2026-05-04T20:59:03.765Z http://ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai

ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai:1:11589 2026-03-27T12:39:00Z 1

Modified Metaheuristics for Intrusion Detection in the Internet of Things 2025 http://ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai/record/1/11589 https://ieeexplore.ieee.org/document/11158122 L. Jovanovic M. Stankovic M. Zivkovic M. Antonijevic S. Andjelic N. Bacanin IoT networks are exposed to a variety of cyberthreats because of low processing power and storage capacity of IoT devices, which make it difficult to deploy strong and thorough security solutions. This study addresses these issues by including a modified metaheuristic optimization technique to enhance the performance of security classifiers, therefore proposing an AI-driven security framework specifically designed for IoT networks. In particular, the framework optimizes the accuracy and efficiency of CatBoost classification models by adjusting hyperparameter settings with a modified artificial bee colony (ABC) method. The models were trained and assessed on an actual dataset from a real-world scenario. The findings show that by striking a balance between sophisticated threat mitigation capabilities and computing efficiency, this framework presents an effective approach to enhancing security in IoT contexts. conferenceObject 1 6 10.1109/ASSIC64892.2025.11158122 2025 International Conference on Advancements in Smart, Secure and Intelligent Computing (ASSIC), IEEE, Bhubaneswar, India