On novelty detection for multi-class classification using non-linear metric learning

作者：

Highlights：

• A novelty detection method based on learning non-linear distances is proposed.

• Non-linear distances are learnt from datasets using a metric learning algorithm.

• Normal data is attracted to clusters, while abnormalities become isolated.

• Our approach is shown to outperform previous work on novelty detection.

摘要

•A novelty detection method based on learning non-linear distances is proposed.•Non-linear distances are learnt from datasets using a metric learning algorithm.•Normal data is attracted to clusters, while abnormalities become isolated.•Our approach is shown to outperform previous work on novelty detection.

论文关键词：Novelty detection,Outlier detection,Anomaly detection,Metric learning,Multi-class classification

论文评审过程：Received 24 December 2019, Revised 28 October 2020, Accepted 28 October 2020, Available online 1 November 2020, Version of Record 10 February 2021.

论文官网地址：https://doi.org/10.1016/j.eswa.2020.114193