摘要
多源域自适应是迁移学习中的一项重要技术,其目标是利用多个源域的知识来提升目标域的学习性能。然而,目前的多源域自适应方法大多关注于源域与目标域之间的差异,忽略了源域的选取问题。为了解决上述问题,提出了基于多目标优化的多源域自适应算法,使用多目标优化来增强各源域之间不相似域的效应和源域与目标域之间相似域的效应。此外,使用粒子群优化算法来优化以上两个目标。对五个基准的评估表明了所提出的模型的有效性。
Multi source domain adaptation is an important technology in transfer learning,with the goal of improving the learning performance of the target domain by utilizing knowledge from multiple source domains.However,current multi-source domain adaptation methods mostly focus on the differences between the source domain and the target domain,neglecting the selection of the source domain.To address the above issues,this paper proposes a multi source domain adaptive algorithm based on multi-objective optimization,which enhances the effects of dissimilar domains between source domains and similar domains between source and target domains.
出处
《工业控制计算机》
2024年第7期121-122,128,共3页
Industrial Control Computer
基金
贵州省教育厅自然科学研究项目(黔教技[2023]012号及061号)
贵州省科技计划项目(黔科合基础ZK[2022]195)。
关键词
深度学习
多源域自适应
粒子群优化算法
多目标优化
deep learning
multi source domain adaptation
particle swarm optimization
multi-objective optimization