About Us

Welcome!

Thank you for visiting the Nakayama Laboratory website.

Our laboratory is currently active in two main research areas: Computer Vision (CV) and Natural Language Processing (NLP). We conduct research on a wide range of topics, from fundamental theories to practical applications, focusing on technologies that enable computers to recognize, understand, and generate multimodal data consisting of images, videos, text, and more.

In recent years, CV and NLP technologies have made remarkable advances thanks to the development of machine learning techniques, particularly deep learning. However, most of these advances are achieved through brute-force approaches that rely on massive amounts of task-specific supervised data and computational resources. To build more general intelligent systems, it is crucial to determine how data and knowledge should be acquired, and what should be taught to computers—questions for which established methodologies do not yet exist. We believe that the key to answering these questions lies in frameworks that allow systems to learn common-sense knowledge through multimodal interactions, much like how humans acquire knowledge through their five senses in daily life. This is why we focus on both CV and NLP while conducting cross-disciplinary research. Furthermore, with the development of robots and sensor networks, sensor information automatically collected by computer systems open to the real world will continue to overflow. We believe that flexibly recognizing, understanding, and effectively utilizing this diverse information will be key to next-generation intelligent information processing systems.

This field is still developing, with many unsolved challenges. However, this makes it extremely attractive as a research subject, and we feel that many business opportunities lie dormant here. Although fierce development competition continues worldwide in this field, we are eagerly awaiting the participation of students who think they have what it takes. We are also willing to actively engage in joint research and other collaborations, so please feel free to contact us.

Laboratory Policy

The Department of Creative Informatics aims to foster human resources who can produce excellent creative ideas and elevate them into practical products. Our laboratory also emphasizes the following keywords:

"Create" things that don't exist in the world

Most students who graduate from our laboratory and department will be active in fields deeply involved with software. This field sees rapid research and development, with advanced technologies spreading in very short periods. It is not uncommon for content presented at cutting-edge international conferences to be implemented in open source within six months.

Against this background, software engineers are increasingly polarized between those on the "creating" side of new technologies and frameworks and those on the "using" side. To be clear, this is not about which is superior or inferior. Catching up with and mastering emerging new tools is an essential ability for living as a researcher or engineer, and this must be continuously refined. However, if you only learn how to use ready-made technologies, there is little point in spending time at a university. We want students studying in our laboratory to always have the consciousness of being on the "creating" side, finding and questioning values that don't yet exist anywhere.

Creating new things is not easy; it takes a long time and involves hardship. However, no matter how small a step it may be, the joy of accomplishing something that no one else in the world has done yet, or obtaining knowledge that no one else in the world knew, is irreplaceable. We want you to thoroughly enjoy that appeal.

"Practice" what you create

New technologies and knowledge only demonstrate their true value when conveyed to many people. Of course, in university activities, research presentations and paper submissions are the first step, but there are various other forms. For example, publishing source code or services, and in some cases, even starting a business are possible. Challenging programs like the "Mitou" (Exploratory IT Human Resources Project) would also be good. We strongly encourage such activities in our laboratory. However, what's important is that, as mentioned above, you should always be on the "creating" side, with original new technology at the core. If you emphasize practical activities too much and end up with just a collection of existing tools, it loses meaning. We expect the development of innovative products that will surprise the world.

"Show" what you've done

Recently, the world has become increasingly unpredictable, and even large companies are no longer secure. The era of being protected for life as a member of an organization has ended, and from now on, it's necessary to actively polish individual abilities, widely appeal them to the world, and survive. For example, the contents of GitHub, SlideShare, blogs, etc., are already commonly used as criteria for judging individual skills. Related to the above emphasis on practice, we want you to actively open up your work, abilities, and knowledge, and acquire your own way of presenting them. Research activities at university are also a valuable opportunity to learn such skills. We want you to become someone who is respected as an individual, not just as a member of a laboratory or university.

For students who want to join our lab

Seeking students who are serious about research and development and who can enjoy freedom

We believe that the ability expected of a graduate is to take ideas and shape them as an expert in their field, and to disseminate them externally. Self-satisfying results are not worthy of a master's degree, nor is there any need to come to graduate school to do so. In our laboratory, we encourage students to present their work at peer-reviewed international conferences or journals by the time they complete their master's degree. Alternatively, for those who prefer practical development over academic research, publishing open source projects may also be acceptable. We want to support motivated students so that they can gain as many opportunities as possible and grow.

Regarding the setting and approach to research topics, we respect the autonomy of students as much as possible. If you have the motivation, you can freely work on topics that are not being done in the laboratory (although some relevance is necessary). For example, Nakayama Lab used to only do image recognition research, but when students with strong motivation for natural language processing joined and enthusiastically conducted research activities while leading junior students, we now have a large natural language processing group, and the scope of what we can do as a laboratory has expanded dramatically. Such chemical reactions caused by students' independent activities and the broadening of the laboratory's perspective are the greatest joy for faculty. In this history, the laboratory has developed a culture of respecting each other's ideas, learning from and improving each other constructively, and for those who can think and act autonomously and freely, it should be a comfortable place.

Conversely, this may be a somewhat challenging laboratory for those who just chose it because of trendy keywords like artificial intelligence or machine learning, or for those who always need specific instructions to feel secure.

For those who want to do deep learning research

Deep learning has become commonly used in all fields related to artificial intelligence, and various frameworks have become widespread, so it is not as difficult as before to use it as one of the research tools. However, when conducting research on the principles of deep learning itself, caution is necessary, and unless you are prepared to spend two to three times as much time on research as others, it is safer not to get involved. The main reasons are: (1) Deep learning itself is very computationally intensive, and because it is a technology that requires a lot of tedious adjustment work, much time is inevitably spent on non-essential parts of the research; (2) It is an extremely competitive field where technological innovation occurs on a monthly or weekly basis, and meaningful research results cannot be produced at a half-hearted pace. If you understand these points and still want to take on the challenge, you are very welcome.

Desirable skills

As this is a laboratory that mainly deals with software, some programming skills are necessary (C++, Python, Matlab, etc.). In addition, since we develop and implement methods based on mathematical theory as core technologies, mathematical skills (linear algebra, probability and statistics, etc.) are also quite important. We assume that you already have a certain level of basic knowledge in these areas. Of course, you don't necessarily need to have high skills from the beginning of your assignment, but since we don't provide hand-holding care, it is important to enjoy self-study.

For those who want to enter the doctoral program

We expect those joining the laboratory from the doctoral program to already have a full set of experience and skills in planning, conducting, and externally presenting research independently. For this reason, we make it a condition for application to have at least one paper accepted at a mid-level or higher international conference or journal in our research field. Specifically, for example, international conferences listed here. Needless to say, the same standards are expected for those who continue from the master's program to the doctoral program.

About laboratory tours

We accept laboratory tour requests by email at any time, so please feel free to contact us. However, since we would like to have an in-depth discussion about research in individual interviews, please be prepared to discuss one paper published by our laboratory that you found interesting. Also, since we receive many tour requests just before the entrance exam application period every year and have less free time, we recommend contacting us early. Note that we hold a laboratory open house on the day of the graduate school entrance exam briefing in May every year, with introductions to research content, schedules, environment, etc., and opportunities to talk with current students, so if you just want to hear general information about the laboratory, please participate in that.

Information about entrance exam

Please visit the Department of Creative Informatics page of the Graduate School of Information Science and Technology. Note that the Department of Creative Informatics also conducts winter entrance exams for master's programs.