2026-05-10T08:08:54.084Z http://ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai

ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai:3:11410 2025-05-15T23:07:32Z 3

Optimizing XGBoost for Blockchain Malicious Node Detection by Metaheuristics, Chapter in CCIS Communications in Computer and Information Science: icSoftComp 2024: International Conference on Soft Computing and its Engineering Applications, Springer, volume 2430 2025 http://ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai/record/3/11410 https://link.springer.com/chapter/10.1007/978-3-031-88039-1_3 L. Jovanovic M. Protic N. Bacanin M. Mravik M. Zivkovic M. Antonijevic J. Cadjenovic Milovanovic Concerns over proof of work and proof of stake techniques that are fundamental to the blockchain yet also susceptible to certain security flaws have grown as a result of increased adoption of the blockchain. Notably, malevolent nodes might work together to create a fake consensus, which can cause the blockchain to be incorrectly validated. This vulnerability affects both PoW and PoS algorithms. Creating strong defenses against malevolent nodes for these systems is crucial to resolving this pressing problem. This work investigates the possibilities of using metaheuristic approaches to optimize artificial intelligence systems for the detection of rogue nodes. An altered form of a metaheuristic optimization technique was presented and used on a semi-synthetic dataset that was made available to the general public. With the help of the suggested method, the top-performing classification algorithms were optimized, and their accuracy surpassed 87% suggest suitable potential. bookPart Springer, Cham 30 43 10.1007/978-3-031-88039-1_3 CCIS Communications in Computer and Information Science: icSoftComp 2024: Soft Computing and its Engineering Applications, volume 2430