Graph regularized multiset canonical correlations with applications to joint feature extraction

作者：

Highlights：

• We propose a novel algorithm called GrMCC for joint feature extraction.

• GrMCC considers both discriminative and intrinsic geometrical structure in multi-representation data.

• The extracted features by GrMCC have strong discriminant power for recognition.

• Experimental results show GrMCC can provide encouraging recognition results in contrast to the state-of-the-art algorithms.

摘要

•We propose a novel algorithm called GrMCC for joint feature extraction.•GrMCC considers both discriminative and intrinsic geometrical structure in multi-representation data.•The extracted features by GrMCC have strong discriminant power for recognition.•Experimental results show GrMCC can provide encouraging recognition results in contrast to the state-of-the-art algorithms.

论文关键词：Pattern recognition,Canonical correlation analysis,Multiset canonical correlations,Graph embedding,Feature extraction

论文评审过程：Received 23 February 2013, Revised 24 March 2014, Accepted 19 June 2014, Available online 30 June 2014.

论文官网地址：https://doi.org/10.1016/j.patcog.2014.06.016