A New Method on Personalized Privacy Preserving Multi-classification

Abstract

Privacy-preserving data mining has become important since data mining has been widely used in many fields. Various privacy preserving techniques have been proposed to preserve the sensitive data. In this paper, we address two algorithms which can build classifiers accurately with less privacy disclosure in distributed system. These schemes can satisfy the different privacy disclosure level need of every client, which can meet clients' personalized needs. Besides this, our methods can be used for multi-classification.