Song, Lan, Xing, Zeng, Liu, 2016. An End-to-End Spatio-Temporal Attention Model for Human Action Recognition from Skeleton Data.