Constellation Loss: Improving the efficiency of deep metric learning loss functions for optimal embedding | Publicación