Weight Averaging Improves Knowledge Distillation under Domain Shift
This work addresses domain generalization in knowledge distillation, offering a practical improvement for model compression in non-i.i.d. settings, though it is incremental as it applies existing techniques to a new context.
The paper tackles the problem of knowledge distillation under domain shift, showing that weight averaging techniques improve student network performance on unseen domains, with the proposed WAKD method performing on par with existing strategies without needing validation data.
Knowledge distillation (KD) is a powerful model compression technique broadly used in practical deep learning applications. It is focused on training a small student network to mimic a larger teacher network. While it is widely known that KD can offer an improvement to student generalization in i.i.d setting, its performance under domain shift, i.e. the performance of student networks on data from domains unseen during training, has received little attention in the literature. In this paper we make a step towards bridging the research fields of knowledge distillation and domain generalization. We show that weight averaging techniques proposed in domain generalization literature, such as SWAD and SMA, also improve the performance of knowledge distillation under domain shift. In addition, we propose a simplistic weight averaging strategy that does not require evaluation on validation data during training and show that it performs on par with SWAD and SMA when applied to KD. We name our final distillation approach Weight-Averaged Knowledge Distillation (WAKD).