Learning from a Lightweight Teacher for Efficient Knowledge Distillation
This work addresses the training time bottleneck in knowledge distillation for model compression, offering an incremental improvement over teacher-free methods.
The paper tackles the inefficiency of training complex teacher models in knowledge distillation by proposing LW-KD, which uses a lightweight teacher trained on a synthesized dataset to guide student learning, achieving effective results on multiple public datasets.
Knowledge Distillation (KD) is an effective framework for compressing deep learning models, realized by a student-teacher paradigm requiring small student networks to mimic the soft target generated by well-trained teachers. However, the teachers are commonly assumed to be complex and need to be trained on the same datasets as students. This leads to a time-consuming training process. The recent study shows vanilla KD plays a similar role as label smoothing and develops teacher-free KD, being efficient and mitigating the issue of learning from heavy teachers. But because teacher-free KD relies on manually-crafted output distributions kept the same for all data instances belonging to the same class, its flexibility and performance are relatively limited. To address the above issues, this paper proposes en efficient knowledge distillation learning framework LW-KD, short for lightweight knowledge distillation. It firstly trains a lightweight teacher network on a synthesized simple dataset, with an adjustable class number equal to that of a target dataset. The teacher then generates soft target whereby an enhanced KD loss could guide student learning, which is a combination of KD loss and adversarial loss for making student output indistinguishable from the output of the teacher. Experiments on several public datasets with different modalities demonstrate LWKD is effective and efficient, showing the rationality of its main design principles.