Probabilistic latent variable models for unsupervised many-to-many object matching

作者：

Highlights：

• We propose a probabilistic model for matching clusters in different domains without correspondence information.

• The proposed method can handle data with more than two domains, and the number of objects in each domain can be different.

• We extend the proposed method for a semi-supervised setting.

• We demonstrate that the proposed method achieve better matching performance than existing methods using synthetic and real-world data sets.

摘要

•We propose a probabilistic model for matching clusters in different domains without correspondence information.•The proposed method can handle data with more than two domains, and the number of objects in each domain can be different.•We extend the proposed method for a semi-supervised setting.•We demonstrate that the proposed method achieve better matching performance than existing methods using synthetic and real-world data sets.

论文关键词：Latent variable model,Mixture model,Object matching

论文评审过程：Received 3 March 2015, Revised 22 December 2015, Accepted 25 December 2015, Available online 5 February 2016, Version of Record 17 May 2016.

论文官网地址：https://doi.org/10.1016/j.ipm.2015.12.013