Multi independent latent component extension of naive Bayes classifier

作者：

Highlights：

• Adopting a flexible and parsimonious structure extension approach.

• Local conditional dependent features are clustered into non-overlapping components.

• Our model diminishes the underflow problem to consider the probability of rare events.

• We prove that the components perform independently during learning and classification.

摘要

•Adopting a flexible and parsimonious structure extension approach.•Local conditional dependent features are clustered into non-overlapping components.•Our model diminishes the underflow problem to consider the probability of rare events.•We prove that the components perform independently during learning and classification.

论文关键词：00-01,99-00,Naive Bayes classifier,Latent variable,Conditional mutual information,Conditional mutual dependency

论文评审过程：Received 14 June 2020, Revised 19 October 2020, Accepted 29 November 2020, Available online 24 December 2020, Version of Record 24 December 2020.

论文官网地址：https://doi.org/10.1016/j.knosys.2020.106646