Image classification with multi-view multi-instance metric learning

作者：

Highlights：

• MVMIML integrates multi-view multi-instance features into metric learning framework.

• A new distance function is designed for bags to measures the relation between bags.

• Alternating iteration optimization algorithm is adopted to solve MVMIML.

• Comprehensive experiments verify the effectiveness of our proposed models.

摘要

•MVMIML integrates multi-view multi-instance features into metric learning framework.•A new distance function is designed for bags to measures the relation between bags.•Alternating iteration optimization algorithm is adopted to solve MVMIML.•Comprehensive experiments verify the effectiveness of our proposed models.

论文关键词：Metric learning,Multi-view learning,Multi-instance learning,Image classification

论文评审过程：Received 24 April 2021, Revised 17 August 2021, Accepted 16 October 2021, Available online 24 October 2021, Version of Record 3 November 2021.

论文官网地址：https://doi.org/10.1016/j.eswa.2021.116117