Threshold Garbled Circuits and Ad Hoc Secure Computation
1 view
Skip to first unread message
Dionysis Zindros
Apr 20, 2021, 3:17:25 PM4/20/21
to decr...@googlegroups.com
Tomorrow, Wednesday, April 21, 2021, 03:00 PM UTC, Michele Ciampi from the University of Edinburgh will present "Threshold Garbled Circuits and Ad Hoc Secure Computation".
Abstract: Garbled Circuits (GCs) represent fundamental and powerful tools in cryptography, and many variants of GCs have been considered since their introduction. An important property of the garbled circuits is that they can be evaluated securely if and only if exactly 1 key for each input wire is obtained: no less and no more. In this work we study the case when: 1) some of the wire-keys are missing, but we are still interested in computing the output of the garbled circuit and 2) the evaluator of the GC might have both keys for a constant number of wires. We start to study this question in terms of non-interactive multi-party computation (NIMPC) which is strongly connected with GCs. In this notion, there is a fixed number of parties (n) that can get correlated information from a trusted setup. Then these parties can send an encoding of their input to an evaluator, which can compute the output of the function. Similarly to the notion of ad hoc secure computation proposed by Beimel et al. [ITCS 2016], we consider the case when less than n parties participate in the online phase, and in addition we let these parties collude with the evaluator. We refer to this notion as Threshold NIMPC. In addition, we show that when the number of parties participating in the online phase is a fixed threshold l ≤ n then it is possible to securely evaluate any l-input function.
Join here: https://zoom.us/j/97127367194?pwd=M3FsY2FUamFrdVl2UFNiRTh4MEFIUT09
Password: km53TR
More seminars: https://decrypto.org/seminar
Abstract: Garbled Circuits (GCs) represent fundamental and powerful tools in cryptography, and many variants of GCs have been considered since their introduction. An important property of the garbled circuits is that they can be evaluated securely if and only if exactly 1 key for each input wire is obtained: no less and no more. In this work we study the case when: 1) some of the wire-keys are missing, but we are still interested in computing the output of the garbled circuit and 2) the evaluator of the GC might have both keys for a constant number of wires. We start to study this question in terms of non-interactive multi-party computation (NIMPC) which is strongly connected with GCs. In this notion, there is a fixed number of parties (n) that can get correlated information from a trusted setup. Then these parties can send an encoding of their input to an evaluator, which can compute the output of the function. Similarly to the notion of ad hoc secure computation proposed by Beimel et al. [ITCS 2016], we consider the case when less than n parties participate in the online phase, and in addition we let these parties collude with the evaluator. We refer to this notion as Threshold NIMPC. In addition, we show that when the number of parties participating in the online phase is a fixed threshold l ≤ n then it is possible to securely evaluate any l-input function.
Join here: https://zoom.us/j/97127367194?pwd=M3FsY2FUamFrdVl2UFNiRTh4MEFIUT09
Password: km53TR
More seminars: https://decrypto.org/seminar
Reply all
Reply to author
Forward
0 new messages