem and has demonstrated that dynamic modelling can enhance both the diversity of members and the accuracy of the ensemble.