Theory Seminar, September 13th, 3:30pm
Warren Weaver Hall, Room 201
Kai-Min Chung, Cornell University
Cryptographic protocols have been developed for a variety of tasks,
including verifiable delegation of computation, electronic voting
system, privacy preserving data mining and more. A central method for
constructing such protocols is to first construct a "basic" protocol
satisfying a weak level of security, and then amplify the security of
the protocol by running multiple instances of the basic protocol in
parallel; this is referred to as a "parallel repetition".
In this talk we present several general *parallel repetition
theorems*, identifying general classes of protocols for which such
parallel repetition amplifies the security level, while at the same
time determining the optimal number of parallel repetitions, and the
amount of extra randomness needed, to perform such security
amplification. In addition, we will briefly mention several
applications of parallel repetition in cryptography.