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