VISION DIFFMASK: Faithful Interpretation of Vision Transformers with Differentiable Patch Masking
This addresses the problem of interpretability for users of Vision Transformers in critical applications, though it is incremental as it builds on existing attribution methods.
The paper tackles the lack of interpretability in Vision Transformers by proposing VISION DIFFMASK, a post-hoc method that identifies minimal input subsets preserving class predictions, achieving compelling results on CIFAR-10 and ImageNet-1K.
The lack of interpretability of the Vision Transformer may hinder its use in critical real-world applications despite its effectiveness. To overcome this issue, we propose a post-hoc interpretability method called VISION DIFFMASK, which uses the activations of the model's hidden layers to predict the relevant parts of the input that contribute to its final predictions. Our approach uses a gating mechanism to identify the minimal subset of the original input that preserves the predicted distribution over classes. We demonstrate the faithfulness of our method, by introducing a faithfulness task, and comparing it to other state-of-the-art attribution methods on CIFAR-10 and ImageNet-1K, achieving compelling results. To aid reproducibility and further extension of our work, we open source our implementation: https://github.com/AngelosNal/Vision-DiffMask