Playing with blocks: Toward re-usable deep learning models for side-channel profiled attacks | Publicación