Collaborative Differentially Private Outlier Detection for Categorical Data

Collaborative analytics is crucial to extract value from data collected by different organizations and stored in separate silos. However, privacy and legal concerns often inhibit the integration and joint analysis of data. One of the most important data analytics tasks is that of outlier detection, which aims to find abnormal entities that are significantly different from the remaining data. In this paper, we define privacy in the context of collaborative outlier detection and develop a novel method to find outliers from horizontally partitioned categorical data in a privacy-preserving manner. Our method is based on a scalable outlier detection technique that uses attribute value frequencies. We provide an end-to-end privacy guarantee by using the differential privacy model and secure multiparty computation techniques. Experiments on real data show that our proposed technique is both effective and efficient.