I was born in Kumamoto-Prefecture in 1945. My childhood home was located in the countryside, surrounded by beautiful nature. Every day, I played with my friends in the fields and hills, without worrying about my future. I have no memory about my schooling at the time. I realized later that my childhood experiences playing in Kumamoto-Prefecture were actually quite similar to the research I would later engage in, when I worked for a national laboratory and a university.
Education
1964-1968 Nagoya Institute of Technology, Faculty of Engineering (Bachelor of Science)
1968-1970 The University of Tokyo Graduate School (Master's Degree)
1985 Dissertation Doctorate (Doctor of Engineering)
Experience
1970-1992: As a researcher, I worked for the Electro-technical Laboratory, which belongs to the Ministry of International Trade and Industry of Japan (MITI). The main research topics I worked on were character, image, and speech recognition. Regarding the topic of character recognition, I proposed a method of feature extraction called cellular feature which was licensed to a company that used it in their commercial Optical Character Reader. Concerning the topic of speech recognition, I proposed a matching algorithm called continuous dynamic programming (CDP), which fits somewhere between a reference sequence pattern with a fixed interval and an input sequence pattern without a fixed interval. Essentially, it is an endless stream of sequence patterns. This was the first algorithm in a family of CDP algorithms which I would go on to create.
1985-1986: I stayed for one year at the National Research Council of Canada in Ottawa as a visiting research fellow. I enjoyed my time there, a time in which I conducted research concerning image processing, but I did not obtain any remarkable outcomes during the stay.
1993-2002: I joined a big ten-year national project called Real World Computing (RWC) sponsored by MITI. RWC had two main topics. The first one was developing new functions applicable to the real world, including multi-media retrieval, robotics, and neural networks, while the other was the construction of a massive parallel computer. I worked on the former along with many excellent researchers from different industries. One of the outcomes of the project was the RWC music database, which was a collaboration with a team of AIST researchers lead by Dr. Masataka Goto. This massive music database is now widely used by researchers in the field of music.
2003-present: I am currently working at the University of Aizu as professor, dean, and president.
Areas of Expertise
My research areas include character recognition, speech recognition and retrieval, image retrieval, image processing, image understanding, computer vision (3D scene reconstruction from video footage), data mining, data visualization, mobile robots, wired drone networks ("dronet"), etc. I proposed a family of methods called Continuous Dynamic Programming (CDP) for segmentation-free optimal matching between two patterns. Two-dimensional CDP is applicable to images for segmentation-free recognition of image and pixel-wise tracking, as well as to segmentation-free recognition of a single speech spoken by multiple speakers. Time-space CDP is applicable to move images for time-segmentation-free and also space-segmentation-free recognition of the motion of moving target objects belonging to a moving background scene, as well as moving non-target objects in the background. I am also proposing a method for 3D reconstruction of scenes including wide city scenes from video streams (High Definition, 4K size) without using stereo vision. I am also proposing a new type of wired drone network called “Dronet” in which a number of drones are connected to each other by cables. This stabilizes the drone network against external disturbances like the wind. It also allows the drones to distribute payloads amongst themselves, allowing heavier objects to be carried.
What have been the main lessons you have learned in your career and how do you implement them in your management?
Research is sometimes thought of as being random, but I believe that there are in fact several methodologies one can use. As with any creative activity, the first methodology begins by finding the incredibly small features and signs that are hidden within natural phenomena. When doing this, you must lightly touch these things without preconceived idea , otherwise you might miss them entirely. The second step is continuous trial and error and detailed investigation. The final step is a three-stage process in which one reveals the phenomenon, substance, and essence of the research subject. The three-stage concept was proposed by the physicist Mitsuo Taketani as a means of understanding nature, but it can be applied to any discipline, including engineering. The second methodology is to make use of mathematical tools. Mathematical tools are used to create models of the subject, and to dynamically describe it. The third methodology is to undergo a final evaluation through experiments using the methods proposed by the researcher. It is more than reasonable to conduct comparisons of existing methods with those you have proposed during your research. Conducting too much survey work before starting a research project can actually do more harm than good because of the strong influence of the past work. I believe that conducting surveys of past methods after your work is perfectly acceptable. The merit of conducting surveys after your research is complete is that even if your research partially overlaps with past research, you have an opportunity to see how your original work has grown.
Mentors
I have many mentors who have influenced me in both direct and indirect ways. One example was to watch how mathematical tools were used for both modeling and analyzing the subjects of research. Most excellent researchers have mastered their own unique techniques to handle their research themes. Reading good books or articles is another way to enrich our own productive thinking. Dr Goro Shimura, a famous Japanese mathematician is one of my mentors even though his research area is completely different from mine and I have never met him. However, I was very influenced by reading Shimura’s books and essays. Apart from academic research works, I am very impressed by the works of the authors Ryotaro Shiba and Nanao Shiono. They write historical fiction and non-fiction history books about Japan and the ancient Greco-Roman world.
Quotes
“Any creative work requires the pure and honest heart of a young boy." Ryotaro Shiba, Japanese novelist.
