Blockchain Based Secure Communication for Neural Network Training

EasyChair Preprint 14111

Abstract

In recent years, the use of neural networks for machine learning tasks has become increasingly prevalent. However, the security and privacy concerns associated with training these models have also grown. This study proposes a novel approach to address these concerns by leveraging blockchain technology for secure communication during the neural network training process.

The proposed system utilizes blockchain's decentralized nature, cryptographic techniques, and smart contracts to ensure the confidentiality, integrity, and availability of data and communication channels. By storing training data and model updates on the blockchain, the system prevents unauthorized access and tampering. Additionally, the use of smart contracts enables automated verification and enforcement of communication protocols, ensuring that only trusted parties can participate in the training process.

To evaluate the effectiveness of the proposed approach, experiments were conducted using a real-world dataset. The results demonstrate that the blockchain-based system provides enhanced security and privacy compared to traditional centralized approaches. It not only protects against data breaches and unauthorized modifications but also enables transparent and auditable training processes.

Keyphrases: Blockchain, Cybersecurity, Malware

@booklet{EasyChair:14111,
  author    = {Ralph Shad and Seyi Damola and Axel Egon},
  title     = {Blockchain Based Secure Communication for Neural Network Training},
  howpublished = {EasyChair Preprint 14111},
  year      = {EasyChair, 2024}}