2026-05-04T18:33:30.575Z http://ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai

ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai:3:11799 2026-01-13T10:47:24Z 3

Optimizing Gated Recurrent Units with a Hybrid Crayfish Algorithm for Enhanced Unemployment Rate Prediction, Chapter in LNNS Lecture Notes in Networks and Systems: CVR 2025: Computer Vision and Robotics, Springer, volume 1643 2026 http://ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai/record/3/11799 https://link.springer.com/chapter/10.1007/978-3-032-06250-5_11 L. Babic S. Ivanovic A. Petrovic L. Jovanovic M. Zivkovic A. Zivaljevic M. Antonijevic N. Bacanin The economy is the backbone of the society, and one of key factors that influences it is the unemployment rate. The well-being of society is also influenced by this factor making it undoubtedly paramount. The traditional methods for predicting such trends provide some insights, but are far from the potential that can be achieved through the use of artificial intelligence (AI). Therefore, the research proposed in this article employs AI framework for solving this important issue, which is influenced by many complex relationships. The problem of unemployment rate prediction is formulated as a time series challenge and for such issues recurrent neural networks (RNNs) may obtain satisfying performance. Research proposed in this manuscript utilizes gated recurrent unit (GRU) networks, a subtype of RNNs, coupled with an attention mechanism. However, this approach necessitates hyperparameter tuning, prompting the use of metaheuristics as an effective solution. The crayfish optimization algorithm (COA) further enhanced to suit the specific requirements of the task is employed to tune GRU models. The proposed solution demonstrates superior performance compared to both state-of-the-art methods and the original COA in terms of standard time-series prediction metrics. bookPart Springer, Cham 128 139 10.1007/978-3-032-06250-5_11 LNNS Lecture Notes in Networks and Systems: CVR 2025: Proceedings of International Conference on Computer Vision and Robotics, volume 1643