NEC - AIST AI Cooperative Research Laboratory

As social systems become ever more complex, advanced needs that can’t be met through individual technologies also increase. To meet these growing needs, the Cooperative Research Laboratory combines simulation technology and AI technology. Regarding technology for understanding special, extreme, and irregular phenomena, advanced decision-making technology that transcends previous data, and technology for coordinating behavior among many AI systems, the Cooperative Research Laboratory pursues a diverse range of projects, from research on fundamental principles with practical goals in mind, to R&D on industrial applications.
Click here to read about the background to the opening of the NEC – AIST AI Cooperative Research Laboratory, and the types of research it conducts.

Overview of Cooperative Research Laboratory

As social systems become ever more complex, advanced needs that can’t be met through individual technologies also increase. To meet these growing needs, the Cooperative Research Laboratory combines simulation technology and AI technology. Regarding technology for understanding special, extreme, and irregular phenomena, advanced decision-making technology that transcends previous data, and technology for coordinating behavior among many AI systems, the Cooperative Research Laboratory pursues a diverse range of projects, from research on fundamental principles with practical goals in mind, to R&D on industrial applications.

Project 1: Convergence of machine learning and simulation

Support for social system design using AI technology

Supplementing big data through social simulation

Enabling data mining with reference to even rare phenomena

Verifying results of new measures

By combining artificial intelligence technologies such as data mining and simulation technology, we aim to establish technologies that make it possible to understand complex, large-scale social systems and support their design.
In addition to big data regarding frequently occurring phenomena in society and their associated structures, we supplement such data with multi-agent social simulations for phenomena that only occur rarely, such as disasters and major events, and for unknown situations, and we analyze mining technologies such as interdisciplinary learning. In this way, we develop methods for understanding complex social phenomena and verifying the effects of new measures using an engineering approach.
Through such methods, we provide frameworks that help to create pleasant, convenient urban environments, organically coordinate transport and distribution networks with all kinds of services, and devise measures for flexible emergency response.
The research results of PJ1 aiming to solve the problems in the design for production and physical distribution are socially implemented as assimee.

Project 2: Convergence of automatic inference and simulation

By combining simulation and automatic inference technologies, we develop technologies to support the operation of complex, dynamic systems such as plants and IT systems. Through planning and inference using simulators, we analyze the causes of problems and devise response policies, to suggest solutions in forms that people can understand and implement.

Project 3: Negotiation AI

As the automatic control and planning optimization of individual systems by AI advances, it will be extremely important to adjust the behavior and interests between multiple systems. Even between system entities that cannot be centralized control nor cooperated control, we are developing AI technology to coordinate them through automatic negotiation.