LFM: Leveraging Foundation Models for Source-Free Universal Domain Adaptation
LFM:利用基础模型进行无源通用域适应
机构 * School of Computer Science and Engineering, Tianjin University of Technology(天津理工大学计算机科学与工程学院) ; Engineering Research Center of Learning-Based Intelligent System, Ministry of Education of the People’s Republic of China, Tianjin University of Technology(中华人民共和国教育部基于学习的智能系统工程研究中心,天津理工大学) ; School of Artificial Intelligence, Tianjin University(天津大学人工智能学院) ; Engineering Research Center of City Intelligence and Digital Governance, Ministry of Education of the People’s Republic of China, Tianjin University(中华人民共和国教育部城市智能与数字治理工程研究中心,天津大学)
专题命中 VLM训练与架构 :vision-language model(abstract);VLM(abstract);分类 cs.CV、cs.LG
AI总结 研究在无源数据时将预训练源模型适应到目标域的问题,提出LFM框架,利用视觉语言模型计算相似度确定标签转移类型、识别未知样本,通过共识策略精炼伪标签训练目标模型,实验验证了该框架的有效性和优越性。
Comments Accepted by IEEE Transactions on Multimedia (2026)