School of Information and Communication Engineering, Guilin University of Electronic Technology, China

Biography: ZhaoYu SHOU is an associate professor and PhD candidate at School of Information and Communication Engineering, Guilin University of Electronic Technology. He received his Master’s degree from Guilin University of Electronic Technology in 2004. His research interest covers pattern recognition and big data engineering. Currently more attention is being paid to data mining and outlier detection.

Speech Title: A outlier detection algorithm based on differential privacy

Abstract: Aiming at the problem that personal privacy is vulnerable to damage during outlier detection, this paper proposes an outlier detection method based on differential privacy. The algorithm uses prim algorithm to generate minimum spanning tree (MST), adds Laplace noise to the weight of the edge of the minimum spanning tree. At the same time, by combining the degree of dissimilarity and reverse similarity number, a new anomaly judgment method is proposed, which improves the outlier detection rate and effectively resists background knowledge attacks. The experimental analysis shows that the algorithm can effectively protect the sensitive attributes of the data, improve the true positive rate(TPR) of outlier detection and reduce the false positive rate(FPR).

Keywords: outlier detection,differential privacy,MST