基于博弈论的隐私保护分布式数据挖掘

隐私保护的分布式数据挖掘问题是数据挖掘领域的一个研究热点，而基于经济视角，利用博弈论的方法对隐私保护分布式数据挖掘进行研究只是处于初始阶段。基于收益最大化，研究了完全信息静态博弈下分布式数据挖掘中参与者(两方或多方)的策略决策问题，得出了如下结论:数据挖掘在满足一定的条件下，参与者(两方或多方)的准诚信攻击策略是一个帕累托最优的纳什均衡策略;在准诚信攻击的假设下，参与者(多方)的非共谋策略并不是一个纳什均衡策略。同时给出了该博弈的混合战略纳什均衡，它对隐私保护分布式数据挖掘中参与者的决策具有一定的理论和指导意义。

Privacy Preserving Distributed Data Mining Based on Game Theory

Privacy preserving distributed data mining has become an important issue in the data mining. Based on economic perspectives, game theory has been applied to privacy preserving data mining, which is a relatively new area of research. This paper studied the strategies of partics(two-party or multi-party) by using a complete information static game theory framework for the privacy preserving distributed data mining, where each party tries to maximize its own utility. Research results show that the semi-honest adversary strategy of partics(two-party or multi-party) is Pareto dominance and Nash equilibrium under certain conditions in distributed data mining; and non-collusion strategy of parties(multi-party) is not a Nash equilibrium under the assumption of semi-honest adversary behavior, then the mixed strategy Nash equilibrium was given. So this paper has some theoretical and practical implication for the strategy of partics in privacy preserving distributed data mining.