Attacking Binarized Neural Networks

Anonymous

Nov 03, 2017 (modified: Dec 05, 2017) ICLR 2018 Conference Blind Submission readers: everyone Show Bibtex
  • Abstract: Neural networks with low-precision weights and activations offer compelling efficiency advantages over their full-precision equivalents. The two most frequently discussed benefits of quantization are reduced memory consumption, and a faster forward pass when implemented with efficient bitwise operations. We propose a third benefit of very low-precision neural networks: improved robustness against some adversarial attacks, and in the worst case, performance that is on par with full-precision models. We focus on the very low-precision case where weights and activations are both quantized to $\pm$1, and note that stochastically quantizing weights in just one layer can sharply reduce the impact of iterative attacks. We observe that non-scaled binary neural networks exhibit a similar effect to the original \emph{defensive distillation} procedure that led to \emph{gradient masking}, and a false notion of security. We address this by conducting both black-box and white-box experiments with binary models that do not artificially mask gradients.
  • TL;DR: We conduct adversarial attacks against binarized neural networks and show that we reduce the impact of the strongest attacks, while maintaining comparable accuracy in a black-box setting
  • Keywords: adversarial examples, adversarial attacks, binary, binarized neural networks

Loading