A framework for private statistics between two participants

Private statistics between two parties is a sort of computation in which each party has their own private input (e.g., an integer) to a statistical function f of two inputs, and both want to know the result of applying the function to the two inputs, but neither wishes to disclose the value of his/her own input. At the end of the computation, the parties will learn only the result of f. Here we describe a new framework for implementing protocols for private statistical computations among two parties, such as the calculation of correlation, variance, kurtosis, and some others.