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