[No. 161] Shunta Saito

I was introduced to the C programming language in elementary school and came to love programming. My interest in computers continued, so I chose the Faculty of Science and Technology with the intention of studying computer science at university. However, right after entering university, I became absorbed in fun and hobbies, joining soccer and classical guitar clubs and participating as a guest violinist in another university's orchestra. As a result, I failed many required courses, and due to poor grades, I was unable to advance to my first choice, the Department of Information and Computer Science, and ended up in the Department of Electronics and Electrical Engineering. I was devastated when I received the news and holed up at my parents' house. But then, by chance, I discovered an old film camera in a closet that had belonged to my grandfather, who passed away before I was born. I decided to use this as an opportunity to start something new and change my mindset, so I joined the Keio Camera Club and began taking photographs. This later became the catalyst for my interest in images, which would become my specialty. Also, when I became a third-year student, I couldn't give up my long-held desire to play jazz and improvise, so I joined the Modern Jazz Society (the so-called "jazz club"). I feel that my interactions around this time with various people, including alumni, who gathered in the basement of the Student Union Building, allowed me to see many different ways of life. This inspired me to want to live a life where I could also pursue what I love. That fall, I had to choose a laboratory. Because I was interested in image processing and thought I could hone my programming skills, I requested to join Professor Yoshimitsu Aoki's lab, and my research life began.

My graduation research involved estimating the 3D positions of finger joints while grasping an object, using multiple cameras. At that time, I found it enjoyable that the harder I worked on programming, the more I was recognized for it, and I became completely immersed in my research. I realized that I felt a sense of pleasure in concentrating deeply on something and began to think that it would be the best life ever if I could continue living like that, which led me to consider advancing to the Doctoral Programs. When I entered the master's program, another research project I had started for my master's thesis took shape early on, so I began to apply my skills in activities outside the university, following my interests. I devoted a lot of time to activities not directly related to my lab, such as participating in a theater unit at another university by providing technical support to create something like media art, performing a piece with a sculptor that converted a live painting into music in real time, and collaborating on the production of a stage work using image sensing and projection technology. However, through these experiences, I was able to learn a great deal technically, including visualization techniques, methods for accelerating programs using GPUs, and ways to prevent program crashes from affecting performance. Furthermore, having the opportunity during this period to interact with people active in the arts and those studying philosophy at the university sparked my interest in how humans perceive the world through their eyes and what it means to perceive the world in the first place. This also became a catalyst for my desire to conduct research in fields like artificial intelligence and machine learning.

Alongside these activities, I also co-founded a company with a friend to develop web services, where I learned about the entrepreneurial process and front-end/back-end development of web services as vice president and engineer. We sold the company, along with its services, to another company before I graduated from my master's program. Those two years in the master's program were a time when I tried my hand at everything that interested me while contemplating what I wanted to establish as my specialty.

Afterward, I entered the Ph.D. program with the desire to work on image recognition using machine learning, but for a while, I struggled without a clear research topic. However, in my second year of the doctoral program, I got the chance to study abroad for a year at UC Berkeley in the United States. During that time, I arrived at a research topic that involved applying deep learning to automatically detect buildings, roads, and other features from satellite images, and I made this the theme of my doctoral dissertation.

After returning to Japan, I withdrew from the university after completing my course requirements and underwent the examination for my doctorate while working as a researcher on a CREST project. When I published the results of this work on GitHub, a researcher from the American company Facebook contacted me, saying my code was useful and asking if I was interested in working there. I decided to work as a contractor in Singapore until my visa application for the US was ready, and I moved to Singapore right after my commencement. Later, while I was preparing to apply to the US headquarters, I received a direct invitation from the vice president of a Japanese company called Preferred Networks (PFN). I decided to change jobs and return to Japan because joining a startup in its early stages seemed exciting and because PFN was the company developing Chainer, the deep learning framework I was most interested in at the time. I am still working there today.

At PFN, right from the start, I have been given a great deal of freedom to work on various things, such as promoting collaborations with global companies like Microsoft, engaging in R&D projects related to image recognition, and activities for the development and popularization of Chainer itself. Recently, I have also been involved in the field of education, which I have long been interested in, by leading a project to create online learning materials called the "Chainer Tutorial." I continue to live a life where I dive into everything that interests me, in an environment that allows me to develop not only my research and development skills but also a wide range of other abilities.

In closing, while I tried many different things during my student years, I never accomplished any of them to the point of being recognized in a tangible way, such as winning an award or a competition. One reason for this is that you can't win an award or a championship without first applying or participating. However, I lacked the confidence to enter competitions. Instead, whenever I gained something from an endeavor, I did my best to output it in a form that others could recognize. For me, this meant things like properly presenting a work created with people from completely different fields, or publishing the code I wrote on GitHub. Although these were all incomplete and immature, they all became stepping stones to the next thing. I believe that the reason I am fortunate enough to continue doing what I love today is thanks to those efforts and to the people who saw them and gave me a chance. Even if you don't have the confidence to compete, I believe that if you gain something, sharing it with others in a visible form can lead to unexpected opportunities. I hope that people who have something they love but lack confidence will find that by sharing what they have gained—even quietly at first—it can become a catalyst for them to continue pursuing it for a long time.