UCLA researchers and co-inventors have developed a method and system for multi-party computation (MPC) that enables multiple parties to jointly compute a function over their inputs while keeping those inputs private. The scheme aims to improve efficiency, reduce interaction or rounds of communication, and support public verification or stronger privacy guarantees.
Multi-party computation is crucial in settings where sensitive data from multiple sources need to be processed jointly (e.g. medical data, financial data, federated learning) without revealing individual inputs. Existing MPC protocols often trade off between round complexity (number of back-and-forth messages), computational/communication overhead, trust assumptions, or require semi-trusted setups. There is a need for MPC methods that are more efficient, leaner in communication, robust, and better suited for practical deployment with strong privacy.
This application introduces a system in which parties can engage in a protocol to compute a function while preserving input privacy. Key features described include:
Reducing communication rounds or interaction steps compared to classical MPC protocols;
Efficient handling of input privacy, possibly via commitment schemes, zero-knowledge proofs, or other cryptographic primitives;
Support for public verification of output correctness without exposing private inputs;
Modularity so that different functions or use-cases can be plugged in;
Emphasis on both efficiency and strong privacy guarantees under well-known assumptions.
Strong input privacy with possibly reduced trust requirements among participating parties.
Lower communication overhead or fewer rounds can reduce latency and improve usability.
Public verifiability allows third-party auditors or external verifiers to check outputs without seeing private inputs.
Modular/flexible architecture may support multiple use-cases without redesign.
Better scalability for real-world applications (e.g. federated learning, joint data analysis) due to improved efficiency.
Federated learning among institutions (e.g. healthcare providers) where data privacy is essential.
Secure joint computation on financial or sensitive datasets without revealing individual inputs.
Auditable computation services (e.g. for verification in supply chain, compliance, or regulatory reporting).
Cryptographic cloud services allowing clients to outsource computation securely.
Privacy-preserving analytics in biotech, insurance, telecom, or governmental data use cases.
WO2024/206968 A2 — Multi-Party Computation Method and System Patent Link
I did not find a specific peer-reviewed publication with DOI that exactly corresponds to this WO application in the searched sources. If you want, I can broaden the search (author names, related MPC cryptography literature) to find prior or related work by the inventors.