Hierarchical combining of classifiers in privacy preserving data mining

Abstract

In privacy preserving classification there are several different types of classifiers with different parameters. We cannot point out the best type of classifiers and its default parameters. We propose the new solution in privacy preserving classification, namely a framework for combinig classifiers trained over data with preserved privacy - the hierarchical combining of classifiers. This solution enables a miner to obtain better results than those achieved with single classifiers. © 2014 Springer International Publishing.