Motivated by this, our work is to propose one 'knowledge reverse distillation' strategy for the confidence calibration of deep neural networks.
確定! 回上一頁