Facial expression recognition using multi-class minimax probability machine
Ping He, Guofeng Pan, Yatong Zhou, Hongdong Zhao · 2008
Facial expression recognition has received more and more attentions during the last two decades. A variety of recognition techniques have been applied in various applications. In this paper, a novel expression recognition technique is proposed based on a state-of-the-art classifier called minimax probability machine (MPM). After introducing some technical details of preprocessing and feature extraction, we present a multi-class MPM classification for expression recognition. In the experiments, we compare MPM-based recognition technique to other traditional techniques including nearest neighbor (NN) and support vector machine (SVM). The results illustrate that the MPM-based technique is competitive and promising for expression recognition, and it can provide a low bound on classification accuracy.