[No. 65] Masanori Kawakita

I am currently at Kyushu University, conducting research in fields such as machine learning (artificial intelligence), bioinformatics, and cryptanalysis. When the 2008 Japanese Joint Statistical Meeting was held at the Keio University Yagami Campus, I visited Professor Honda of the Department of Electronics and Electrical Engineering, to which I had belonged. This led to the opportunity to contribute to this column for Keio University alumni.

My motivation for choosing the Faculty of Science and Technology was very simple: I had wanted to be a scientist since I was a child, and that was it. I was particularly interested in artificial intelligence (understanding the mechanisms of the human brain and whether they can be replicated in a machine), perhaps because my parents gave me a computer, which was rare at the time, when I entered junior high school, and I studied programming. However, back then, I had no idea which universities were actively researching AI. I chose Keio because I had the impression that it fostered more social skills than other universities and produced many successful people in the world.

As it turned out, when I enrolled, many of the people around me were smart, cool, and fun, and they were absorbed in having a good time. Naturally, my grades were not great, and I was not assigned to the Department of Information and Computer Science, which was the most popular and my first choice. Instead, I was placed in the Department of Electronics and Electrical Engineering, my third choice. I was extremely shocked at the time. I thought I would only do research in artificial intelligence, so I even considered intentionally repeating a year to transfer to the Department of Information and Computer Science, but it was in vain. I was even prepared to quit university if I couldn't do the research I wanted. As a result, as you can infer from my brief history below, I did repeat a year. However, when I spoke with Professor Ikuji Honda of the Department of Electronics and Electrical Engineering, he made me an incredible offer: "You can do the research you want to do here. I have connections with the artificial intelligence lab, so I'll introduce you, and you can work with them as well."

My lab life from then until I completed my master's program was incredibly enjoyable. In addition to the Honda Lab, I moved to the Nakanishi-Saito Lab in the Computer Science course for my master's, and everyone was passionately engaged in their research. My research theme as an undergraduate was an associative memory model using neural networks (a machine that can trace memories through association, like a human). For my master's, I researched independent component analysis (a machine for extracting a specific person's voice from a mix of many people talking at once). When I get absorbed in something, I lose all restraint, and there were long periods when I stayed up all night and hardly ever went home. But at the time, there were many friends like me, and spending time consulting with and encouraging each other made even the all-nighters feel enjoyable and not like a hardship.

What was particularly noteworthy about our lab was that everyone was also enthusiastic about drinking parties and having fun. This is very important. No matter how dedicated you are to your research, if you get stuck on the same problem for too long, your ability to come up with new ideas or find a way out seems to decline significantly. Taking appropriate breaks and building physical stamina are actually very important, even in theoretical research. In fact, these things were a huge help in overcoming the various challenges I later faced. Afterward, I decided to spend my doctoral program at The Institute of Statistical Mathematics to learn statistical science and information geometry, which are necessary and powerful tools for artificial intelligence research.

Finally, I would like to talk about something from my university research life that I now realize was very important. To do so, let me first list my research themes from after my master's graduation to the present. These include research on artificial intelligence using statistical science; research to predict the conditions under which specific fish are likely to be caught based on fish catch data (fisheries data analysis); research for diagnosing and treating diseases based on gene function (bioinformatics); cryptanalysis; and research to protect networks from various attacks over the internet (network security). It is not because I have superior abilities that I am able to conduct such a wide variety of research. There are two main reasons.

One is that I studied the profound theory of statistical science. Statistical theory is a practical theory that can be applied to a very wide range of fields. Even when conducting applied research, it is important to thoroughly study the foundational theory that is most likely to be useful. The other reason is that I have studied and researched various fields as my interests led me through my bachelor's, master's, and doctoral degrees, which allows me to adapt flexibly to new fields. I believe this is a major advantage of a university where top-tier researchers from a wide range of fields are gathered in one place.

The Keio University Faculty of Science and Technology was undoubtedly an environment that met these demands. The current trend is to fuse techniques from multiple fields, rather than focusing on a single one, and this is beginning to produce results that were unimaginable within single-field research. This type of research is called interdisciplinary research. To achieve good results in interdisciplinary research, I hope that those of you considering enrolling in the Keio University Faculty of Science and Technology will steadily learn useful fundamental theories, but without being confined by them, also take an interest in diverse fields and view things from a broad perspective. After all, the environment to do so is more than adequate.