BLACK-BOX ATTACK USING NEURO-IDENTIFIER
Mahmood Khalel Ibrahem Al-Ubaidy · Cryptologia · 2004
The cryptanalysis problem can be described as an unknown (Black-Box), system identification problem, and the goal of the cryptography is to build a cryptographic system that is hard to identify. Neural networks are ideal tools for Black-Box system identification. A black-box attack against classical and stream cryptosystems is presented by constructing a Black-Box Neuro-Identifier model to achieve two objectives; the first objective is to determine the key from given plaintext-ciphertext pair, while the second objective is to construct a Neuro-model for the target cipher system.