Transformers—Messages in disguise

Compared to human-designed algorithms, Neural Network (NN)-based cryptography can create a new NN-specific cryptographic scheme that changes every time the NN is (re)trained. NN retraining is advantageous because it requires an adversary to restart its process(es) to learn or break the cryptographic...

詳細記述

書誌詳細

出版年:	Machine Learning with Applications
主要な著者:	Joshua H. Tyler, Donald R. Reising, Mohamed M.K. Fadul
フォーマット:	論文
言語:	英語
出版事項:	Elsevier 2025-06-01
主題:	Adversarial Neural Cryptography (ANC) One-Time Pad (OTP) Perfect Forward Security (PFS) Neural Network (NN) Deep Learning (DL)
オンライン･アクセス:	http://www.sciencedirect.com/science/article/pii/S2666827025000520