Towards the creation of innovative computing technologies

The human brain, despite being composed of the seemingly unstable biological elements known as neurons, demonstrates the ability to perform advanced information processing autonomously and adaptively with remarkable energy efficiency. Understanding the information processing architecture of the brain is a critical interdisciplinary challenge spanning biology, engineering, and information science.

This Research Area aims to pioneer a new academic discipline to unravel the brain's information processing mechanisms through in vivo and in vitro experiments, formulate the findings in mathematical models, and explore their potential applications in systems applications. The achievements herein would deepen our understanding of the nervous system and also lead to a more efficient, robust, and adaptable information and communication technology that supports next-generation super-smart societies.

Outline of the Research

The human brain is composed of unstable biological elements called neurons. Even so, it realizes complex information processing with high energy efficiency and adaptability. Such properties does not appear in single cells and cannot be explained as their linear summation. It is only through the multicellular network formed by the precise wiring of a diverse array of neurons that the system-level function of a brain emerge for the first time.

In pursuit of a super-smart society, referred as Society 5.0, the brain is being modeled to develop better machine learning technologies. However, current technologies fall short in fully mimicking the brain. A deeper understanding of the nervous system could lead to what might be called "biological supremacy," akin to quantum supremacy in quantum computing. This advancement could drive the development of next-next generation information and communication technology (ICT).

Research strategy

We set the following three domains to integrate different research fields towards understanding the fundamental basis of multicellular computing:

• Multicellular modeling: Formulating bioplausible models of sensory-motor control
• Multicellular hardware: Developing ultra-low power neuromorphic systems
• Multicellular wetware: Building artificial systems using living neurons

Expected Research Achievements

Expected achievements include a formulation of the interplay between collective behavior of biological elements (cells) and their function as a system (neural system) and understanding its engineering advantages. Our framework seamlessly connects top-down research, i.e., detailed analysis of the entire system, with bottom-up research, i.e., understanding the system by building the system from its elements. The hybrid approaches will lead to the development of a new computing technology with high energy efficiency, robustness, and flexibility.