CL-DMDF:Dynamic Multimodal Data Fusion Model Based on Contrastive Learning

In three linesCL-DMDF introduces a dynamic multimodal data fusion model using contrastive learning to handle missing or uncertain modalities. It features a dual-dimension attention mechanism (features and modalities) and entity-centroid contrastive learning module for enhanced discrimination. Validated across three datasets.

Summary generated by Claude — human-verified