RTCRelief-F: an effective clustering and ordering-based ensemble pruning algorithm for facial expression recognition

Abstract

Ensemble pruning is effective for improving the accuracy of expression recognition. This paper proposes a novel ensemble pruning algorithm called RTCRelief-F and applies it to facial expression recognition. RTCRelief-F uses a novel classifier-representation method that accounts for the interaction among classifiers and bases the classifier selection upon not only diversity but accuracy. Additionally, for the first time, RTCRelief-F, applies the Relief-F algorithm to evaluate the classifiers’ ability and resets the fusion order. Finally, the combination of a clustering-based ensemble pruning method and the ordering-based ensemble pruning method can both alleviate the dependence of a selected subset S on the adopted clustering strategies and guarantee the diversity of the selected subset S. The experimental results show that this method outperforms or competes with the original ensemble and some major state-of-the-art results on the data sets Fer2013, JAFFE, and CK $$+$$ .