Marginalised Stacked Denoising Autoencoders for Robust Representation of Real-Time Multi-View Action Recognition | Publicación