A Density Peaking Clustering Algorithm for Differential Privacy Preservation

Abstract

The privacy protection problem in data mining has received increasingly attention and is a hot topic of current research. To address the problems of large accuracy loss and instability of clustering results of clustering algorithms under differential privacy protection requirements, a density peak clustering algorithm for differential privacy protection (DP-chDPC) is proposed. Firstly, the original DPC algorithm is improved, by using the dichotomy method to automatically determine the truncation distance to avoid the subjectivity of manual selection, and by setting the threshold of local density and center offset distance to automatically obtain the clustering center, which overcomes the uncertainty of the original DPC algorithm to select the clustering center based on the decision graph. Then, noise is added to the local density by using the Laplace mechanism to realize the differential privacy protection of the algorithm during the clustering analysis. Finally, the Chebyshev distance is used to replace the Euclidean distance to calculate the distance matrix, which reduces the interference on the clustering results after the algorithm adds noise, and reduces the loss of clustering accuracy, so that the stability of the algorithm is improved. The experimental results show that the DP-chDPC algorithm can effectively reduce the loss of clustering accuracy after the algorithm adds noise, and the clustering results are more stable.

The novel hierarchical clustering approach using self‐organizing map with optimum dimension selection

“…From the experiments performed and the results obtained (Tables 3 and 4 ), it is revealed that the ARI obtained on the given data sets is better than other approaches [ 15 , 18 , 28 , 29 , 30 ]. The other observed phenomenon is that whenever clustering is performed using the SOM, the most optimum 2D configuration would be preferable by considering the ST or CH score to obtain optimized results because the WM of the SOM is also very meaningful.…”

LoHDP: Adaptive local differential privacy for high‐dimensional data publishing

“…To address concerns regarding the LDA model and the potential exposure of textual information in the training process, Zhao et al 16 proposed several differential privacy LDA algorithms tailored to typical training scenarios. Chen et al 17 introduces a differential privacy‐preserving density peak clustering algorithm to address privacy protection concerns in data mining. Han et al 18 proposes a cluster‐based hierarchical federated learning framework with both differential privacy and secure aggregation.…”

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

“…Using the dichotomy method [30] to determine the cutoff distance d c can solve the subjectivity of the traditional method by setting the truncation percentage p, the main process is as follows:…”

“…Based on the literature [29,30], this paper proposes the use of cosine distance instead of Euclidean distance to measure the similarity of high-dimensional datasets, and the use of dichotomy to determine truncation distance adaptively. Aiming at the subjective selection of clustering centers, from the perspective of the decision graph, this paper establishes the constraint conditions of the decision graph statistics by Chebyshev inequality, and then the threshold of statistics ρ and δ are obtained, and the selection of clustering centers is achieved by adjusting the constraint parameters, the improved DPC algorithm based on Chebyshev inequality (CDPC) is proposed.…”

“…This methodology facilitates the selection of appropriate clustering centers. H. Chen et al proposed two improved density peak clustering algorithms based on differential privacy [29,30] with distinct perspectives. These approaches address various challenges, including weak clustering effects on highdimensional datasets, manual selection of clustering centers and truncation distances, as well as privacy leakage during the clustering analysis in the DPC algorithm.…”