$V_kD:$ Improving Knowledge Distillation using Orthogonal Projections
This work addresses the problem of improving knowledge distillation for training efficient models across diverse tasks and modalities, offering a novel method with broad applicability.
The paper tackles the limitation of knowledge distillation methods degenerating when transferred to other tasks or architectures by proposing a constrained feature distillation method with orthogonal projection and task-specific normalization, resulting in transformer models outperforming previous methods on ImageNet with up to a 4.4% relative improvement and showing consistent gains in object detection and image generation.
Knowledge distillation is an effective method for training small and efficient deep learning models. However, the efficacy of a single method can degenerate when transferring to other tasks, modalities, or even other architectures. To address this limitation, we propose a novel constrained feature distillation method. This method is derived from a small set of core principles, which results in two emerging components: an orthogonal projection and a task-specific normalisation. Equipped with both of these components, our transformer models can outperform all previous methods on ImageNet and reach up to a 4.4% relative improvement over the previous state-of-the-art methods. To further demonstrate the generality of our method, we apply it to object detection and image generation, whereby we obtain consistent and substantial performance improvements over state-of-the-art. Code and models are publicly available: https://github.com/roymiles/vkd