5ZE-3 Anti-Collusion Privacy-Preserving Data Mining

概要

論文の詳細を見る
Up to now, numerous methods of privacy-preserving data mining have been proposed. Most of them are under the assumption of semi-honest and without collusion, which means that every party follows the protocol properly with the exception that it keeps a record of all its intermediate computations but does not share the record to any others. In this study, we consider the problem of multiparty secure computation of some functions on secure summations of data spread around multiple parties. In particular, we focus our attention on the problem of collusions, in which some parties may collude to deduce the privacy of other parties. We propose a new method satisfying a high level security, full-privacy. By this method, nothing in a party is revealed even though all other parties collude. This method is also an efficient one with running time of O(m), and it can solve a large number of problems in privacy-preserving data mining.
一般社団法人情報処理学会の論文
2010-03-08