Distributed Privacy-Preserving Minimal Distance Classification
The paper focuses on the problem of preserving privacy for a minimal distance classifier working in the distributed environment. On the basis of the study of available works devoted to privacy aspects of machine learning methods, we propose the novel definition and taxonomy of privacy. This taxonomy was used to develop new effective classification algorithms which can work in distributed computational environment and assure a chosen privacy level. Instead of using additional algorithms for secure computing, the privacy assurance is embedded in the classification process itself. This lead to a significant reduction of the overall computational complexity what was confirmed by the computer experiments which were carried out on diverse benchmark datasets.
Keywordsprivacy preserving distributed data mining classification k-NN
Unable to display preview. Download preview PDF.
- 2.Asuncion, A., Newman, D.J.: UCI Machine Learning Repository. University of California, School of Information and Computer Science, Irvine (2007), http://www.ics.uci.edu/~mlearn/MLRepository.html
- 3.Chitti, S., Liu, L., Xiong, L.: Mining Multiple Private Databases using Privacy Preserving kNN Classifier, Technical Reports TR-2006-008, Emory University (2006)Google Scholar
- 4.Clifton, C., Kantarcioglu, M., Vaidya, J., Lin, X., Zhu, M.Y.: Tools for privacy preserving data mining. SIGKDD Explorations, 28–34 (2002)Google Scholar
- 7.Han, J.: Data Mining: Concepts and Techniques. Morgan Kaufmann Publ. Inc., San Francisco (2005)Google Scholar
- 11.Nissenbaum, H.: Can we Protect Privacy in Public? In: Computer Ethics Philosophical Enquiry ACM/SIGCAS Conference, Rotterdam, The Netherlands (1997)Google Scholar
- 12.Westin, A.F.: Privacy and Freedom. The Bodley Head Ltd. (1970)Google Scholar