Sign in Subscribe
Privacy

Distributed Privacy-Preserving k-Means Algorithm Ensures Data Concealment

Strategic Insights

— 1 min read

Researchers from the Guilin University of Electronic Technology have developed a distributed privacy-preserving k-means algorithm, DTK-means, designed to address the issue of intermediate cluster details exposure in traditional k-means methods. The algorithm involves multiple users and two non-colluding servers, ensuring that neither the participants nor the servers have knowledge of the specific details. The scheme includes four algorithms for compute local centroids, data aggregation, compute global centroids, and output final cluster centroids, implemented using Paillier homomorphic encryption.

Key Takeaways:

  • The DTK-means algorithm is a distributed privacy-preserving k-means method addressing the issues of intermediate cluster details exposure and data concealment.
  • The algorithm involves multiple users and two non-colluding servers, ensuring that neither the participants nor the servers have knowledge of the specific details.
  • The scheme includes four algorithms for compute local centroids, data aggregation, compute global centroids, and output final cluster centroids, implemented using Paillier homomorphic encryption.
  • The complexity analysis and numerical experiments show that the algorithm has good efficiency.
  • The DTK-means algorithm can resist collusion attacks, even if one server colludes with all participants except one.
  • The participants can accurately perform k-means clustering using the hidden global centroids without any information loss.
  • The research was funded by the National Key R&D Program of China, National Natural Science Foundation of China (NSFC), National Natural Science Foundation of Guangxi Province, and Open Project Program of Guangxi Key Laboratory of Digital Infrastructure.
  • The research included the contributions of Xue-Feng Duan, Zeng-Ao Tang, Yong Ding, and Rong-Hua Liang from the Guilin University of Electronic Technology.

Statistics:

  • The DTK-means algorithm has a computational complexity of O(n) and communication complexity of O(m), demonstrating good efficiency.
  • The numerical experiments show that the algorithm can accurately conceal intermediate data and private data from all parties involved.
  • The algorithm can resist collusion attacks, even if one server colludes with all participants except one.

Sources:

  • Efficient Multi-party Privacy Preserving Federated k -means Based On Homomorphic Encryption. Information Sciences, 2025;717.
  • National Key R&D Program of China
  • National Natural Science Foundation of China (NSFC)
  • National Natural Science Foundation of Guangxi Province
  • Open Project Program of Guangxi Key Laboratory of Digital Infrastructure
  • Information Technology Newsweekly, November 4, 2025, p 241
  • Xue-Feng Duan, Zeng-Ao Tang, Yong Ding, and Rong-Hua Liang, Guilin University of Electronic Technology

Read more