DiscoVL: Unveiling Disentangled C ross-Modal Representation Learning via Orthogonal Adversarial Regularization for V ision-Language Models

作者：Mengping Dong, Jinbao Li, Fei Li · 单位：Shandong Artificial Intelligence Institute, Qilu University of Technology (Shandong Academy of Sciences), Jinan, China, University of Florida, Florida, USA · 会议/期刊：arXiv preprint · 方向：cs.CV · 发布日期：2026-10-09