2026-05-10T08:08:34.662Z http://ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai

ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai:3:11492 2025-07-20T20:21:48Z 3

Multi-headed Metaheuristic-Optimized Gated Recurrent Unit Networks for Ethereum Price Forecasting, Chapter in LNNS Lecture Notes in Networks and Systems: CIS 2024: Fifth Congress on Intelligent Systems, Springer, volume 1277 2025 http://ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai/record/3/11492 https://link.springer.com/chapter/10.1007/978-981-96-2700-4_16 N. Bacanin L. Jovanovic M. Mravik M. Zivkovic M. Milovanovic M. Antonijevic D. Kavitha As the cryptocurrency market continues to develop, accurately predicting trends has become more crucial and fascinating for both academic and practical applications. Nevertheless, forecasting cryptocurrency prices is challenging due to the market’s volatility, influenced by sudden regulatory shifts, technological innovations, and changes in investor sentiment. These factors add a significant level of uncertainty and complexity to prediction models. Artificial intelligence has proven to be a potent tool for addressing these challenges, showing considerable promise in time-series forecasting, particularly within the ETH and cryptocurrency markets. Advanced AI models, like recurrent neural networks (RNNs), have demonstrated exceptional predictive capabilities. To address the complex task of forecasting ETH prices, this study proposes a multi-headed RNN model with gated recurrent unit (GRU) heads. Recognizing the substantial impact of hyper-parameters on RNN performance, an altered particle swarm optimization (PSO) metaheuristics was developed. This optimization method is tailored to calibrate the hyper-parameters of the multi-headed GRU model for this specific task. Simulations findings reveal that the metaheuristics-tuned multi-headed GRU models are able to obtain satisfactory performance in ETH price prediction according to standard time-series regression metrics. bookPart Springer, Singapore 207 221 10.1007/978-981-96-2700-4_16 LNNS Lecture Notes in Networks and Systems: CIS 2024: Fifth Congress on Intelligent Systems, volume 1277