Credal classification rule for uncertain data based on belief functions

作者：

Highlights：

• A new credal classification rule (CCR) is proposed to deal with uncertain data.

• A justified use of meta-class is introduced to model the imprecision of data.

• A new method is presented for the determination of bba׳s of the object.

• The computation burden of CCR is quite low.

• The experiments with real data are given to illustrate the interest of CCR.

摘要

Highlights•A new credal classification rule (CCR) is proposed to deal with uncertain data.•A justified use of meta-class is introduced to model the imprecision of data.•A new method is presented for the determination of bba׳s of the object.•The computation burden of CCR is quite low.•The experiments with real data are given to illustrate the interest of CCR.

论文关键词：Credal classification,Data classification,Belief functions,Evidence theory,Uncertain data

论文评审过程：Received 24 May 2013, Revised 23 September 2013, Accepted 23 January 2014, Available online 3 February 2014.

论文官网地址：https://doi.org/10.1016/j.patcog.2014.01.011