Loading...

Secure- multiparty Computation Protocol for Privacy Preserving Data Mining

Maftouni, Mahya | 2015

1694 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 48768 (19)
  4. University: Sharif University of Technology
  5. Department: Computer Engineering
  6. Advisor(s): Amini, Morteza
  7. Abstract:
  8. Privacy preserving data mining helps organizations and companies not only to deal with privacy concerns of customers and regular limitations, but also to benefit from collaborative data mining. Utilizing cryptographic techniques and secure multiparty computation (SMC) are among widely employed approaches for preserving privacy in distributed data mining. The general purpose of secure multiparty computation protocols to compute specific functions on private inputs of parties in a collaborative manner and without revealing their private inputs. Providing rigorous security proof of secure multiparty computation makes it a good choice for privacy preservation, despite of its cryptographic overheads. Self-Organizing Map (SOM) is a kind of competitive learning neural network, which is frequently used for clustering applications, such as anomaly detection. To the best of our knowledge, there are no considerable conducted research studies in privacy preserving SOM, except for the protocol proposed by Han et al. In this thesis a secure protocol for privacy preserving SOM has been proposed. The proposed protocol uses an additive homomorphic encryption scheme for computing square of Euclidean distance. Performance analysis and implementation results show that secure computation of Euclidean distance in our proposed protocol has linear complexity with respect to the number of parties, in contrast with the Han’s protocol which has quadratic complexity
  9. Keywords:
  10. Self-Organizing Map (SOM) ; Data Mining ; Privacy Preserving Date Mining ; Secure Multiparty Computation (SMC) ; Hommorphic Encryption

 Digital Object List