Learning to Protect Communications with Adversarial Neural Cryptography

Martín Abadi, David G. Andersen

We ask whether neural networks can learn to use secret keys to protect information from other neural networks. Specifically, we focus on ensuring confidentiality properties in a mu…