Associative Interacting Intelligence

Understanding and Learning from Human

Although the computer has been developed as a numerical computational tool, it has become a core processor of many of our daily tools brought about by the harmonized contribution from programming and processor technologies. In fact, former super-computer technology has become indispensable in our daily lives. Even though we have access to powerful and fast processors, we are a long way from realizing an intelligent communication link between a human and a machine. It is definitely not for the lack of numerical computational power, but because the human brain and a conventional computer have very different process objectives. Communication by conventional computers is defined as a transfer data function between agents. On the other hand, human-like communication is the result of learned behavior or from shared state interactions between agents. It is our intent to replicate the human-like interaction system by understanding and learning from human. To achieve this goal we focus on the following three research topics;

Brain-like Computer

— The architecture of the brain is completely different from current computer systems. The ultimate brain-like computer will appear most likely with very specialized hardware. We will clearly build and explain its architecture by developing hypothetical brain models in parallel to creating potential hardware. We will then evaluate its interactive intelligence in a real world situation.

Brain-Machine Interface (BMI)

— BMI is a new interface concept utilizing brain activities to control machines. To understand BMI, we will focus on the non-invasive measurement and decoding of natural brain activities that accompanies human states such as recognition, emotion, intention and motion. The anticipated results will provide fundamental novel man-machine communication technologies and novel algorithms for brain-like computing.

Human-Robot Interaction

— Despite recent technological development of intelligent robots in areas such as speech, image, environmental understanding and behavior control, we are still far from building robots that can naturally, robustly, and adaptively interact with humans in real world situations. We are pursuing a novel intelligence model that will integrate individual technologies and learning techniques to achieve such human-robot interactions.