Graph-Attention Fusion with VAE Cross-Modal Mapping and Reinforcement-Learning Visualization for Real-Time AR. IJCAI. 2025;49(14). doi:10.31449/inf.v49i14.11191