I am doing research on artificial intelligence (AI). I am looking at two areas. One is AI that can recognize gymnastics skills and automatically rate performances. The other can recognize human body movements from pictures of elderly people or preschoolers taken at nursing homes or daycare facilities and convert them into text data to create nursing care or daycare journals.
We began this research by properly and accurately measuring people's movements. This requires collecting data such as how much each limb and every joint in the whole body bends. We also used machine learning and other solutions to turn the collected data into text.

As the amount of accumulated data increases, AI can recognize movements of a person that we show it like, "This movement is similar to one that I learned in the past." Additionally, we can use the accumulated data to drive a humanoid robot that mimics human movements.
A robot's movements can be reproduced exactly, and robots can move in the same way repeatedly. On the other hand, human movements, even if it is just a walking motion, differ slightly each time and have variations. We use statistical mathematics and data science to bridge these differences and variations into a form that can be handled by AI.

Data science, which can efficiently process huge amounts of data, is based on relatively traditional disciplines such as statistics. Hence, there are many tools available for data science, including various theorems and algorithms. I strive to combine these tools to recognize human movements and convert them into text. Additionally, I would like to use these tools to control robots. This is my research field.

One of the courses that I am teaching is called "Mathematical Sciences for the Integration of Arts and Sciences," which is designed to attract students interested in the basics of data science and AI from a wide range of majors across arts and sciences. The goal is to increase scientific literacy about mathematical sciences by having as many students as possible take this course.
In 2022, eight classes will be offered per week. Students can choose the class that fits into their schedule (i.e., the same class is offered in the 5th class slot on Mondays as well as the 4th and 5th class slots on Tuesdays), without overlapping with their required classes.
I have also created e-learning content so that students can view the class content on demand. Each module includes quizzes so that students can check their understanding and increase their learning effectiveness. Students can watch when it is convenient as the e-learning content is posted on a video server. This provides flexibility not only for students but also for me because I don't have to go to classrooms eight times a week to teach. There are about 3300 students per year at Osaka University, and I hope that about 1400 students will take this course.

In "Mathematical Sciences for the Integration of Arts and Sciences," students practice simple analysis using national census data. For example, analyzing the relationship between the amount of money spent on cheese per household by prefecture and wine consumption per household by prefecture reveals that prefectures spending a lot of money on cheese tend to also spend a lot of money on wine. In this way, I help students experience data analysis and visualization of the results in a practical manner using statistics.

As a student, I took a statistics course when I was studying at university. However, its usefulness was often too abstract for me to understand. It was not interesting at all. In my undergraduate and master's courses, I majored in mechanical precision engineering. There were also classes on control and programming in graduate school, but I avoided them because it was unclear what was interesting about them.
After completing my master's program and getting a job at an automobile company, I worked on driver assistance systems where I actually drove a car with such a system. This was the first time I realized that AI and control technology were interesting. So, I went back to graduate school and joined a robotics laboratory, where I earned my degree.
There are many universities where faculty members specializing in mathematics teach data science, but I think one of the roles of faculty members like me, who have a working business career, is to stimulate students' curiosity by giving them something concrete.
Some students from departments other than data science have taken more than one of my classes because they were surprised by my lecture, which opened a new and interesting world to them. Often my students ask me questions about what classes they should take or what books to read. It makes me happy to connect and share knowledge with many students.

Yes. In 2021, I gave a series of classes to second-year high school students. It was part of the high school-university connection effort. I taught five 3-hour classes on Saturdays, which were equivalent to two credits, covering a high volume of content. These classes were taught at the same level as those for first-year university students, but I modified the mathematics content to be suitable for second-year high school students so that they could see how high school math can be used in data science.
Many universities desire to teach data science, but the limited number of faculty members who can teach data science is an obstacle. Therefore, we are working to make the e-learning content produced by MMDS available to local universities.
Within Osaka University, classes for first-year students have been enriched, and practical training and exercises are now available. We are looking into the possibility of involving more senior undergraduate and graduate students in the form of class or research support. We are currently thinking of accepting PhD students from other majors such as biology at MMDS for a few months to help guide their research using data science.

Interview as of January 2022
*Interview and photography were conducted with countermeasures against COVID-19.