A game-theoretic framework for classifier ensembles using weighted majority voting with local accuracy estimates
This work addresses the classifier combination problem for machine learning practitioners, but it appears incremental as it builds on existing weighted majority voting methods.
The paper tackled the problem of optimally combining binary classifiers by proposing a game-theoretic framework using weighted majority voting with local accuracy estimates, and experiments showed that this adaptive model outperformed common combination rules on benchmark datasets.
In this paper, a novel approach for the optimal combination of binary classifiers is proposed. The classifier combination problem is approached from a Game Theory perspective. The proposed framework of adapted weighted majority rules (WMR) is tested against common rank-based, Bayesian and simple majority models, as well as two soft-output averaging rules. Experiments with ensembles of Support Vector Machines (SVM), Ordinary Binary Tree Classifiers (OBTC) and weighted k-nearest-neighbor (w/k-NN) models on benchmark datasets indicate that this new adaptive WMR model, employing local accuracy estimators and the analytically computed optimal weights outperform all the other simple combination rules.