摘要
针对迁移环境,提出了一种基于内部簇结构重构的迁移模糊聚类算法。首先,通过为每个对象分配权重,利用所有对象信息来描述内部簇结构。其次,为了重建目标域中的簇结构,应对目标域中对象稀疏分布或被噪声污染的问题,将源域和目标域联合在一起并重新计算目标域中每个对象的权重。第三,目标域中更新后的权重被认为是用于指导目标域学习的迁移知识。对不同类型数据集的实验结果表明,与其他类似算法相比,本研究所提出的迁移模糊聚类算法具有良好性能。
With regards to transfer environments in this study,it proposed a transfer fuzzy clustering algorithm based on inner-cluster-structure reconstruction.Firstly,all objects were used to capture the inner cluster structure by assigning a weight to each object.Secondly,in order to reconstruct the cluster structure in the target domain in which objects distribute sparsely or are contaminated by noises,the source domain and the target domain were jointed together and recalculated the weight of each object in the target domain.Thirdly,the updated weights in the target domain were considered as transfer knowledge that is used for guiding the target domain learning.Experimental results on different kinds of data sets indicated that the proposed fuzzy clustering algorithm has good performance compared with other similar algorithms.
作者
王锦
尹汪宏
WANG Jin;YIN Wang-hong(Department of Information and Intelligent Engineering,Anhui Vocational College of Electronics&InformationTechnology,Bengbu,233030,Anhui)
出处
《蚌埠学院学报》
2021年第5期46-53,共8页
Journal of Bengbu University
基金
安徽省高校优秀青年人才支持计划项目(gxyq2021281)
安徽省高校自然科学研究项目(KJ2020A1085),安徽省质量工程校企合作示范实训中心项目(2019xqsxzx56)
安徽省质量工程教学研究项目(2020jyxm0124)。
关键词
无监督学习
迁移学习
模糊聚类
unsupervised learning
transfer learning
fuzzy clustering