Secure multi-party computation (MPC) allows multiple parties to jointly evaluate a function over their private inputs while keeping those inputs secret. Garbled-circuit protocols are a common way to achieve this, but their cost has traditionally been tied to the number of non-linear gates in the circuit being evaluated.

According to the paper's abstract, previous optimization efforts in garbled-circuit protocols have largely concentrated on reducing that non-linear gate count. The new work takes a different angle: it suggests maximizing the amount of local processing each party can do before the joint computation begins.

Because the source is only an abstract, the specific algorithm and measured speedups are not described in detail. The paper is titled "Think Locally Before Computing Jointly: Efficient Secure Multi-Party Computation via Maximizing Local Processing" and is available on arXiv under identifier 2610.17072.