Preservation Of Patterns and Input-Output Privacy

Abstract breaches. To do so, the data custodian needs to transform its data. To determine the appropriate transforPrivacy preserving data mining so far has mainly mation, there are two critical considerations: minimizfocused on the data collector scenario where individu- ing disclosure versus minimizing outcome change (i.e., als supply their personal data to an untrusted collector the deviation between the outcome obtained by minin exchange for value. In this scenario, random per- ing the original data and the outcome of mining the turbation has proved to be very successful. An equally transformed data). Among the transformations studcompelling, but overlooked scenario, is that of a data ied in the literature, random perturbation is a domicustodian, which either owns the data or is explicitly nant approach, i.e., transforming data values by adding entrusted with ensuring privacy of individual data. In random noise in a principled way [2]. The more noise this scenario, we show that it i...

• Type: Conference
• Year: 2007
• Where: ICDE
• Authors: Shaofeng Bu, Laks V. S. Lakshmanan, Raymond T. Ng, Ganesh Ramesh
• Keywords: Data Collector Scenario, Database, ICDE 2007, Outcome Change, Versus Minimizing Outcome