Whenever I am asked to write an article or give a lecture like this, I have always titled it "An Encouragement of Fundamental Studies." By explaining how interesting applied research can be developed from the mathematical foundations of my specialty, the field of signal processing, I have consistently advocated for the importance of fundamentals, asserting that a solid foundation allows one to advance in any direction. As an educator, I believe I have been saying something quite reasonable. However, in the last 10 years, society has undergone a major transformation. This is due to the development of artificial intelligence research.
For nearly 30 years, since my student days, I have been conducting basic research in signal processing. Signal processing is a fundamental academic field that involves converting and processing signals such as images, audio, and communications to acquire information or restore signals, and it is a crucial technology that supports our current advanced information society. It involves mathematically describing and modeling phenomena to formulate problems, and then solving these problems using various optimization techniques to obtain solutions. For this, deep knowledge and long years of experience are essential, and when combined with a passion and intuition for research, new research is accomplished. That is why I have always told my students to thoroughly study the fundamentals to cultivate the ability to develop in any direction.
However, with the evolution of artificial intelligence over the past 10 years, the story I used to tell is no longer as valid. Speaking for myself, an opportunity led me to change my research theme at the age of 50, shifting from signal processing to image restoration using deep learning. It is impossible to completely describe real images and audio with mathematical models, and there were many cases where even if an optimal solution was found, the results were not sufficient. However, when solving the same problems with a deep learning-based approach, unbelievably superior results were obtained. Now, on the verge of retirement, I am once again thrilled by the fun of research, and I can't help but be excited about what I can achieve in the future.
On the other hand, research in image processing using deep learning is something that even beginners can fully engage in, as long as they have programming skills and a spirit of inquiry (though basic knowledge is, of course, necessary). Rather than being constrained by preconceptions based on knowledge and experience, amazing results can sometimes be achieved from fresh and novel ideas. Some students in my laboratory have achieved astonishing research results as early as their fourth year of undergraduate studies, and looking around the world, the achievements of young researchers are tremendous. There are many opportunities for young people to succeed, and the Fourth Industrial Revolution is starting right now.
As for myself, while still maintaining the spirit of "I won't lose to the young ones," I want to help my students grow and support them, and witness amazing research results together. "An Encouragement of Fundamental Studies" is of course important, but from now on, let's make it "An Encouragement of *jitsugaku* (science)"!