Polarizing Front Ends for Robust CNNs

Abstract

The vulnerability of deep neural networks to small, adversarially designed perturbations can be attributed to their "excessive linearity." In this paper, we propose a bottom-up strategy for attenuating adversarial perturbations using a nonlinear front end which polarizes and quantizes the data. We observe that ideal polarization can be utilized to completely eliminate perturbations, develop algorithms to learn approximately polarizing bases for data, and investigate the effectiveness of the proposed strategy on the MNIST and Fashion MNIST datasets.

Published in 45th International Conference on Acoustics, Speech, and Signal Processing (ICASSP 2020)

References in corpus (3)

• Adversarial Machine Learning at Scale
• ME-Net: Towards Effective Adversarial Robustness with Matrix Estimation
• Discretization based Solutions for Secure Machine Learning against Adversarial Attacks