Country / Region
Graduate School
Graduate School of Science and Engineering
Systems Engineering and Science
Year of Graduation
2020 (Master’s degree)
Prof. Kazuhiro Yasuda
Research Theme
(Master thesis) Error analysis of stochastic optimal control problem with deep fictitious play (Current topic) Finite state Mean-field game
Current Workplace / School (Occupation)
Osaka University (Doctoral Student)

Q1 What kind of research did you do at IIST?

Basically, my research involves error analysis of stochastic optimal control problems. Such problems can be generalized by solving Forward-Backward Stochastic Differential Equation. My work is to approximate FBSDEs numerically in a time-discrete way based on a learning process strategy called deep fictitious play. This is a special mean-field game strategy, and our model is based on the Linear-Quadratic case. Under the appropriate assumption of strong and weak convergence rate of FBSDEs, I estimate error of cost function (of stochastic optimal control problem), meanwhile approximate boundedness of time interval & iteration stage so as to provide available parameters.

Q2 How was your campus life?

Two unfulfilled wishes: one is that I have not experienced the Tama campus life (I heard it is very large); the other is that I haven’t applied for co-op credit card within Hosei’s logo, which is cool to use, I reckon.
I’m in Koganei campus. It has a good multi-lingual function, even if some staff are not good at English, they’ll find professional English supporters, no need to worry at all. (Real personal experience)
Highly recommended to enjoy Ippeisoba, a special noodle restaurant near Koganei campus. VERY DELICIOUS!!! At Higashi-Koganei station, a Chinese restaurant called Kōen, is worth a try!!

Q3 What are the good points of IIST?

I have compared the English program courses of many top universities. For the IIST course field, not all universities have a wide range of English courses such that in many cases you have to learn on your own. And even some national universities cannot offer such a wide range of English courses as IIST.
Second, this is very international. The courses at IIST are not isolated, if there are only Japanese guys, the professor may use the native language. But once an IIST student has joined, the course will be given in English, which is very impressive!!!
Third, almost all the professors at IIST graduated from top universities in Japan and even taught courses at top’s as well. Therefore, whether the academic style or the tolerance in dealing with people or the multilingual level, you will never be disappointed. Meanwhile, the staff in the IIST department are kind enough, and any problem you can discuss with them, that makes it easy to live and communicate.

Q4 What kind of job or research are you doing now? How are you making use of your learning and experience that you earned at IIST?

My small research is about “Finite state Mean-field game” related topics, but recently I have been working slowly. One difference between the past is the fact that the dynamic, now, is generated by Continuous-Time Markov Chain within semi-martingale representation and the state is finite.
In fact, in most cases of a probability/mathematical finance major, one must learn stochastic calculus & some measure theory-based probability theories at least. In IIST courses, especially in Yasuda sensei’s lab, such courses are provided, and sensei will change the lecture depending on the students’ background year by year. This is of great importance for further study.

Q5 Give a message to prospective IIST students.

Please see all the answers about question 4 and the extra following point:
It is superb but challenging for a foreigner whose English is a second language to enjoy an English-only environment in IIST.
The IIST program is relatively new. Everything is energetic. Work hard, play hard. You can write your own story from here.

The above information is at the time of the interview.