Active learning through label error statistical methods

作者：

Highlights：

• We define two label error statistics functions and build clustering-based practical statistical models to guide block splitting.

• We propose a center-and-edge instance selection strategy to choose critical instances.

• We design an algorithm called active learning through label error statistical methods (ALSE).

• Results of significance test verify the superiority of ALSE to state-of-the-art algorithms.

摘要

•We define two label error statistics functions and build clustering-based practical statistical models to guide block splitting.•We propose a center-and-edge instance selection strategy to choose critical instances.•We design an algorithm called active learning through label error statistical methods (ALSE).•Results of significance test verify the superiority of ALSE to state-of-the-art algorithms.

论文关键词：Active learning,Clustering,Label error statistical model,Probabilistic lipschitzness

论文评审过程：Received 31 May 2019, Revised 17 October 2019, Accepted 19 October 2019, Available online 24 October 2019, Version of Record 16 January 2020.

论文官网地址：https://doi.org/10.1016/j.knosys.2019.105140