Achieving Generalizable Robustness of Deep Neural Networks by Stability Training
This work addresses the need for generalizable robustness in deep learning for applications like image classification, though it appears incremental as it builds on existing stability training methods.
The paper tackled the problem of improving deep neural network robustness against input perturbations by exploring stability training as an alternative to data augmentation, finding that it performs comparably or better for specific transformations and offers improved robustness to unseen distortions with less hyperparameter dependence and side effects.
We study the recently introduced stability training as a general-purpose method to increase the robustness of deep neural networks against input perturbations. In particular, we explore its use as an alternative to data augmentation and validate its performance against a number of distortion types and transformations including adversarial examples. In our image classification experiments using ImageNet data stability training performs on a par or even outperforms data augmentation for specific transformations, while consistently offering improved robustness against a broader range of distortion strengths and types unseen during training, a considerably smaller hyperparameter dependence and less potentially negative side effects compared to data augmentation.