To Actively Initialize Active Learning

作者：

Highlights：

• The initialization problem of active learning is addressed, which is how to find a set of labeled samples which contains at least one instance per category.

• A new active initialization criterion, the Nearest Neighbor Criterion, is proposed for the initialization task.

• The impacts of different initialization strategies on the whole active learning process are further investigated.

• Excellent performance of the proposed method in comparison with state-of-the-art approaches is demonstrated.

摘要

•The initialization problem of active learning is addressed, which is how to find a set of labeled samples which contains at least one instance per category.•A new active initialization criterion, the Nearest Neighbor Criterion, is proposed for the initialization task.•The impacts of different initialization strategies on the whole active learning process are further investigated.•Excellent performance of the proposed method in comparison with state-of-the-art approaches is demonstrated.

论文关键词：active learning,active initialization,nearest neighbor criterion,minimum nearest neighbor distance

论文评审过程：Received 2 November 2021, Revised 2 June 2022, Accepted 3 June 2022, Available online 12 June 2022, Version of Record 18 June 2022.

论文官网地址：https://doi.org/10.1016/j.patcog.2022.108836