Speaker: Aggelos Kiayias
Day:  Wednesday, 1/30/2008
Room: ITEB 336
Time: 1:00-2:00pm

Title: How to Collect a Function Securely?

Consider a set of private values x1, ..., xn dispersed over a set of
computing units. An analyst wishes to evaluate f(x1,...,xn) for some
possibly private function f. The analyst has no other access to the units
other than through the dispatching of agents that can perform a pass over
the units. We call the above scenario "mobile data collection & processing"
and we are interested in privacy preserving solutions where the relevant
measures of efficiency are

1. the number of agents the analyst needs to dispatch.
2. the memory required by an agent.
3. the computation required by the units.
In this talk we present a number of different protocols for general
functions as well as for specific functions relevant to private statistical
calculations.