A novel mid-level distinctive feature learning for action recognition via diffusion map
Keywords: Human action recognition; Diffusion map; Spatio-temporal structures; Feature learning; Feature fusion;