Multi-conditional Latent Variable Model for Joint Facial Action Unit Detection | IEEE Conference Publication | IEEE Xplore