A classification framework for multivariate compositional data with Dirichlet feature embedding

作者：

Highlights：

• Dirichlet feature embedding is proposed for compositional data classification.

• The framework removes the constraints of compositional data and improves classification.

• Multivariate compositional data analysis with diverse number of parts is studied.

• Results suggest that Dirichlet feature embedding performs better than alr and ilr.

摘要

•Dirichlet feature embedding is proposed for compositional data classification.•The framework removes the constraints of compositional data and improves classification.•Multivariate compositional data analysis with diverse number of parts is studied.•Results suggest that Dirichlet feature embedding performs better than alr and ilr.

论文关键词：Multivariate compositional data,Classification,Feature embedding,Dirichlet distribution,Support vector machine

论文评审过程：Received 13 July 2020, Revised 11 November 2020, Accepted 13 November 2020, Available online 16 November 2020, Version of Record 26 November 2020.

论文官网地址：https://doi.org/10.1016/j.knosys.2020.106614