Generalized RBF kernel for incomplete data

作者：

Highlights：

• We construct a generalization of RBF kernel to incomplete data.

• It is the expected value of classical RBF.

• We do not perform any imputations but model missing values by probability distributions.

• It is easy to implement and can be used together with any kernel method.

摘要

•We construct a generalization of RBF kernel to incomplete data.•It is the expected value of classical RBF.•We do not perform any imputations but model missing values by probability distributions.•It is easy to implement and can be used together with any kernel method.

论文关键词：Incomplete data,Missing values,SVM,Kernel methods,RBF

论文评审过程：Received 22 February 2018, Revised 22 December 2018, Accepted 24 February 2019, Available online 5 March 2019, Version of Record 21 March 2019.

论文官网地址：https://doi.org/10.1016/j.knosys.2019.02.034