Multimodal Attention-Based Mixture of Experts for Video-Level Activity Recognition using the YouTube-8M Dataset. (2026). ITEGAM-JETIA, 12(61), 1490-1499. https://doi.org/10.5935/jetia.v12i61.4267