JunZhang, C., GuanLi, Y. (2026). A hybrid framework combining adaptive graph learning and global temporal attention for skeleton-based action recognition. Scientific Reports. https://doi.org/10.1038/s41598-026-49915-z