Wenqi Niu

Rethinking the Dark Knowledge and Kullback-Leibler Divergence Loss in Knowledge Distillation Under Capacity Mismatching

Knowledge Distillation (KD) aims to transfer the dark knowledge that encodes inter-class similarity, semantic structure, and decision boundaries from a powerful teacher model to a compact student model by minimizing the Kullback-Leibler (KL) divergence between their output distributions. While effec