Generative Adversarial Networks (GANs) have been a hot topic in Deep Learning ever since their initial invention and publication at NIPS 2014. There's a good reason for it all: GANs can create new content based on only a small bit of guidance. It's that sort of creativity which makes them so powerful. Just to name a few there. With all of this, massive resources are being poured into GAN research to figure out both how they work and how to design the absolute best GAN networks.

Generative Adversarial Networks (GANs) have been a hot topic in Deep Learning ever since their initial invention and publication at NIPS 2014. There's a good reason for it all: GANs are able to create totally new content based on only a small bit of guidance. It's that sort of creativity which makes them so powerful. Just to name a few there. With all of this, massive resources are being poured into GAN research to figure out both how they work and how to design the absolute best GAN networks.

Thanks to the creation of AutoML -- which is essentially automated neural architecture search (NAS) -- AI can now design better deep neural networks than human researchers for computer vision tasks such as image classification and object detection. AutoML's tremendous success has prompted AI researchers to explore its efficacy in additional areas, such as generative adversarial networks (GANs). Researchers from Texas A&M University and MIT-IBM Watson AI Lab recently presented a paper that applies NAS to GANs. Their "AutoGAN" is an architecture search scheme specifically tailored for GANs that outperforms current state-of-the-art hand-crafted GANs on the task of unconditional image generation. The associated paper has been accepted by ICCV 2019.

Classifiers fail to classify correctly input images that have been purposefully and imperceptibly perturbed to cause misclassification. This susceptability has been shown to be consistent across classifiers, regardless of their type, architecture or parameters. Common defenses against adversarial attacks modify the classifer boundary by training on additional adversarial examples created in various ways. In this paper, we introduce AutoGAN, which counters adversarial attacks by enhancing the lower-dimensional manifold defined by the training data and by projecting perturbed data points onto it. AutoGAN mitigates the need for knowing the attack type and magnitude as well as the need for having adversarial samples of the attack. Our approach uses a Generative Adversarial Network (GAN) with an autoencoder generator and a discriminator that also serves as a classifier. We test AutoGAN against adversarial samples generated with state-of-the-art Fast Gradient Sign Method (FGSM) as well as samples generated with random Gaussian noise, both using the MNIST dataset. For different magnitudes of perturbation in training and testing, AutoGAN can surpass the accuracy of FGSM method by up to 25\% points on samples perturbed using FGSM. Without an augmented training dataset, AutoGAN achieves an accuracy of 89\% compared to 1\% achieved by FGSM method on FGSM testing adversarial samples.