Research conducted within the Horizon Europe TWINNING project ARTIFACT was presented at NeurIPS 2025, one of the world’s leading international conferences in artificial intelligence.

The paper entitled “TC-GTN: Temporal Convolution Graph Transformer Network for Hydrological Forecasting” explores the application of advanced AI models for river flow forecasting. The proposed approach contributes to improved flood risk management, enhanced preparedness for extreme climate events, and more sustainable hydropower system operation.

The model was evaluated using hydrological data from the Drina–Lim river basin, where it demonstrated improved performance in predicting both regular river flows and extreme hydrological events, highlighting its potential for real-world applications in climate-resilient water management. This work reflects ARTIFACT’s objective of strengthening research excellence and fostering the application of cutting-edge artificial intelligence methods to address complex societal and environmental challenges.

Authors: Ana Dodig Samac, Dr Milan Dotlić, Luka Vinokić, Dr Milan Stojković, Dr Veljko Prodanović