Clustering via binary embedding

作者：

Highlights：

• A novel clustering approach is proposed.

• The approach encodes the objects to be clustered with binary signatures.

• The binary signatures are extracted through a set of one-class models.

• The approach is agnostic to the shape of clustering.

• The proposed method favourably compares with state of the art.

摘要

•A novel clustering approach is proposed.•The approach encodes the objects to be clustered with binary signatures.•The binary signatures are extracted through a set of one-class models.•The approach is agnostic to the shape of clustering.•The proposed method favourably compares with state of the art.

论文关键词：Clustering,Binary embedding,Finite mixture models,Biclustering,One-class classification

论文评审过程：Received 24 August 2017, Revised 30 April 2018, Accepted 13 May 2018, Available online 14 May 2018, Version of Record 24 May 2018.

论文官网地址：https://doi.org/10.1016/j.patcog.2018.05.011