Optimizing Machine Learning for Malicious Blockchain Node Detection Using Modified Metaheuristic

Abstract

The decentralized architecture of blockchain tech-nology introduces various challenges, particularly in Proof of Stake (PoS) networks, where malicious nodes can undermine the consensus process and threaten the integrity of the network. Nodes with significant stakes may disproportionately influence validation, weakening the network's decentralization. Despite ongoing advancements in blockchain security, there remains a notable gap in research on the integration of artificial intelli-gence (AI) and optimization techniques for enhancing blockchain security. This paper proposes an approach, utilizing a CatBoost classifier optimized with a modified Particle Swarm Optimization (PSO) algorithm, designed specifically for this study. A compar-ative analysis, performed on a real-world dataset, demonstrates the effectiveness of the proposed method, with the most accurate models achieving an accuracy of 87.13%. The findings highlight the potential of AI -driven optimization in improving blockchain security. Future research will focus on the application of AI and optimization techniques to further refine blockchain protocols, enhance detection and prevention of security threats, and improve overall network performance.