Distilling Out-of-Distribution Robustness from Vision-Language Foundation Models
This work addresses robustness challenges in computer vision for real-world deployment, though it appears incremental as it builds on existing knowledge distillation and data augmentation techniques.
The paper tackles the problem of improving out-of-distribution robustness in vision models by proposing Discrete Adversarial Distillation (DAD), which distills knowledge from pretrained foundation models and uses adversarial examples discretized via VQGAN for data augmentation, achieving strong gains in both robustness and clean accuracy across different student architectures.
We propose a conceptually simple and lightweight framework for improving the robustness of vision models through the combination of knowledge distillation and data augmentation. We address the conjecture that larger models do not make for better teachers by showing strong gains in out-of-distribution robustness when distilling from pretrained foundation models. Following this finding, we propose Discrete Adversarial Distillation (DAD), which leverages a robust teacher to generate adversarial examples and a VQGAN to discretize them, creating more informative samples than standard data augmentation techniques. We provide a theoretical framework for the use of a robust teacher in the knowledge distillation with data augmentation setting and demonstrate strong gains in out-of-distribution robustness and clean accuracy across different student architectures. Notably, our method adds minor computational overhead compared to similar techniques and can be easily combined with other data augmentations for further improvements.