As skeleton data becomes increasingly available, Graph Convolutional Networks (GCNs) are popularly adapted to extract the spatial and temporal features for skeleton-based action recognition. However, ...