TL-SVM_一种迁移学习新算法.pdfVIP

第29 卷 第1 期 控 制 与 决 策 2014 年 1 月 Vol. 29 No. 1 Control and Decision Jan. 2014 文章编号: 1001-0920 (2014) 01-0141-06 DOI: 10.13195/j.kzyjc.2012.1450 TL-SVM: 一种迁移学习算法 1,2 1 1 许 敏 , 王士同 , 顾 鑫 (1. 江南大学数字媒体学院，江苏无锡214122 ；2. 无锡职业技术学院物联网技术学院，江苏无锡214121) 摘 要: 迁移学习旨在利用大量已标签源域数据解决相关但不相同的目标域问题. 当与某领域相关的新领域出现 时, 若重新标注新领域, 则样本代价昂贵, 丢弃所有旧领域数据又十分浪费. 对此, 基于SVM 算法提出一种新颖的迁 移学习算法—– TL-SVM, 通过使用目标域少量已标签数据和大量相关领域的旧数据来为目标域构建一个高质量的 分类模型, 该方法既继承了基于经验风险最小化最大间隔SVM 的优点, 又弥补了传统SVM 不能进行知识迁移的缺 陷. 实验结果验证了该算法的有效性. 关键词: 迁移学习；分类；支持向量机 中图分类号: TP273 文献标志码: A TL-SVM: A transfer learning algorithm 1,2 1 1 XU Min , WANG Shi-tong , GU Xin (1. School of Digital Media，Jiangnan University ，Wuxi 214122 ，China ；2. School of Internet of Things Engineering ，Wuxi Institute of Technology ，Wuxi 214121 ，China ．Correspondent ：XU Min ，E-mail ：xum@wxit. ) Abstract: Transfer learning(TL) aims to solve related but different target domain problems by using plenty of labeled source domain data. When the task from one new domain comes, new domain samples are relabeled costly, and it would be a waste to discard all the old domain data. Therefore, an algorithm called TL-SVM based on the SVM algorithm is proposed, which uses a small amount of target domain tag data and a large number of source domain old data to build a high-quality classification model. The method inherits the advantages of the maximum interval SVM based on empirical risk minimization and makes up for the defects that traditional SVM can not migrate knowledge. Experiment