Attacking Bayes: On the Adversarial Robustness of Bayesian Neural Networks

2024-04-27 · Yunzhen Feng, Tim G. J. Rudner, Nikolaos Tsilivis, Julia Kempe

Adversarial examples have been shown to cause neural networks to fail on a wide range of vision and language tasks, but recent work has claimed that Bayesian neural networks (BNNs) are inherently robust to adversarial perturbations. In this work, we examine this claim. To study the adversarial robustness of BNNs, we investigate whether it is possible to successfully break state-of-the-art BNN inference methods and prediction pipelines using even relatively unsophisticated attacks for three tasks: (1) label prediction under the posterior predictive mean, (2) adversarial example detection with Bayesian predictive uncertainty, and (3) semantic shift detection. We find that BNNs trained with state-of-the-art approximate inference methods, and even BNNs trained with Hamiltonian Monte Carlo, are highly susceptible to adversarial attacks. We also identify various conceptual and experimental errors in previous works that claimed inherent adversarial robustness of BNNs and conclusively demonstrate that BNNs and uncertainty-aware Bayesian prediction pipelines are not inherently robust against adversarial attacks.

📄 PDF Abstract BibTeX arXiv:2404.19640

Code (0)

등록된 구현이 없습니다.

Tasks

Adversarial Robustness PredictionSemantic Shift Detection

Attacking Graph Classification via Bayesian Optimisation

2021-06-18 · ICML Workshop AML 2021 7 · Xingchen Wan, Henry Kenlay, Binxin Ru, Arno Blaas 외

Graph neural networks have been shown to be vulnerable to adversarial attacks. While the majority of the literature focuses on such vulnerability in node-level classification tasks, little effort has been dedicated to at…

Adversarial Robustness Bayesian Optimisation Classification Graph Classification

Adversarial attacks against Bayesian forecasting dynamic models

2021-10-20 · Roi Naveiro

The last decade has seen the rise of Adversarial Machine Learning (AML). This discipline studies how to manipulate data to fool inference engines, and how to protect those systems against such manipulation attacks. Exten…

regression Time Series Time Series Analysis Time Series Forecasting

Bayesian Optimization with Robust Bayesian Neural Networks

2016-12-01 · NeurIPS 2016 12 · Jost Tobias Springenberg, Aaron Klein, Stefan Falkner, Frank Hutter

Bayesian optimization is a prominent method for optimizing expensive to evaluate black-box functions that is prominently applied to tuning the hyperparameters of machine learning algorithms. Despite its successes, the pr…

Bayesian Optimization Deep Reinforcement Learning Hyperparameter Optimization Reinforcement Learning+1

Improving Transferability of Adversarial Examples via Bayesian Attacks

2023-07-21 · Qizhang Li, Yiwen Guo, Xiaochen Yang, WangMeng Zuo 외

This paper presents a substantial extension of our work published at ICLR. Our ICLR work advocated for enhancing transferability in adversarial examples by incorporating a Bayesian formulation into model parameters, whic…

Comment on "Adv-BNN: Improved Adversarial Defense through Robust Bayesian Neural Network"

2019-07-01 · Roland S. Zimmermann

A recent paper by Liu et al. combines the topics of adversarial training and Bayesian Neural Networks (BNN) and suggests that adversarially trained BNNs are more robust against adversarial attacks than their non-Bayesian…

Adversarial Attack Adversarial Defense