• Basic understanding of high dimensional data and develop algorisms to support the effective use of big data.
• Especially, focusing on suppressing the appearance of ‘hub data’ which is one of the high dimensional data problem, behave like spam that resembles any data.
• By understanding these high dimensional specific problem, we successfully utilize real world data, such as document, image, recommendation, and biological data.

In the left figure, edges are concentrated to one specific data called “hub data”. On the other hand by applying our hub suppressing method we could reconstruct the original cluster structure.

• Research and development of algorisms for flexible robots to perform human intended operations
• As an example, the artificial muscle made of rubber is activated by applying air pressure. The advantage of this muscle is flexible movement, however it is difficult to obtain movements because the muscle often acquire hysteresis.
• By learning the relation between the air pressure and the movement of the muscle by applying machine learning techniques, we could obtain the movement that we desired for the robots.