K-Means Clustering With Local dᵪ-Privacy for Privacy-Preserving Data Analysis

Abstract

Privacy-preserving data analysis is an emerging area that addresses the dilemma of performing data analysis on user data while protecting users' privacy. In this paper, we consider the problem of constructing privacy-preserving K-means clustering protocol for data analysis that provides local privacy to users' data. To enable a desirable degree of local privacy guarantee while maintaining high accuracy of the clustering, we adopt a generalized differential privacy definition, dχ-privacy, which quantifies the distinguishability level based on the distance between data records defined by the distance function dχ. In our work, we consider the space of data points as a metric space imbued with Euclidean distance and propose a bounded perturbation mechanism (BPM) with bounded sampling space of the perturbed data points, which is formally shown to achieve dχ-privacy. BPM perturbs the data as a whole instead of treating each dimension independently, which is desirable since the privacy budget is no longer required to be split among different dimensions. Bounded output space also means that we will not get into the case where the report or the statistical result is so far out of the data domain that it is hard to interpret. Furthermore, it can also help in limiting the amount of bandwidth needed to send such report to the server. The design of BPM is based on a probability density function which decreases exponentially as the Euclidean distance with respect to the true value grows. It is also designed with the aim of ensuring that BPM produces perturbed data that provides the claimed privacy guarantee while ensuring high utility response. To guarantee the efficiency of the perturbation method, we propose an efficient algorithm to sample from the proposed distribution and apply BPM to the design of dχ-private K-means clustering algorithms. Lastly, we analyse the privacy and utility guarantee provided by the proposed method and provide its experimental results.

Efficient FHE-based Privacy-Enhanced Neural Network for Trustworthy AI-as-a-Service

“…The test set contain 37611 trial pairs, and the task is to verify whether the trial pair of utterances are from the same speaker. We use MFA-Conformer model 5 to do the feature extraction. After obtaining the embedding feature vector of the input audio, the vector is encrypted and send to the private network.…”

“…Typically, AI models, such as deep neural networks (DNNs), apply a sequence of evaluations on the input data and model parameters to obtain an inference output. Many techniques are studied for privacy-preserving machine learning, such as differential privacy [4], [5] and federated learning [6]- [9]. These techniques address the concern of privacy issues of the training data and of the data sharing during model training.…”

K-Means Clustering with Local Distance Privacy

“…References [21,22] focused on providing local privacy for users who participate in the K-means clustering analysis. The general idea of GoodCenters is utilizing some hash functions to map the original data records to another space where the collision happens with a high chance to users' data records being closer to each other and less chance to the data records far away from each other.…”

A Multi-Agent Deep-Reinforcement-Learning-Based Strategy for Safe Distributed Energy Resource Scheduling in Energy Hubs

“…Particularly, when agents engage in data transactions with external power and gas grids, the internal parameters of the agents become more susceptible to leakage. In order to safeguard data privacy in EH, we adopt an efficient and computationally simple approach known as local differential privacy [49][50][51][52][53][54]. This approach not only allows for quantifying the strength of privacy protection but also enables the application of the noise addition process at each EH node.…”