A Differentially Private K-means Clustering Scheme for Smart Grid

Shuai Guo ； Mi Wen ； Xiaohui Liang

《International Journal of Network Security》 23卷1期 (2021/01) Pp. 126-134

https://doi.org/10.6633/IJNS.202101_23(1).15

摘要

Cluster analysis via data mining is of great significance to smart grid for enabling power load analysis. However, the privacy-sensitive electricity consumption data may be leaked in the process of data mining. To address this problem, privacy-preserving data mining has been widely studied. This paper proposes an improved differentially private K-means clustering scheme while considering the unique characteristics of smart grid data. Using dimensionality reduction, the proposed scheme improves the effectiveness of data mining while preserving the data privacy. Performance evaluations are further conducted to illustrate that the proposed scheme outperforms the existing differential privacy preserving k-means clustering schemes.

關鍵字

Differential Privacy ； Dimensionality Reduction ； K-means Clustering ； Smart Grid