Towards Deep Learning Models Resistant to Adversarial Attacks

📖 Description

Towards Deep Learning Models Resistant to Adversarial Attacks Aleksander M ? adry MIT [email removed]Aleksandar Makelov MIT [email removed]Ludwig Schmidt MIT [email removed] Dimitris Tsipras MIT [email removed]Adrian Vladu MIT [email removed] Abstract Recent work has demonstrated that deep neural networks are vulnerable to adversarial examples?inputs that are almost indistinguishable from natural data and yet classi?ed incor- rectly by the network. In fact, some of the latest ?ndings suggest that the existence of adversarial attacks may be an inherent weakness of deep learning models. To address this problem, we study the adversarial robustness of neural networks through the lens of robust optimization. This approach provides us with a broad and unifying view on much of the prior work on this topic. Its principled nature also enables us to identify methods for both training and attacking neural networks that are reliable and, in a certain sense, universal. In particular, they spe...

📊 Game Impacts

💭 Reactions

🔗 Sources

• [1] https://arxiv.org/abs/1706.06083