Give me your attention: dot-product attention considered harmful for adversarial patch robustness
Neural architectures based on attention such as vision transformers are revolutionizing image recognition. Their main benefit is that attention allows reasoning about all parts of a scene jointly. In this paper, we show how the global reasoning of (scaled) dot-product attention can be the source of...
Autori principali: | Lovisotto, G, Finnie, N, Munoz, M, Mummadi, CK, Metzen, JH |
---|---|
Natura: | Conference item |
Lingua: | English |
Pubblicazione: |
IEEE
2022
|
Documenti analoghi
Documenti analoghi
-
AttentionGAN: unpaired image-to-image translation using attention-guided generative adversarial networks.
di: Tang, H, et al.
Pubblicazione: (2021) -
Adversarial patch detection
di: Yeong, Joash Ler Yuen
Pubblicazione: (2022) -
Adversarial robustness without perturbations
di: Rodríguez Muñoz, Adrán
Pubblicazione: (2024) -
ME-Net: Towards effective adversarial robustness with matrix estimation
di: Yang, Yuzhe, et al.
Pubblicazione: (2022) -
ME-Net: Towards effective adversarial robustness with matrix estimation
di: Yang, Y, et al.
Pubblicazione: (2021)