SVM based multi-label learning with missing labels for image annotation

作者:

Highlights:

• Our loss function guarantees the large margin and minimum number of samples which live in margin area.

• Our approach takes into account both example smoothness and label consistence when learning the mapping function in SVM.

• We propose a SVM based method for multi-label learning with missing label problems.

摘要

•Our loss function guarantees the large margin and minimum number of samples which live in margin area.•Our approach takes into account both example smoothness and label consistence when learning the mapping function in SVM.•We propose a SVM based method for multi-label learning with missing label problems.

论文关键词:Multi-label learning,Missing labels,SVM,Image annotations

论文评审过程:Received 27 December 2016, Revised 5 November 2017, Accepted 24 January 2018, Available online 1 February 2018, Version of Record 8 February 2018.

论文官网地址:https://doi.org/10.1016/j.patcog.2018.01.022