Two Wavelet-Based Data Distortion Methods for Privacy Preserving Classification

With rapid development of modern data collection and data warehouse technologies, data mining is becoming more and more a standard practice. Accompanying this trend, preserving privacy in certain data becomes a challenge for data mining applications in many fields, especially in medical, financial and homeland security fields. The main objective of data mining is generalization of information and not to popularization of private data. We can implement Data mining and maintain the privacy preserving capability of data simultaneously. Data mining can deal with private data privately. Therefore, the true problem is not data mining, but the way of doing it. Privacy of preserving data mining algorithms has been recently introduced with the aim of preventing the discovery of sensible information. This paper introduces a privacy-preserving data distortion method in the collaborative analysis situations based on wavelet transformation, which provide an efficient balance between data utilities and privacy preserving beyond its fast run time. In suggested method a dynamic technique was used for engaging between criteria of privacy and accuracy. It provides data owners the ability to make decision for each attribute independently in terms of level of their sensitivity. Wavelet transform and inverse wavelet transform keep Euclidean distance and quality of man data, in this paper KNN algorithm was used for classifying data which are compatible with Euclidean distance for keeping statistical features of data. Experiments on real-life datasets indicate high potential of the suggested method in terms of accuracy, run time and level of privacy.

Keywords:

cardiovascular disease , neural network , learning algorithms