Skip to main content
placeholder image

Effects of normalization on rotation based perturbation techniques for privacy preservation

Conference Paper


Abstract


  • Data is increasing rapidly with the vast development of technologies. Mining is mandatory to retrieve knowledge from these data. But, the data may contain sensitive information of the individuals such as medical diagnostic reports which they don't want to reveal. Privacy Preserving Data Mining (PPDM) helps in this regard to keep sensitive data private as well as preserving the data utility. Besides other PPDM techniques, rotation based perturbation contributes to satisfying both aspects of PPDM i.e., individuals' privacy and data utility. Also, normalization techniques have a significant impact in the field of PPDM. In this work, the effects of three well-known normalization approaches (i.e., z-score, min-max and decimal scaling) on rotation based perturbation techniques called 2-Dimensional Rotation Transformation (2DRT) and 3-Dimensional Rotation Transformation (3DRT), are analyzed. The effects of normalization over rotation based perturbation techniques are investigated using three benchmark classifiers with five UCI data set. The empirical analysis elucidates that 3DRT provides significant privacy preservation and data utility maintenance than 2DRT for z-score normalization.

Publication Date


  • 2021

Citation


  • Paul, M. K., & Islam, M. R. (2021). Effects of normalization on rotation based perturbation techniques for privacy preservation. In 2021 International Conference on Automation, Control and Mechatronics for Industry 4.0, ACMI 2021. doi:10.1109/ACMI53878.2021.9528134

Scopus Eid


  • 2-s2.0-85115695899

Web Of Science Accession Number


Abstract


  • Data is increasing rapidly with the vast development of technologies. Mining is mandatory to retrieve knowledge from these data. But, the data may contain sensitive information of the individuals such as medical diagnostic reports which they don't want to reveal. Privacy Preserving Data Mining (PPDM) helps in this regard to keep sensitive data private as well as preserving the data utility. Besides other PPDM techniques, rotation based perturbation contributes to satisfying both aspects of PPDM i.e., individuals' privacy and data utility. Also, normalization techniques have a significant impact in the field of PPDM. In this work, the effects of three well-known normalization approaches (i.e., z-score, min-max and decimal scaling) on rotation based perturbation techniques called 2-Dimensional Rotation Transformation (2DRT) and 3-Dimensional Rotation Transformation (3DRT), are analyzed. The effects of normalization over rotation based perturbation techniques are investigated using three benchmark classifiers with five UCI data set. The empirical analysis elucidates that 3DRT provides significant privacy preservation and data utility maintenance than 2DRT for z-score normalization.

Publication Date


  • 2021

Citation


  • Paul, M. K., & Islam, M. R. (2021). Effects of normalization on rotation based perturbation techniques for privacy preservation. In 2021 International Conference on Automation, Control and Mechatronics for Industry 4.0, ACMI 2021. doi:10.1109/ACMI53878.2021.9528134

Scopus Eid


  • 2-s2.0-85115695899

Web Of Science Accession Number