dynamic-fusion

This paper proposes CL-DMDF, a dynamic multimodal data fusion model that uses contrastive learning and a dual-dimensional attention mechanism to handle missing modalities and improve discriminative learning.