Privacy Preserving Naïve Bayes Classifier for Horizontally Distribution Scenario Using Un-trusted Third Party

Abstract

The aim of the classification task is to discover some kind of relationship between the input attributes and the output class, so that the discovered knowledge can be used to predict the class of a new unknown tuple. The problem of secure distributed classification is an important one. In many situations, data is split between multiple organizations. These organizations may want to utilize all of the data to create more accurate predictive models or classifier while revealing neither their training data nor the tuples to be classified. The Naive Bayes classifier is a simple but efficient baseline classifier. In this paper, we present a privacy preserving Naive Bayes classifier for horizontally partitioned data. Our three layer protocol uses an Un-trusted Third Party (UTP). We study how to calculate model parameters for privacy preserving three-layer Naive Bayes classifier for horizontally partitioned databases and communicate their intermediate results to the UTP not their private data. In our protocol, an UTP allows to meet privacy constraints and achieve acceptable performance.