Robust Convolutional Neural Networks under Adversarial Noise
This addresses the security and reliability problem of CNNs for applications like image classification, though it appears incremental as it builds on existing noise-based defenses.
The paper tackles the vulnerability of Convolutional Neural Networks (CNNs) to adversarial noise by proposing a new feedforward CNN that adds stochastic noise to inputs and models, resulting in improved robustness that outperforms other methods on CIFAR-10 and ImageNet tests, especially for difficult tasks or strong noise.
Recent studies have shown that Convolutional Neural Networks (CNNs) are vulnerable to a small perturbation of input called "adversarial examples". In this work, we propose a new feedforward CNN that improves robustness in the presence of adversarial noise. Our model uses stochastic additive noise added to the input image and to the CNN models. The proposed model operates in conjunction with a CNN trained with either standard or adversarial objective function. In particular, convolution, max-pooling, and ReLU layers are modified to benefit from the noise model. Our feedforward model is parameterized by only a mean and variance per pixel which simplifies computations and makes our method scalable to a deep architecture. From CIFAR-10 and ImageNet test, the proposed model outperforms other methods and the improvement is more evident for difficult classification tasks or stronger adversarial noise.