基于马尔可夫逻辑网的关联规则迁移学习  

Association Rule Transfer Learning Based on Markov Logic Network

作　　者：张倩[1] 李海港[1] 李明[1] 程玉虎[1] 

机构地区：[1]中国矿业大学信息与电气工程学院,江苏徐州221116

出　　处：《信息与控制》2014年第6期715-721,共7页Information and Control

基　　金：国家自然科学基金资助项目(61273143);教育部新世纪优秀人才支持计划(NCET-10-0765);教育部高等学校博士学科点专项科研基金资助项目(20120095110025);江苏省自然科学基金资助项目(BK20130207)

摘　　要：针对源领域和目标领域共享知识是规则、结构和逻辑等关联规则的情况,提出一种基于马尔可夫逻辑网的关联规则迁移学习方法.首先利用伪对数似然函数将源领域中马尔可夫逻辑网表示的知识迁移到目标领域中,建立两个领域之间的关联;再通过对源领域进行自诊断、结构更新和目标领域搜索新子句,来优化映射得到的结构,进而适应目标领域的学习.实验结果表明,算法成功地映射了迁移知识,提高了学习模型的精确度.An association rule transfer learning method based on Markov logic networks is presented specific to the situation wherein shared knowledge between the source domain and the target domain is associated with knowl- edge containing rules, structure, and logic. Having applied this method by means of a pseudo log-likelihood function, the knowledge in the source domain expressed in a Markov logic network is transferred into the tar- get domain while the link between the two domains is established. By means of a self-diagnosis and structure update in the source domain and a new clause surf in the target domain, the mapped structure is optimized so that it can be adapted to learning in the target domain. The experimental results show that the given algorithm successfully maps the transferred knowledge, and improves the precision of the learning model.

关 键 词：迁移学习 马尔可夫逻辑网 关联规则 结构优化 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]