Introduction to Secure Multi-Party Computation
summary
Overview of practical secure multi‑party computation: three families exist — homomorphic encryption (Gentry’s 2009 fully homomorphic breakthrough initially costs ~10^10 slowdown, now closer to ~10^2–10^3), garbled circuits, and secret sharing. Shamir secret sharing gives information‑theoretic secrecy with k-of-n recovery; additive sharing makes additions local and free but multiplying two secrets incurs communication (roughly two network round trips) unless using precomputed SPDZ triples (a,b,c). Production examples: Sharemind runs 3‑node MPC with a C dialect (GCC fork); PSI products use commutative encryption for private set intersection and outcome‑based drug pricing.
AI-written summary · about
chapters (14)
- 0:00Introduction
- 0:30About Ahto
- 2:58What is Secure MultiParty Computation
- 5:18History of Secure MultiParty Computation
- 8:37How can we do it
- 10:37Homomorphic Encryption
- 16:49Fully Homomorphic Encryption
- 19:42Garbage Circuits
- 27:53Secret Sharing
- 30:35Security Properties
- 35:48Information Theory Security
- 47:00Additive Secret Sharing
- 52:31Additive Equal Sharing
- 54:29Resharing
related talks
22mNext-Gen PeT: the Future of Multi-Party Computation Software
45mPrice of Active Security in Multiparty Computation
51mDistributed Zero-Knowledge Proofs and Applications to Secure Computation
44mIntroduction to confidential computing – protecting sensitive workloads
33mInteraction Preserving Compilers for Secure Computation
34mPrivacy Preserving Aggregation of Ad Conversions using MPC
38mDemystifying Privacy Preserving Computing
18mSECRECY: Secure collaborative analytics in untrusted clouds