A COMPLETE PRIVACY PRESERVATION SYSTEM FOR DATA MINING USING FUNCTION APPROXIMATION

Data privacy has become the primary concern in the current scenario as there are many pioneering methods for efficient mining of data. There are many algorithms to preserve privacy and handle the trade-off between privacy and utility. The ultimate goal of these algorithms is to anonymize the data without reducing the utility of them. A Privacy preserving procedure should have a minimum execution time, which is the overhead of clustering algorithms implemented using classical methods. There is also no single procedure that completely handles the trade-off and also updates itself automatically. In this work, the anonymization is implemented using Radial Basis Function [RBF] network, which provides both maximum privacy and utility with a proper tuning parameter specified between privacy and utility. The network also updates itself when the trend of data changes by controlling the maximum amount of error with a threshold value.