Data Perturbation Methods Implementation Using Matlab

Abstract:

One important class of data perturbation methods is the class of additive data perturbation methods for masking sensitive datasets containing linear relationships among their variables. Theses methods try to maintain original linear relationships in masked datasets while preserving data security and privacy. Three important methods under this class are GADP (Muralidhar et al. 1999), IPSO (Burridge 2003) and EGADP (Muralidhar and Sarathy 2005). To facilitate the work in this important area of research, we provide brief discussion of each of the three methods along with their Matlab implementations.