Resilient Vision-Tabular Multimodal Learning under Modality Missingness
在模态缺失下的视觉-表格多模态学习的鲁棒性
机构 * Research Unit of Artificial Intelligence and Computer Systems, Department of Engineering, Università Campus Bio-Medico di Roma(人工智能与计算机系统研究单位,工程系,罗马生物医学大学) ; Department of Diagnostics and Intervention, Radiation Physics, Biomedical Engineering, Umeå University(诊断与干预系,辐射物理,生物医学工程,乌梅大学)
专题命中 医疗多模态 :medical image(abstract);分类 cs.CV、cs.LG
AI总结 本文提出一种多模态Transformer框架,用于在普遍存在模态缺失的情况下进行视觉-表格联合学习,通过可学习的模态标记和中间融合实现鲁棒的多模态推理。