Neural cryptography

From Wikipedia, the free encyclopedia

Neural cryptography is a branch of cryptography dedicated to analyzing the application of stochastic algorithms, especially neural network algorithms, for use in encryption and cryptanalysis.

Contents 1 Definition 2 Applications 3 See also 4 References

[edit] Definition

The ability of neural networks to explore the solution space could also be used in the field of Cryptanalysis. It also could be possible to generate new kind of attacks on existing algorithms based on the idea that any function could be reproduced by a neural network, so it will be possible to find the exact solution, at least theoretically, breaking the algorithm.

The ideas of mutual learning, self learning, and stochastic behavior of neural networks and similar algorithms can be used for different aspects of cryptography, like public-key cryptography, solving the key distribution problem using neural network mutual synchronization, hashing or generation of pseudo-random numbers.

Another idea is the ability of Neural Network to separate space in non-linear pieces using "bias". By the way, it gives different probabilities of activating or not the neural network. This is very useful in the case of the Cryptography or Crytanalysis.

Two names are used to design the same domain of researches : Neuro-Cryptography and Neural Cryptography. We can add followings derivated names: Neuro-Cryptanalysis, Neuro-Encoding, Neuro-Key, ... as well for Neural.

To try to bring you a beginning date, Sebastien Dourlens has introduced this domain the first time in his IT Master in 1995. At least 30 years after the first talk on the definition of the basics of Neural Network.

[edit] Applications

Still there are no practical applications due to the recently of the development of the field, but it could be used specially over applications where the keys could be continually generated and the system (both pairs and the insecure media) could be in a continuous evolving mode. In 1995, One application (see reference) made by Sebastien Dourlens concerns the way how to cryptanalyse the famous DES (Data Encryption Standard) by learning how to invert the S-tables of the DES. Results are very interesting. All bias studied by the Differential Cryptanalysis of the DES done by Shamir are highlighted. And the experiment shows about 50% of the bits of the key can be found giving the ability to find the complete key in a very few time. A proposal of hardware application with multi micro-controllers has been explained due to the easy implementation of multilayer neural networks.

[edit] See also

• Neural Network
• Stochastic neural network

[edit] References

• Neuro-Cryptography 1995 - The first definition of the Neuro-Cryptography (AI Neural-Cryptography) applied to DES cryptanalysis by Sebastien Dourlens, France.
• Neural Cryptography - Description of one kind of neural cryptography at the University of Würzburg, Germany.
• Using synchronized neural networks to achieve public key authentication system - One of the leading papers that introduce the concept.
• A remote password authentication scheme for multiserverarchitecture using neural networks - Possible practical application of Neural Cryptography.
• Analysis of Neural Cryptography - Analysis of neural cryptography in general and focusing on the weakness and possible attacks of using synchronized neural networks.

v • d • e   Cryptography History of cryptography | Cryptanalysis | Cryptography portal | Topics in cryptography Symmetric-key algorithm | Block cipher | Stream cipher | Public-key cryptography | Cryptographic hash function | Message authentication code | Random numbers | Steganography