Full Kullback-Leibler-Divergence Loss for Hyperparameter-free Label Distribution Learning | Read Paper on Bytez