Wu Jimmy Ming-Tai, Lin Jerry Chun-Wei, Viger Philippe Fournier, Djenouri Youcef, Chen Chun Hao, Li Zhong Cui
College of Computer Science and Engineering, Shandong University of Science and Technology, Qindao, Shandong, China.
School of Computer Science and Technology, Harbin Institute of Technology (Shenzhen), Shenzhen, China.
Math Biosci Eng. 2019 Feb 27;16(3):1718-1728. doi: 10.3934/mbe.2019082.
Privacy-preserving data mining has become an interesting and emerging issue in recent years since it can, not only hide the sensitive information but still mine the meaningful knowledge at the same time. Since privacy-preserving data mining is a non-trivial task, which is also concerned as a NP-hard problem, several evolutionary algorithms were presented to find the optimized solutions but most of them focus on considering a single-objective function with the pre-defined weight values of three side effects (). In this paper, we aim at designing a multiple objective particle swarm optimization method for hiding the sensitive information based on the density clustering approach (named CMPSO). The presented CMPSO is more flexible to select the most appropriate solutions for hiding the sensitive information based on user's preference. Extensive experiments are carried on two datasets to show that the designed CMPSO algorithm has good performance than the traditional single-objective evolutionary approaches in terms of three side effects.
近年来，隐私保护数据挖掘已成为一个有趣且新兴的问题，因为它不仅可以隐藏敏感信息，同时还能挖掘有意义的知识。由于隐私保护数据挖掘是一项艰巨的任务，也被视为一个NP难问题，因此提出了几种进化算法来寻找优化解决方案，但其中大多数都专注于考虑具有三个副作用预定义权重值的单目标函数。在本文中，我们旨在设计一种基于密度聚类方法的多目标粒子群优化方法来隐藏敏感信息（命名为CMPSO）。所提出的CMPSO能够根据用户偏好更灵活地选择最合适的解决方案来隐藏敏感信息。在两个数据集上进行了大量实验，结果表明，在三个副作用方面，所设计的CMPSO算法比传统的单目标进化方法具有更好的性能。
