Generative Adversarial Trainer: Defense to Adversarial Perturbations with GAN

By Sungyeob Han, Jungwoo Lee. 2017

157 citations recorded by the source corpus.

We propose a novel technique to make neural network robust to adversarial examples using a generative adversarial network. We alternately train both classifier and generator networks. The generator network generates an adversarial perturbation that can easily fool the classifier network by using a gradient of each image. Simultaneously, the classifier network is trained to classify correctly both original and adversarial images generated by the generator. These procedures help the classifier network to become more robust to adversarial perturbations. Furthermore, our adversarial training framework efficiently reduces overfitting and outperforms other regularization methods such as Dropout. We applied our method to supervised learning for CIFAR datasets, and experimantal results show that our method significantly lowers the generalization error of the network. To the best of our knowledge, this is the first method which uses GAN to improve supervised learning.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Vision and Pattern Recognition

Citations in the corpus

The Robust Manifold Defense: Adversarial Training using Generative Models (2017, methodology): “The GAT-Trainer work by Lee et al. [28] uses generative models to perform adversarial training but in a very diﬀerent way from our work and further without projecting on the range of a GAN.”
Recent Progress on Generative Adversarial Networks (GANs): A Survey (2019, background): “To against adversarial attacks, [80] and [81] use GANs to make the right defense.”
Threat of Adversarial Attacks on Deep Learning in Computer Vision: A Survey (2018, methodology): “[101] used the popular framework of Generative Adversarial Networks [151] to train a network that is robust to FGSM [23] like attacks.”
Constructing Unrestricted Adversarial Examples with Generative Models (2018, neutral): “Examples include AdvGAN [23], DeepDGA [47], ATN [48], GAT [49] and Defense-GAN [14].”
Disentangling Adversarial Robustness and Generalization (2018, methodology): “Defenses: Proposed defenses include detection and rejection methods [31, 25, 54, 60, 3, 63], pre-processing, quantization and dimensionality reduction methods [12, 73, 7], manifold-projection methods [39, 72, 82, 86], methods based on stochasticity/regularization or adapted architectures [109, 7, 68”
A Review on Generative Adversarial Networks: Algorithms, Theory, and Applications (2020, background): “To be against adversarial attacks [118], [119], references [120], [121] utilize GANs to conduct the right defense.”
Adversarial Examples on Object Recognition (2020, background): “Many conjectures have been proposed and are discussed in this section.”
Adversarial Attacks and Defenses in Deep Learning (2020, background): “[65] first propose to exploit a nondeterministic generator to generate adversarial samples in the process of adversarial training.”
High-Frequency Component Helps Explain the Generalization of Convolutional Neural Networks (2019, background): “In order to defend the model against the de-ception, another group of researchers proposed a wide range of methods (known as defense methods ) [1, 38, 44, 45, 24].”
Distortion Agnostic Deep Watermarking (2020, methodology): “, using approximations to the optimization step [27], or using generative models in place of iterative optimization [4, 18].”
Generative Adversarial Examples (2018, unknown): “Examples include AdvGAN [44], ATN [45], GAT [43] and Defense-GAN [14].”
Adversarial Spheres (2018, background): “Several defenses against adversarial examples have been proposed recently which are motivated by the assumption that adversarial examples are off the data manifold (Anonymous, 2018b,a; Lee et al., 2017).”