On the Structural Sensitivity of Deep Convolutional Networks to the Directions of Fourier Basis Functions
This work addresses a security issue in deep learning by revealing structural vulnerabilities, with incremental contributions to understanding and exploiting network sensitivity.
The study tackled the problem of deep convolutional networks' sensitivity to data-agnostic perturbations, finding that these networks are particularly sensitive to the directions of Fourier basis functions, which was empirically validated. As a result, they proposed an algorithm for creating shift-invariant universal adversarial perturbations in black-box settings.
Data-agnostic quasi-imperceptible perturbations on inputs are known to degrade recognition accuracy of deep convolutional networks severely. This phenomenon is considered to be a potential security issue. Moreover, some results on statistical generalization guarantees indicate that the phenomenon can be a key to improve the networks' generalization. However, the characteristics of the shared directions of such harmful perturbations remain unknown. Our primal finding is that convolutional networks are sensitive to the directions of Fourier basis functions. We derived the property by specializing a hypothesis of the cause of the sensitivity, known as the linearity of neural networks, to convolutional networks and empirically validated it. As a by-product of the analysis, we propose an algorithm to create shift-invariant universal adversarial perturbations available in black-box settings.