SDCOR: Scalable density-based clustering for local outlier detection in massive-scale datasets

作者：

Highlights：

• A novel scalable approach for local outlier detection in massive data is presented.

• The input data is not required to be entirely loaded in the memory as it is processed in chunks.

• Our assessments prove that the proposed method has a linear time complexity with a low constant.

• The proposed approach is superior to the state-of-the-art density- and distance-based methods.

摘要

•A novel scalable approach for local outlier detection in massive data is presented.•The input data is not required to be entirely loaded in the memory as it is processed in chunks.•Our assessments prove that the proposed method has a linear time complexity with a low constant.•The proposed approach is superior to the state-of-the-art density- and distance-based methods.

论文关键词：Local outlier detection,Massive-scale datasets,Scalable,Density-based clustering,Anomaly detection

论文评审过程：Received 30 June 2020, Revised 21 June 2021, Accepted 24 June 2021, Available online 1 July 2021, Version of Record 6 July 2021.

论文官网地址：https://doi.org/10.1016/j.knosys.2021.107256