NeuroCrypto = Feistel network + Deep Learning
seeking investment partners for an existing product
The Feistel network was chosen as the foundation for the design of encryption algorithms, leveraging the Feistel structure—one of the most robust methods for constructing block ciphers. The network comprises units known as Feistel cells. Each cell receives data and a key as inputs, producing modified data and a modified key as outputs. All cells within the network are identical, creating an iterated structure that repeats uniformly.
The Feistel network features the following cryptographic primitives:
- Substitution Block (S-block)
- Permutation Block (P-block)
NeuroCrypto customizes cryptographic primitives for each client, adhering to cryptographic principles that minimize statistical dependencies and significantly enhance resistance to cryptanalysis. This innovative approach, combined with the Feistel network’s design, enables the training of models embedded within the Feistel framework, ensuring high reliability on par with established encryption algorithms such as BlowFish. Consequently, each client benefits from a unique set of cryptographic primitives, resulting in a distinct encryption algorithm.
The efficacy of this approach is underscored by the results from the NIST test suite for the algorithm in counter (CTR) mode, which is recognized as the most vulnerable mode for block encryption algorithms.