|Title||Privacy-preserving Multiparty Collaborative Mining with Geometric Data Perturbation|
|Publication Type||Journal Article|
|Year of Publication||2009|
|Authors||Keke Chen, Ling Liu|
|Journal||IEEE Transactions on Parallel and Distributed Systems|
In multiparty collaborative data mining, participantscontribute their own datasets and hope to collaborativelymine a comprehensive model based on the pooled dataset. Howto efficiently mine a quality model without breaching eachparty's privacy is the major challenge. In this paper, we proposean approach based on geometric data perturbation and datamining-service oriented framework. The key problem of applyinggeometric data perturbation in multiparty collaborative miningis to securely unify multiple geometric perturbations that arepreferred by different parties, respectively. We have developedthree protocols for perturbation unification. Our approach hasthree unique features compared to the existing approaches. (1)With geometric data perturbation, these protocols can work formany existing popular data mining algorithms, while most ofother approaches are only designed for a particular mining algorithm.(2) Both the two major factors: data utility and privacyguarantee are well preserved, compared to other perturbationbasedapproaches. (3) Two of the three proposed protocols alsohave great scalability in terms of the number of participants,while many existing cryptographic approaches consider only twoor a few more participants. We also study different features of thethree protocols and show the advantages of different protocolsin experiments.
|Full Text|| |
Keke Chen and Ling Liu, "Privacy-preserving Multiparty Collaborative Mining with Geometric Data Perturbation," IEEE Transactions on Parallel and Distributed Systems, 2009.