Privacy Preserving Density-Based Outlier Detection

Outlier detection can find its tremendous applications in areas such as intrusion detection, fraud detection, and image processing. Among many outlier detection algorithms, LOF is a very important density-based algorithm in which one critical step is to find the k-distance neighbors. In some privacy preserving circumstances, the cooperation between data holders is necessary while the privacy of the participators should be guaranteed. In this paper, we focus on privacy preserving LOF. We propose a novel algorithm for privacy preserving k-distance neighbors search. Combining it with other secure multiparty computation techniques, we detect outliers by LOF in a privacy preserving way.