Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Guided perturbation: towards private and accurate mining

Guided perturbation: towards private and accurate mining There have been two methods for privacy- preserving data mining: the perturbation approach and the cryptographic approach. The perturbation approach is typically very efficient, but it suffers from a tradeoff between accuracy and privacy. In contrast, the cryptographic approach usually maintains accuracy, but it is more expensive in computation and communication overhead. We propose a novel perturbation method, called guided perturbation . Specifically, we focus on a central problem of privacy-preserving data mining—the secure scalar product problem of vertically partitioned data, and give a solution based on guided perturbation, with good, provable privacy guarantee. Our solution achieves accuracy comparable to the cryptographic solutions, while keeping the efficiency of perturbation solutions. Our experimental results show that it can be more than one hundred times faster than a typical cryptographic solution. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png The VLDB Journal Springer Journals

Guided perturbation: towards private and accurate mining

The VLDB Journal , Volume 17 (5) – Aug 1, 2008

Loading next page...
 
/lp/springer_journal/guided-perturbation-towards-private-and-accurate-mining-yl5KYRveac

References (40)

Publisher
Springer Journals
Copyright
Copyright © 2008 by Springer-Verlag
Subject
Computer Science; Database Management
ISSN
1066-8888
eISSN
0949-877X
DOI
10.1007/s00778-007-0056-z
Publisher site
See Article on Publisher Site

Abstract

There have been two methods for privacy- preserving data mining: the perturbation approach and the cryptographic approach. The perturbation approach is typically very efficient, but it suffers from a tradeoff between accuracy and privacy. In contrast, the cryptographic approach usually maintains accuracy, but it is more expensive in computation and communication overhead. We propose a novel perturbation method, called guided perturbation . Specifically, we focus on a central problem of privacy-preserving data mining—the secure scalar product problem of vertically partitioned data, and give a solution based on guided perturbation, with good, provable privacy guarantee. Our solution achieves accuracy comparable to the cryptographic solutions, while keeping the efficiency of perturbation solutions. Our experimental results show that it can be more than one hundred times faster than a typical cryptographic solution.

Journal

The VLDB JournalSpringer Journals

Published: Aug 1, 2008

There are no references for this article.