Efficient Generalized Selective Private Function Evaluation With Applications in Biometric Authentication

In a Selective Private Function Evaluation (SPFE) protocol, the client privately computes some predefined function on his own input and on m out of server's n database elements. The authors propose two new generalized SPFE protocols that are based on the new crypto-computing protocol by Ishai and Paskin and an efficient CPIR. The first protocol works only for constant values of m, but has 2 messages, and is most efficient when m=1. The second SPFE protocol works for any m, has 4 messages, and is efficient for a large class of functionalities. They, then propose an efficient protocol for private similarity test, where one can compute how similar client's input is to a specific element in server's database, without revealing any information to the server.