Boosting Graph Embedding with Application to Facial Expression Recognition

Lei Xiong, Duyan Bi, Xu Zhou, Shiping Ma

Abstract: As a general framework in feature extract technology, graph embedding has been paid much attention. Integrate graph embedding with boosting method is a potential developing way. We are the first successfully attempt to incorporate these two technology together. Further more, a new adjacency graph weighting method called ldquoclassification graphrdquo was proposed. By using this more suitable graph, the performance of the boosting graph embedding was improved. Experiment results demonstrate the efficient of our approach.