Idealised Bayesian Neural Networks Cannot Have Adversarial Examples: Theoretical and Empirical Study

06/02/2018
by   Yarin Gal, et al.
0

We prove that idealised discriminative Bayesian neural networks, capturing perfect epistemic uncertainty, cannot have adversarial examples: Techniques for crafting adversarial examples will necessarily fail to generate perturbed images which fool the classifier. This suggests why MC dropout-based techniques have been observed to be fairly robust to adversarial examples. We support our claims mathematically and empirically. We experiment with HMC on synthetic data derived from MNIST for which we know the ground truth image density, showing that near-perfect epistemic uncertainty correlates to density under image manifold, and that adversarial images lie off the manifold. Using our new-found insights we suggest a new attack for MC dropout-based models by looking for imperfections in uncertainty estimation, and also suggest a mitigation. Lastly, we demonstrate our mitigation on a cats-vs-dogs image classification task with a VGG13 variant.

READ FULL TEXT

page 8

page 9

page 12

research
10/02/2022

Understanding Adversarial Robustness Against On-manifold Adversarial Examples

Deep neural networks (DNNs) are shown to be vulnerable to adversarial ex...
research
07/06/2023

Quantification of Uncertainty with Adversarial Models

Quantifying uncertainty is important for actionable predictions in real-...
research
02/08/2021

Exploiting epistemic uncertainty of the deep learning models to generate adversarial samples

Deep neural network architectures are considered to be robust to random ...
research
03/09/2018

Detecting Adversarial Examples - A Lesson from Multimedia Forensics

Adversarial classification is the task of performing robust classificati...
research
03/22/2018

Understanding Measures of Uncertainty for Adversarial Example Detection

Measuring uncertainty is a promising technique for detecting adversarial...
research
04/25/2022

When adversarial examples are excusable

Neural networks work remarkably well in practice and theoretically they ...

Please sign up or login with your details

Forgot password? Click here to reset