Reuse-centric k-means configuration

作者：

Highlights：

• Computation reuse is useful to accelerate k-means configuration.

• The proposed reuse-centric techniques can accelerate k-means configuration by 5-9X.

• Reusing distances does not change k-means clustering results.

• Reusing cluster centers causes only little disparity on the quality of k-means results.

摘要

•Computation reuse is useful to accelerate k-means configuration.•The proposed reuse-centric techniques can accelerate k-means configuration by 5-9X.•Reusing distances does not change k-means clustering results.•Reusing cluster centers causes only little disparity on the quality of k-means results.

论文关键词：K-means,Algorithm configuration,Computation reuse

论文评审过程：Received 17 September 2019, Revised 22 December 2020, Accepted 11 April 2021, Available online 16 April 2021, Version of Record 27 April 2021.

论文官网地址：https://doi.org/10.1016/j.is.2021.101787