ATRAS: Adversarially Trained Robust Architecture Search
Not on the map. Why?
Abstract (arXiv)
In this paper, we explore the effect of architecture completeness on adversarial robustness. We train models with different architectures on CIFAR-10 and MNIST dataset. For each model, we vary different number of layers and different number of nodes in the layer. For every architecture candidate, we use Fast Gradient Sign Method (FGSM) to generate untargeted adversarial attacks and use adversarial training to defend against those attacks. For each architecture candidate, we report pre-attack, post-attack and post-defense accuracy for the model as well as the architecture parameters and the impact of completeness to the model accuracies.
Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Builds on 14 of its 19 references that are in the map. First posted on arXiv 13 Jun 2021.
Builds on
14 of 19 references.
References · 14
- 2014Explaining and Harnessing Adversarial Examplescited by 7,364
- 2017Towards Deep Learning Models Resistant to Adversarial Attackscited by 6,415
- 2013Intriguing properties of neural networkscited by 5,155
- 2017Ensemble Adversarial Training: Attacks and Defensescited by 840
- 2019Adversarial Training for Free!cited by 643
- 2020On Adaptive Attacks to Adversarial Example Defensescited by 555
- 2018Adversarial Logit Pairingcited by 224
- 2017Deep Sparse Coding for Invariant Multimodal Halle Berry Neuronscited by 8
- 2020Adversarial Attacks on Convolutional Neural Networks in Facial Recognition Domaincited by 8
- 2019Machine-learned metrics for predicting the likelihood of success in materials discoverycited by 6
- 2020Perfecting the Crime Machinecited by 3
- 2020Adversarial Attacks against Neural Networks in Audio Domain: Exploiting Principal Componentscited by 3
2 more
Built on by
Citing papers per year: 2022: 1.
1 paper.