Learning Clip Representations for Skeleton-Based 3D Action Recognition | IEEE Journals & Magazine | IEEE Xplore