Content of research

In the field of information science and engineering, many problems appear, such as the problem of designing a mathematical system that provides a desired result, the problem of identifying a mathematical model that gives a given input and output, and the inverse problem of estimating unknown inputs from a system and observations. When dealing with these problems, by dividing the analysis into conditions specific to each problem and mathematical models independent of each problem, it becomes possible to theoretically determine the performance and limitations, and to also expand the analysis horizontally to problems that can be described by similar mathematical models. With this unique approach, we have developed various methodological constructs and application techniques in machine learning problems including image and color restoration, separation of individual sounds in acoustic signals, pattern recognition, and sampling theory. By applying our methodology to today's rapidly developing and diverse measurement technologies, we expect to develop a variety of application techniques based on theory.

Potential for social implementation

• ・Color restoration of faded film
• ・Deleting unwanted objects from an image
• ・Three-dimensional sound field control
• ・Mirror surface design for desired brightness
• ・High-precision data acquisition with a small number of sample points

Appealing points to industry and local governments

In needs-oriented projects, there is often no room to devote human resources or time to examine the theoretical validity and appropriateness of the methods to be developed, and as a result, there is a tendency to focus on ad hoc solutions. We believe that we can cooperate mainly in terms of mathematical analysis to develop reliable technology based on theory.