Memoryless inference control methods have been shown to provide effective means of reducing the amount of sensitive information released from a statistical database while maximizing the release of non-sensitive information. Early memoryless inference controls have the additional benefit of providing control at a low computation and storage cost. Two recent extensions to memoryless inference control allow the controls to release more non-sensitive information but do so at a greater cost in terms of computation and storage. This paper describes a proposed hybrid inference control method that can potentially maximize the release of information while holding down computation costs.
/lp/association-for-computing-machinery/hybrid-inferential-security-methods-for-statistical-databases-4s1oG0GwGf