Random Number Generator Based on Hopfield Neural Network with Xorshift and Genetic Algorithms

Lecca C.; Zegarra A.; Santisteban J. · Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (2024) · DOI: 10.1007/978-3-031-47765-2_21

Resumen

Generating random numbers plays a fundamental role in matters of computer security. New high-security standards for data encryption have emerged with the increasing demand for secure communication, storage, and management of sensitive data on public networks. Data encryption techniques have expanded rapidly. In this paper, we present two methods of pseudorandom number generation; the first method combines the existing architectures of Hopfield neural networks and the XORShift algorithm, an the second is based on genetic algorithms. Our approaches pass the NIST and Diehard tests, which give us numbers that satisfy the statistical requirements. Compared with other architectures, the results demonstrated a robust consistency in the quality of the generated random numbers, which shows that it is a reliable method.

Instituciones: Universidad Católica San Pablo