Learning to Protect Communications with Adversarial Neural Cryptography

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.

15 pages

References in corpus (3)

• Explaining and Harnessing Adversarial Examples
• Deep Generative Image Models using a Laplacian Pyramid of Adversarial Networks
• Crypto-Nets: Neural Networks over Encrypted Data

Cited by in corpus (7)

• A Review on Generative Adversarial Networks: Algorithms, Theory, and Applications
• Cooperative Multi-Agent Reinforcement Learning for Low-Level Wireless Communication
• Message Passing Multi-Agent GANs
• Recent Advances of Image Steganography with Generative Adversarial Networks
• Distributed generation of privacy preserving data with user customization
• A new neural-network-based model for measuring the strength of a pseudorandom binary sequence
• Loosely Coupled Federated Learning Over Generative Models