2026-05-04T19:54:18.509Z http://ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai

ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai:1:11816 2026-01-26T16:27:44Z 1

CatBoost Optimized with Adapted Reptile Search Algorithm for Intrusion Detection Systems 2025 http://ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai/record/1/11816 https://ieeexplore.ieee.org/abstract/document/11314223 D. Cvetkovic M. Zivkovic T. Zivkovic N. Bacanin S. Anetic B. Radomirovic The purpose of this paper is to analyze the potential solutions offered by machine learning and artificial intelligence for intrusion detection systems when applied to problems such as the Metaverse and IoT. The exponential growth of such systems has created a need for better and more efficient detection solutions. The purpose of the research is to quantify and verify the improvement in the results of the classification of traffic provided in the dataset. The methodology of the research is an experimental application of the CatBoost machine learning model, which has been further optimized with the use of a proposed modified reptile search metaheuristic optimization algorithm. The models were developed and validated using data drawn from a real operational environment. Results indicate that by carefully balancing advanced threat detection capabilities with computational efficiency, the proposed framework offers a practical and powerful solution for strengthening security within IoT ecosystems. conferenceObject 1 4 10.1109/TELFOR67910.2025.11314223 2025 33rd Telecommunications Forum (TELFOR), IEEE, Belgrade, Serbia