Secure multiparty computation (MPC) is a collection of cryptographic techniques allowing computation on encrypted data without leaking information to anyone involved in the computation except the intended result. MPC facilitates a large spectrum of applications involving confidential data, for instance collaborative machine learning across data from competing companies.
A future where MPC is efficient enough to be in widespread industrial use would disruptively transform how we collect, store and process sensitive data. MPC was discovered in the 1980, but for many years remained mostly of theoretical interest as the computational overhead of MPC was too high. After decades of theoretical research on MPC, the technology has finally begun to creep into practice. However, current industrial applications of MPC are made possible only by applying it to carefully selected problems. The overhead of MPC is still too large for MPC to be a generic technique that can be put into compilers.
The first objective of BETHE is to create a number of ground breaking new MPC techniques, which will take MPC research from its current state-of-the-art towards a new state where MPC can be applied to the massively complex computational problems that are met in real life computational and economic settings.
So far all industrial applications of MPC has had significant involvement of academics with long-term research careers in MPC, as it is still necessary to have at least a PhD in MPC to securely implement MPC in practice and securely combine MPC with other cybersecurity technologies and distributed systems technology. This is in particular so because much of the MPC theory of the past three decades was performed in idealised theoretical models, which makes academic investigations easy but makes it hard to apply the results to practice.
The second objective is to develop a theory of MPC which better aligns with the emerging practice and with the techniques and methodologies of distributed systems research.
