Review of K-means Algorithm Optimization Based on Differential Privacy

Differential Privacy K-means Algorithm (DP K-means), as a privacy-preserving data mining (PPDM) model based on differential privacy technology, is simple, efficient and can guarantee the privacy of data. It has attracted the attention of researchers. The paper first expounds the principle of differential privacy K-means algorithm and privacy attack model to analyze the shortcomings of the algorithm. Then it discusses and analyzes from three perspectives, such as data preprocessing, privacy budget allocation, and cluster division. The advantages and disadvantages of DP K-means algorithm improvement research are summarized, and the relevant data sets and general evaluation indicators in the research are summarized. Finally, the challenging problems to be solved in the DP K-means algorithm improvement research are pointed out, and the DP K-means algorithm is prospected. The future development trend of means algorithm.