A Correlated Topic Modeling Approach for Facial Expression Recognition

Topic Models has been applied to many Computer Vision problems, and in particular, facial expression recognition, with good results. However, the topic found by the topic models may be correlated, and hence the independent assumption may not hold. Correlated Topic Models (CTM) has be developed to tackle this problem. In this paper, we attempted to use Correlated Topic Models (CTM) for facial expression recognition, and preliminary results are reported.