On October 5, Acompany announced that it has independently developed a secure computation engine, QuickMPC, that leverages the secure computation method known as multi-party computation (MPC).
When using various types of big data in machine learning or statistical analysis, a problem arises by which computation must be performed on medical information, location data, customer data, etc. while protecting personal information. MPC is a secure distributed method that processes data while protecting it through the use of a network. Few practical MPC engines currently exist, however. These have primarily been developed for research purposes, and implemention of MPC engines in actual use cases is said to involve issues of flexibility and security.
Acompany has developed QuickMPC as a general-purpose MPC engine that can resolve these problems. Whereas conventional, research-purpose MPC engines have not been able to achieve system portability and integrity, the use of container technology allows the engine to be flexibly configured in line with users' requirements. Acompany has also developed an original SDK that enables external communication, which the company says enables fast securing, analysis, and decryption of data.
The company plans to proceed with demonstration experiments and joint development with partner companies using QuickMPC.