secure multi-party computation
In plain wordsCryptographic methods that let several parties compute a result from their combined private data — say, an average salary — without any of them revealing its own data to the others.
A branch of cryptography that lets several parties obtain the result of a calculation over their combined private inputs while none learns anything about the others' inputs beyond what the result itself reveals, even if some parties cheat, up to a stated threshold. General constructions include Yao's garbled circuits and protocols based on secret sharing, in which each value is split into random shares that individually reveal nothing. In machine learning, MPC supports private inference (a client's input and a provider's model both stay hidden), training across organizations that cannot pool data, and secure aggregationsecure aggregation, n.: A cryptographic trick that lets a server add up updates from many devices while learning only the total, never any single device's contribution. in federated learningfederated learning, n.: Training one shared AI model across many phones, hospitals or companies without collecting their data in one place: each trains on its own data and sends back only model updates.. It provides strong guarantees without trusted hardware, but communication and computation overheads — often orders of magnitude over plaintext — make it costly for large neural networks.
Three banks used secure multi-party computation to train a fraud model without exchanging customer records.
Origin
Andrew Chi-Chih Yao posed the 'millionaires' problem' — two people comparing wealth without disclosing it — and secure two-party computation in 1982 (FOCS); his garbled-circuit technique followed in 1986. Oded Goldreich, Silvio Micali and Avi Wigderson (1987) generalized it to many parties.
Related
- secure aggregationsecure aggregation, n.: A cryptographic trick that lets a server add up updates from many devices while learning only the total, never any single device's contribution.
- homomorphic encryptionhomomorphic encryption, n.: A way of locking data so that a computer can still work with it while it stays locked: the computer never sees the real numbers, and only the owner can open the answer.
- trusted execution environmenttrusted execution environment, n.: A sealed-off part of a computer chip that runs programs which the rest of the machine, even its operating system, cannot spy on or interfere with.
- federated learningfederated learning, n.: Training one shared AI model across many phones, hospitals or companies without collecting their data in one place: each trains on its own data and sends back only model updates.
- privacy-preserving machine learningprivacy-preserving machine learning, n.: The family of techniques for training and using AI models while keeping the underlying personal or confidential data hidden — from adding careful noise to computing on encrypted data.
- confidential computingconfidential computing, n.: Running computations inside a hardware-sealed compartment of a chip, so that data and AI models stay protected even from the cloud company or administrator operating the machine.
Sources
- Yao, A. C. (1982). Protocols for Secure Computations. 23rd IEEE Symposium on Foundations of Computer Science (FOCS).
- Goldreich, O., Micali, S. & Wigderson, A. (1987). How to Play Any Mental Game. ACM STOC 1987.
Mentioned in
confidential computingconfidential computing, n.: Running computations inside a hardware-sealed compartment of a chip, so that data and AI models stay protected even from the cloud company or administrator operating the machine., model predictive controlmodel predictive control, n.: A control method that repeatedly plans the best moves a few seconds ahead using a model of the machine, carries out only the first move, then plans again with fresh measurements., privacy-preserving machine learningprivacy-preserving machine learning, n.: The family of techniques for training and using AI models while keeping the underlying personal or confidential data hidden — from adding careful noise to computing on encrypted data., homomorphic encryptionhomomorphic encryption, n.: A way of locking data so that a computer can still work with it while it stays locked: the computer never sees the real numbers, and only the owner can open the answer., secure aggregationsecure aggregation, n.: A cryptographic trick that lets a server add up updates from many devices while learning only the total, never any single device's contribution., trusted execution environmenttrusted execution environment, n.: A sealed-off part of a computer chip that runs programs which the rest of the machine, even its operating system, cannot spy on or interfere with.