Private Function Evaluation for MPC

Assume that you invent an algorithm that is able to diagnose different diseases very efficiently based on some general inputs about a person’s health. Clearly, this would be valuable to people and different healthcare institutions would like to apply this in their everyday work. On the other hand, you do not want to hand the algorithm over as it could clearly bring you a lot of money if everyone has to buy this service from you. However, the medical institutions can not just give you the data about their patients. Hence, in this situation you would like to keep your algorithm secret and they also need to keep their inputs secret, but you are both interested in still evaluating this algorithm with the given inputs. This is exactly the problem solved by private function evaluation. Circuit hiding multi-party computation (MPC) or private function evaluation (PFE) is a special additional case of multi-party computation where a private function f needs to be evaluated on private inputs xi to obtain f(x1, . . . , xm). In the concrete setting, we consider the case where one party inputs the function and all parties can provide the inputs. This paper introduces a framework for this kind of computation based on [MS13]. The main body of this overview is divided to two parts. First main part in Sec. 3 introduces the ideas put forth by [MS13] whereas the second part in Sec. 4 writes out the concrete protocols required for secure multi-party boolean circuit evaluation. These are supported by the preliminaries in Sec. 2 and summary of related work in Sec. 5