A kernel semi-supervised distance metric learning with relative distance: Integration with a MOO approach
作者：
Highlights：
• A multi-objective optimization based kernel metric learning technique is developed.
• This approach utilizes only few labeled data for generating constraints.
• Our approach selects a good constraint subset for adjusting kernel matrix.
• Kernel matrix is utilized by K-means algorithm for labeling the dataset.
• Several validity indices are optimized for better partitioning of the dataset.
摘要
•A multi-objective optimization based kernel metric learning technique is developed.•This approach utilizes only few labeled data for generating constraints.•Our approach selects a good constraint subset for adjusting kernel matrix.•Kernel matrix is utilized by K-means algorithm for labeling the dataset.•Several validity indices are optimized for better partitioning of the dataset.
论文关键词：Semi supervised classification,Multi objective optimization,Bregman projection,Clustering,Metric learning
论文评审过程：Received 13 May 2018, Revised 27 December 2018, Accepted 29 December 2018, Available online 18 January 2019, Version of Record 11 February 2019.
论文官网地址：https://doi.org/10.1016/j.eswa.2018.12.051
