Attention-based Knowledge Distillation in Multi-attention Tasks: The Impact of a DCT-driven Loss | Publicación