Design and Development Of Privacy Preserving Data Model For Data Publishing

Abstract

Needs of distributed and data intensive applications are growing day by day. These applications involve distributed database, network services, security management and machine learning techniques to satisfy the customer requirements. In current scenarios aspects of security in machine learning applications motivates to explore the scope and utilization of Privacy Preserving Data Mining (PPDM) techniques. The PPDM is an application of data mining, which is utilize to mine multiparty data with privacy. The PPDM techniques are useful in a number of applications like, business, medical and engineering. The aim of presented study is to analyze PPDM techniques for the identification of different issues and challenges and to develop an efficient PPDM model to overcome the identified issues and challenges. newlineIn the presented study a review has been carried out to understand different PPDM techniques. To overcome the identified issues, various aspects of PPDM technique including data dimensionality, data sanitization, utility of PPDM processed information and applicability in real world applications needs to be addressed. newlineTo identify an efficient dimension reduction technique an experimental study has been carried out. The experimental study includes the analysis of Principal Component Analysis (PCA), Linear Discriminant Analysis (LDA), Kernel Principal Component Analysis (KPCA), and Correlation Coefficient (CC) techniques. The aim is to identify the efficient dimensionality reduction technique based on the comparative analysis of all the mentioned techniques. Based on the result analysis it is observed that the correlation coefficient based dimensionality reduction technique is not much influencing the performance of ML algorithms. newlineTo sanitize the data while maintaining privacy, security and data utility a modified random noise inclusion algorithm has been developed. The modified random noise newlineXXIV newlineinclusion algorithm is an extension of random noise based data sanitization algorithm.