2026-05-14T14:47:33.279Z http://ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai

ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai:3:9440 2024-12-03T20:10:47Z 3

Performance of Sine Cosine Algorithm for ANN Tuning and Training for IoT Security, Chapter in LNNS Lecture Notes in Networks and Systems: HIS 2022: Hybrid Intelligent Systems, Springer, volume 647 2023 http://ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai/record/3/9440 https://link.springer.com/chapter/10.1007/978-3-031-27409-1_27 N. Bacanin M. Zivkovic Z. Hajdarevic S. Janicijevic A. Dasho M. Marjanovic L. Jovanovic Recent advances in Internet technology ensured that the World Wide Web is now essential for millions of users, offering them a variety of services. As the number of online transactions grows, the number of hostile users who are trying to manipulate with sensitive data and steal user’s private details, credit card data and money is also rising fast. To fight this threat, security companies developed a variety of security measures, aiming to protect both end user and business offering online services. Nowadays, machine learning methods are common part of the most of the contemporary security solutions. The research goal of this paper is proposal of the hybrid technique that uses multi-layer perceptron tuned by the well-known sine cosine algorithm. Sine cosine metaheuristics is utilized to determine the neural cell count within the hidden layer, and to obtain the weights and biases. The capabilities of the observed method were validated on a public web security benchmark dataset, and compared to the results obtained by other elite metaheuristics that have been tested under the same conditions. The simulation findings indicate that the introduced model surpassed other observed techniques, showing great deal of potential for practical use in this domain. bookPart Springer, Cham 302 312 10.1007/978-3-031-27409-1_27 LNNS Lecture Notes in Networks and Systems: HIS 2022: Hybrid Intelligent Systems, volume 647