2026-05-04T20:59:54.098Z http://ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai

ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai:1:11583 2025-09-09T16:52:35Z 1

The IoT Networks Intrusion Detection by CatBoost Optimized with Altered Firefly Algorithm 2025 http://ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai/record/1/11583 https://ieeexplore.ieee.org/abstract/document/11135393 K. Milojkovic T. Zivkovic M. Zivkovic P. Spalevic L. Jovanovic N. Bacanin Intrusion identification is a pivotal activity in the realm of the Internet of Things (IoT), tasked to protect inter-linked devices, communication frameworks, and data from unauthorized entry and harmful endeavors. Since IoT proliferates quickly over sectors including intelligent residences, self-driving vehicles, industrial automation and medical technology, ensuring the security of these devices has grown vital due to their scarce resources and diverse structural configurations. This study delves into the challenge of intrusion detection in IoT communication systems by employing the CatBoost classifier. A novel adaptation of the well-known firefly algorithm was developed to fine-tune the hyperparameters of the CatBoost classifier, amplifying its efficiency in detecting intrusions within IoT infrastructures. Extensive comparative simulations were executed, evaluating the suggested methodology against a set of potent recent optimizers within an identical setup. The empirical findings underscored the superior effectiveness of the suggested approach, emphasizing its remarkable prospect within this particular field. conferenceObject 1 6 10.1109/ICCSC66714.2025.11135393 2025 International Conference on Circuit, Systems and Communication (ICCSC), IEEE, Fez, Morocco