Graph-based Clustering under Differential Privacy

2018/03/10 by Pinot, Rafael, Morvan, Anne, Yger, Florian +2
#Data Structures and Algorithms (cs.DS) #FOS: Computer and information sciences #Machine Learning (cs.LG)

paper · doi:10.48550/arxiv.1803.03831

Abstract

In this paper, we present the first differentially private clustering method for arbitrary-shaped node clusters in a graph. This algorithm takes as input only an approximate Minimum Spanning Tree (MST) T released under weight differential privacy constraints from the graph. Then, the underlying nonconvex clustering partition is successfully recovered from cutting optimal cuts on T. As opposed to existing methods, our algorithm is theoretically well-motivated. Experiments support our theoretical findings.

Related

• Minimum spanning tree release under differential privacy constraints
• Locally Differentially Private Graph Clustering via the Power Iteration Method
• Scalable Differentially Private Clustering via Hierarchically Separated Trees
• A Differentially Private Clustering Algorithm for Well-Clustered Graphs
• Differentially Private Spectral Graph Clustering: Balancing Privacy, Accuracy, and Efficiency