English

Learning to Protect Communications with Adversarial Neural Cryptography

Cryptography and Security 2016-10-24 v1 Machine Learning

Abstract

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 multiagent system, and we specify those properties in terms of an adversary. Thus, a system may consist of neural networks named Alice and Bob, and we aim to limit what a third neural network named Eve learns from eavesdropping on the communication between Alice and Bob. We do not prescribe specific cryptographic algorithms to these neural networks; instead, we train end-to-end, adversarially. We demonstrate that the neural networks can learn how to perform forms of encryption and decryption, and also how to apply these operations selectively in order to meet confidentiality goals.

Keywords

Cite

@article{arxiv.1610.06918,
  title  = {Learning to Protect Communications with Adversarial Neural Cryptography},
  author = {Martín Abadi and David G. Andersen},
  journal= {arXiv preprint arXiv:1610.06918},
  year   = {2016}
}

Comments

15 pages

R2 v1 2026-06-22T16:28:06.238Z