Differentially private correlation clustering

Correlation clustering is a widely used technique in unsupervised machine learning. Motivated by applications where individual privacy is a concern, we initiate the study of differentially private correlation clustering. We propose an algorithm that achieves subquadratic additive error compared to the optimal cost. In contrast, straightforward adaptations of existing non-private algorithms all lead to a trivial quadratic error. Finally, we give a lower bound showing that any pure differentially private algorithm for correlation clustering requires additive error of Ω(n).

• View at arXiv

Free Full Text ( Final Version , 366kb )

See Also
inProceedings Differentially private correlation clustering M. Bun (Mark), M. Eliáš (Marek) and J. Kulkarni (Janardhan)