PE-means: Improved Differentially Private $k$-means Clustering through Private Evolution
Researchers have developed PE-means, a new algorithm for differentially private k-means clustering. This method improves upon existing techniques by using a private histogram with constant sensitivity, rather than directly summing private data. PE-means achieves an average 20% reduction in clustering loss compared to current state-of-the-art methods. AI
IMPACT Introduces a more efficient method for private clustering, potentially improving data privacy in machine learning applications.