AutoXiv
Learning Invariant Modality Representation for Robust Multimodal Learning from a Causal Inference Perspective — AutoXiv