English

Improving Adversarial Robustness in Weight-quantized Neural Networks

Machine Learning 2021-01-26 v2 Cryptography and Security

Abstract

Neural networks are getting deeper and more computation-intensive nowadays. Quantization is a useful technique in deploying neural networks on hardware platforms and saving computation costs with negligible performance loss. However, recent research reveals that neural network models, no matter full-precision or quantized, are vulnerable to adversarial attacks. In this work, we analyze both adversarial and quantization losses and then introduce criteria to evaluate them. We propose a boundary-based retraining method to mitigate adversarial and quantization losses together and adopt a nonlinear mapping method to defend against white-box gradient-based adversarial attacks. The evaluations demonstrate that our method can better restore accuracy after quantization than other baseline methods on both black-box and white-box adversarial attacks. The results also show that adversarial training suffers quantization loss and does not cooperate well with other training methods.

Keywords

Cite

@article{arxiv.2012.14965,
  title  = {Improving Adversarial Robustness in Weight-quantized Neural Networks},
  author = {Chang Song and Elias Fallon and Hai Li},
  journal= {arXiv preprint arXiv:2012.14965},
  year   = {2021}
}

Comments

10 pages

R2 v1 2026-06-23T21:34:38.209Z