Date: 2012/9/10

      Place: Collaboration Room #7 15:30 -

      Name: Hiroaki Kikuchi
          Tokai University

      Title: Trend and principle - privacy-preserving data mining

      PPDM is a methodology to allow distributed parties to perform a variety of data mining algorithm without revealing each dataset. Studies are classified into 1) cryptographic approach, using homomorphic encryption and secure multiparty computation, 2) random perturbation approach, based on Bayes theorem and randomization algorithms, 3) k-anonymity model, 4) downgrading. After presenting the principals for these approaches, some update topics is given.