Abstract:Density-based clustering algorithms such as DBSCAN, are highly effective in handling large datasets and identifying clusters of arbitrary shapes, playing a crucial role in data analysis fields like outlier detection and social networks. Outsourcing DBSCAN to the cloud brings substantial benefits but also raises major privacy concerns regarding the private input data of data owners. Existing private DBSCAN methods often face challenges of inefficiency or potential privacy leakage, hindering their practical deployment. To address these challenges, we introduce FSS-DBSCAN, a three-server MPC platform designed for outsourced private density-based clustering using function secret sharing (FSS). This solution guarantees clustering quality equivalent to plaintext algorithms, ensures comprehensive privacy protection, and achieves top-tier efficiency. The high performance of FSS-DBSCAN is driven by two pivotal strategies. First, we devise an MPC-friendly DBSCAN algorithm that is highly compatible with efficient secret-sharing-based cryptographic protocols and benefits from GPU acceleration. Second, we construct novel FSS-based protocols tailored for complex operations integral to our DBSCAN variant, such as Euclidean distance comparison and point assignment, and further optimize their computation through tensorization techniques. We implement our platform as an extensible system on top of PyTorch that leverages GPU hardware acceleration for cryptographic and tensorized operations. These innovations enable FSS-DBSCAN to significantly outperform ppDBSCAN (AsiaCCS 2021), reducing the clustering time for 5000 samples to approximately 2 hours, achieving an speed improvement.

PPA-DBSCAN: Privacy-preserving ρ-Approximate Density-based Clustering

Privacy Preserving Distributed DBSCAN Clustering

Effective Privacy Preserved Clustering Based on Voronoi Diagram

FSS-DBSCAN: Outsourced Private Density-Based Clustering via Function Secret Sharing

Privacy-Preserving DBSCAN Clustering over Vertically Partitioned Data.

Homomorphic Encryption Based Privacy Preservation Scheme for DBSCAN Clustering

A Quadtree Density Clustering Algorithm under Differential Privacy

A Parallel Varied Density-Based Clustering Algorithm with Optimized Data Partition

A privacy preserving clustering technique using hybrid data transformation method

A Statistical Information-Based Clustering Approach in Distance Space

A privacy-preserving clustering model based on density

Privacy-Preserving Affinity Propagation Clustering over Vertically Partitioned Data

An Improved Differentially Private DBScan Clustering Algorithm for Vehicular Crowdsensing

Clustering Algorithm For Privacy Preservation On Mapreduce

A Density Peaking Clustering Algorithm for Differential Privacy Preservation.

Privacy-preserving Constrained Spectral Clustering Algorithm for Large-Scale Data Sets.

An Improved Density Peak Clustering Algorithm Based on Chebyshev Inequality and Differential Privacy

Privacy preserving approximate k-means clustering

Distributed Clustering in the Anonymized Space with Local Differential Privacy

Private Hierarchical Clustering and Efficient Approximation

Distributed Clustering Algorithm Based on Privacy Protection