Can graduate students get paid as company employees!? What is Fujitsu's "Excellent Working Doctoral Student Program"?

Hello! This is the Fujitsu PR note editorial department.

Currently, the number of doctoral degree holders per million people is a problem known as "drifting away from doctoral programs," where Japan is the only major country seeing a downward trend(Note 1). Many cite economic reasons for this; compared to Western graduate schools where students can earn a degree while receiving a corresponding salary, graduate students in Japan are generally unpaid, even though support systems have been improving in recent years, which is a cause of the "drift away from doctoral programs," especially in fields where private companies have strong hiring appetites. This leads to a decrease in talented and energetic young graduate students at Japanese universities, and there is concern that research capabilities will be eroded and Japan's status in the field of science and technology will further decline.

Therefore, in 2021, Fujitsu established the "Excellent Working Doctoral Student Program," which allows outstanding graduate students enrolled in doctoral programs to be hired as employees while retaining their university enrollment, enabling them to aim for a degree at graduate school while conducting joint research with Fujitsu researchers. This program primarily recruits students from university master's programs. Those selected can become Fujitsu employees upon entering the doctoral program, allowing them to receive a salary from Fujitsu while continuing their doctoral research at the university.

Seishi Okamoto, Director of Fujitsu Research, states: "The decline in students pursuing doctoral programs is feared to lead to serious problems such as a decline in Japan's international competitiveness and presence. By removing the economic burden and future employment anxiety that are factors in the 'drift away from doctoral programs,' and by providing an attractive system and environment that allows for dual-track research in both academia and industry, we hope to cultivate innovative talent."
As the first in a series of articles introducing this "Excellent Working Doctoral Student Program," we present an interview with an employee who is enrolled at the University of Tokyo while working as a researcher at Fujitsu Research, aiming to obtain a doctoral degree.

Interviewee:
Fujitsu Limited, Fujitsu Research, Artificial Intelligence Laboratory, AI Innovation CPJ
Yuma Ichikawa

First, please tell us about your current research.

I am conducting research to theoretically elucidate the properties of deep learning technologies, which are often treated as black boxes, particularly generative models (Note 2), using statistical mechanics. Based on this theoretical understanding, I am also working on the development of new algorithms and networks. Specifically, I am taking the perspective of physics, which seeks to understand complex phenomena from simplified situational settings, to tackle the black box problem of deep learning and aim for the development of higher-performance and more efficient machine learning models and optimization algorithms.

That sounds difficult... Could you explain it in simpler terms?

Well... you could say it's like using a discipline that has long been developed to study the movement of molecules in the air (statistical mechanics) to analyze generative AI, which is also composed of a vast number of small components, and create new technology. Since the scope of application for statistical mechanics is extremely broad, extending not only to physical phenomena but also to "collections of many elements," it can be said that my research is about understanding things from that perspective.

What was your motivation for applying to Fujitsu's Excellent Working Doctoral Student Program?

While also considering the economic aspect, I thought that by forming a dual track—research at a university that deeply engages in theoretical research, and research at a company that applies university achievements to business while incorporating cutting-edge technology—I could broaden my perspective, so I applied.

Could you tell us if there are any examples of your research themes expanding as a result of being accepted into Fujitsu's Excellent Working Doctoral Student Program?

Since joining Fujitsu, I have had the opportunity to participate in fields that tackle practical problem-solving, not just mathematical research. At the university, I was always focused on mathematical research, and I had never experienced the moment when an algorithm I developed could contribute to solving actual problems, so I was able to experience that for the first time after joining the company. I think it was a wonderful experience to be able to taste this sense of accomplishment at an early stage. Also, Fujitsu has researchers from various fields, and by discussing with them, I feel that my perspective is broadening. Above all, I am spending fun days being able to discuss with various researchers. In the future, I aim to collaborate with people from other laboratories, not just the Artificial Intelligence Laboratory, and engage in revolutionary research while utilizing my expertise. Furthermore, I want to actively host seminars and study groups for Fujitsu Research researchers and contribute to the skill improvement of the entire laboratory.

Is there anything you would like to realize in the future by combining your research content with other Fujitsu technologies and services?

I believe that more fundamental and theoretical understanding is indispensable for causing further innovation in the field of artificial intelligence. To cause further innovation from the development of generative AI in recent years, including other Fujitsu services, engineering research is important, but I believe that research that deeply considers the roots and properties of technology from simplified situational settings to make the next move will definitely become important. In fact, many of the new AIs that have caused innovation in recent years are the result of being developed based on deep mathematical insights. If possible, I would like to do my best to be involved in the development of new generative AI from Fujitsu.

Finally, please tell us about your future dreams.

It may be a matter of course as a researcher, but it is to be a researcher who always faces the questions that well up from within myself sincerely, without being swayed by my environment or situation. This stance is deeply influenced by the words of my mentor at the university I currently belong to. If I were to talk about a slightly bigger dream, I would like to develop a revolutionary generative model that is recognized globally from Japan. Although the word "revolutionary" is abstract, I want to develop influential technology that is recommended, saying, "Let's use the generative model developed by Ichikawa to solve this problem."
Also, visiting various countries, communicating with researchers from diverse cultural backgrounds, and enjoying research is one of my dreams.

[Asked the boss] Please tell us if there are any episodes that speak to Mr. Ichikawa's excellence.

(Comment from Yuhei Umeda, Senior Research Director, Artificial Intelligence Laboratory)
From the moment he joined the team, I had him participate in a joint research project with an overseas university that we were conducting at the time. The professor we collaborated with was a top-class young researcher who contributed significantly to the development of deep learning, but Mr. Ichikawa immediately caught up with the content of the joint research and began to lead the research. The professor also highly praised Mr. Ichikawa's contribution. Since then, he has become the center of team discussions, developed key technology for "Composite AI" (Note 3), which is a focus technology of the AI Laboratory, and has already become a core member of the AI Lab's technology development.

<Notes>
Note 1 Quoted from "Doctoral Talent Activation Plan - Let's Get a Doctorate -".
Note 2 Generative model: A model that generates new data based on data learned by AI
Note 3 Composite AI: A framework that automatically generates AI innovation components that realize more complex problem solving, such as optimizing production scheduling according to demand
・Developed AI technology that automatically generates AI innovation components (Press Release)

Related Links

Japan's first, what is Fujitsu's "Excellent Working Doctoral Student Program"? ~Supporting both academic research and social problem solving (Fujitra News)

Recommended Articles