Distributionally Robust Multimodal Machine Learning

Peilin Yang, Yu Ma

We consider the problem of distributionally robust multimodal machine learning. Existing approaches often rely on merging modalities on the feature level (early fusion) or heuristi…