Independent component analysis by lp-norm optimization

作者：

Highlights：

• We propose a couple of algorithms for independent component analysis using lp-norm optimization.

• By the use of the generalized Gaussian function, the problem of ICA was shown to be equivalent to the optimization of lp-norm.

• The proposed method is robust to the outliers especially for super-Gaussian sources.

摘要

•We propose a couple of algorithms for independent component analysis using lp-norm optimization.•By the use of the generalized Gaussian function, the problem of ICA was shown to be equivalent to the optimization of lp-norm.•The proposed method is robust to the outliers especially for super-Gaussian sources.

论文关键词：ICA,PCA,lp-Norm,Maximum likelihood estimation,Super-Gaussian,Sub-Gaussian

论文评审过程：Received 18 October 2016, Revised 19 September 2017, Accepted 7 October 2017, Available online 13 October 2017, Version of Record 8 January 2018.

论文官网地址：https://doi.org/10.1016/j.patcog.2017.10.006