NEUROSIGN
Abstract
This paper presents NeuroSign --- a neural network digital signature system based on the NeuroHash hash function (8 KB). The system implements a fundamentally new class of cryptographic primitives where keys are not stored separately but are an integral part of the neural network architecture. Comprehensive testing was conducted: 1 million signing operations, 100 thousand uniqueness checks, 50 million collision comparisons, and 7 types of cryptographic attacks. The results confirm 100% correctness, 100% signature uniqueness, an avalanche effect of 49.99% ±2.37%, and no collisions or false positives. The signing speed is 12,500 ops/sec (80 µs), which is 1.5 times faster than Ed25519. The code size is 8 KB (6 times smaller than analogues). Quantum resistance and protection against timing attacks are confirmed.
Fields of Study & Subjects
Cryptography and Data Security
Cryptographic Implementations and Security
Cryptography and Residue Arithmetic
Publication details
- Journal / source
- Zenodo (CERN European Organization for Nuclear Research)
- Publication date
- 13 March 2026
- Language
- EN