A unifying Bayesian framework for adversarial robustness
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , |
|---|---|
| Format: | Preprint |
| Publié: |
2025
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866918534746996736 |
|---|---|
| author | Arce, Pablo G. Naveiro, Roi Insua, David Ríos |
| author_facet | Arce, Pablo G. Naveiro, Roi Insua, David Ríos |
| contents | The vulnerability of machine learning models to adversarial attacks remains a critical societal security challenge. Traditional defenses, such as adversarial training, typically robustify models by minimizing a worst-case loss. These deterministic approaches do not account for uncertainty in the adversary's attack. While stochastic defenses placing a probability distribution on the adversary exist, they often lack statistical rigor and fail to make explicit their underlying assumptions. To resolve these issues, we introduce a formal Bayesian framework that models adversarial uncertainty through a stochastic channel, articulating all probabilistic assumptions. This yields two robustification strategies: a proactive defense enacted during training, aligned with adversarial training, and a reactive defense enacted during operations, aligned with adversarial purification. Several state-of-the-art defenses can be recovered as limiting cases of our model. We empirically validate our methodology, showcasing the benefits of explicitly modeling adversarial uncertainty. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2510_09288 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | A unifying Bayesian framework for adversarial robustness Arce, Pablo G. Naveiro, Roi Insua, David Ríos Machine Learning 68T37 The vulnerability of machine learning models to adversarial attacks remains a critical societal security challenge. Traditional defenses, such as adversarial training, typically robustify models by minimizing a worst-case loss. These deterministic approaches do not account for uncertainty in the adversary's attack. While stochastic defenses placing a probability distribution on the adversary exist, they often lack statistical rigor and fail to make explicit their underlying assumptions. To resolve these issues, we introduce a formal Bayesian framework that models adversarial uncertainty through a stochastic channel, articulating all probabilistic assumptions. This yields two robustification strategies: a proactive defense enacted during training, aligned with adversarial training, and a reactive defense enacted during operations, aligned with adversarial purification. Several state-of-the-art defenses can be recovered as limiting cases of our model. We empirically validate our methodology, showcasing the benefits of explicitly modeling adversarial uncertainty. |
| title | A unifying Bayesian framework for adversarial robustness |
| topic | Machine Learning 68T37 |
| url | https://arxiv.org/abs/2510.09288 |