Soft Computing Integrated System

Group Web Site

[ Professor ] Yoshihiko Horio
Soft Computing Integrated System

Research Activities

Our brain is a highly-structured but very complex network of a vast number of biological neurons. The brain is established on a completely different information processing principle from that of current digital computers, realizing its high cognitive performance through a physicochemical system. As a result, the brain can quickly and efficiently solve real-world problems, which the digital computers are bad at or cannot solve. Inspired by such information processing paradigm of the brain, in particular, focusing on information processing through physical dynamical process, we aim at a novel brain-inspired computer hardware system, which is robust and flexible, and yet quick and efficient.

Fig. 1-1 A chaotic neuron VLSI circuit chip.
Fig. 1-2 A large-scale general purpose chaotic neuro-computer system with 10,000 neurons connected via 100,000,000 synapses.

Soft Computing Integrated System (Prof. Horio)

We are working on a novel high-performance, highly-efficient, flexible, and robust brain-inspired computer hardware system. In particular, we focus on an information processing through physical complex-networked dynamical process, and its implementation as a computational hardware system using an analog VLSI as a core component. Toward to the final goal, we are developing integrated circuit and device technologies suitable for the brain-inspired computer systems, such as VLSI technologies for high-dimensional chaotic networks and large-scale complex systems, VLSI circuits and architectures for ultra-low-power asynchronous neural network systems, and compact and low-power devices/circuits, e.g., spintronics devices for neuron and adaptive synaptic connections. At the same time, we are developing a massively-parallel brain-inspired computational system architecture, which is very much different from that of the conventional digital computers. We further intend to realize an autonomous brain-inspired computer with a sense of self and consciousness based on a complex network with dynamic change in spatiotemporal network state and structure.

Research topics

  • VLSI information processing systems based on complex dynamics
  • Brain-inspired neuromorphic analog VLSI circuits
  • High-performance brain-like information processing system and its applications
  • Brain-inspired VLSI system with consciousness
Fig. 2-1 A chaotic neuro-computer proto-type system for quadratic assignment problems, which uses chaotic wanderings and inspirations to find solutions.
Fig. 2-2 A dedicated VLSI chip for a fast and practical large-scale chaotic neuro-computer system which can solve real-world problems.