Search Results for author:
Found 1 papers, 0 papers with code
Unimodal and Crossmodal Refinement Network for Multimodal Sequence Fusion
Specifically, to improve unimodal representations, a unimodal refinement module is designed to refine modality-specific learning via iteratively updating the distribution with transformer-based attention layers.
Cannot find the paper you are looking for? You can
Submit a new open access paper.
