Global Site
Breadcrumb navigation
Message for students 2025:Yoshiyuki Yajima
April 3, 2025
Enjoying an active role in the world even when starting out from a different background

Yoshiyuki Yajima
Senior Researcher
Advanced Network Research Laboratories
After receiving the doctoral degree in astrophysics in 2022, Yajima joined NEC. After engaging in research for structural health monitoring based on sensor fusion, he has been working on research and development of short-term traffic flow prediction based on highway traffic flow data obtained from optical fiber sensing. Since joining NEC, he has been continuing to play an outstanding role in academic conferences, with research papers being accepted twice at top-class international conferences in three years.

Applying skills honed in research of astrophysics to the area of sensing
I was aspiring to be a researcher of astrophysics in academia. However, this feeling changed after going through two great earthquakes in my life. Seeing how drastically earthquakes can shake the society, I changed my mind that I want to bring values to the society from my research. So I began my search for a researcher position in the private sector by participating in university-organized matching events that connect doctoral candidates and companies.
The reason for choosing NEC is because I had the great impression that NEC was putting a lot of effort into research activities in academia. Many NEC research papers have been accepted at top-class international conferences related to AI and machine learning. NEC also holds one of the largest numbers of patents in the world. So, I applied for a position at NEC with an interest in NEC’s competitive research achievements and technological prowess that have a strong presence in the world.
While I started out from a totally different field, i.e. astrophysics, looking back, I realized that the essence of my research in the doctoral course―observing interstellar medium in galaxies using radio waves and revealing the mechanism of star formation―was based on phenomena understanding. The basic flow of research starts from analysis using observation data, to understand why and how phenomena is occurred. Thinking that this process has something in common with sensing, one of NEC’s areas of research, I applied for a research position in this area and joined the company.
I almost stumbled on some differences in the first year when I joined NEC. At that time, I started the research to identify damaged parts on a bridge and quantifying their damage level as a next step of anomaly detection. Then, I had no knowledge of civil engineering and structural mechanics. I wasn't even aware of bridge structure then. I caught up by reading papers, discussing with a professor at universities conducting collaborative research, analyzing the actual measurement data, and carrying out structural mechanics simulations.
Once I managed to catch up and learned the specific expertise of the field, I didn't feel that there was a huge difference from what I used to do in the research of astrophysics. There were many similarities in the process to understand the phenomena from observation data, which helped me quickly get on track. In fact, my research paper about this topic in the first year at NEC was accepted at a top-class international conference in structural health monitoring.

Data assimilation based on physics modeling to become a new trend next with machine learning
From the second year, I has been tackling to model and simulate observed traffic flow using optical fiber laid out on expressways based on NEC’s proprietary optical fiber sensing technology. My research involves prediction of traffic conditions up to about 30 minutes ahead in real-time. What I used here is a modeling technique that applies simulation based on a theoretical physics model rather than machine learning. Namely, data assimilation technology is applied to integrate observation and simulation data. There are two benefits to data assimilation. The first one is that it works even when data is limited. Since optical fiber sensing is still a new technology for traffic monitoring, massive data which is necessary for machine learning cannot be prepared in practical cases. The second one is that explainability is ensured. Unlike machine learning, which tends to be a black box, the data assimilation approach can clearly explain the reasons behind the analysis and prediction with the theoretical model. This is a very important point for expressway service providers that are required to ensure safety of the users. The paper about these research results was accepted at the Transportation Research Board Annual Meeting, the top-class international conference in the field of traffic engineering.
While NEC has many achievements in machine learning, data assimilation is still a relatively new approach with few cases. Hereafter, I hope to work on research that adopts this technology to sensing applications in other areas beyond expressways. I want to bring out more successful projects so as to grow data assimilation into another trend.
Ultimately, I would like to explore combining this technology with machine learning, generative AI, and LLMs. In my future challenge, I hope to build a complete digital twin system that handles the entire cycle from sensing to modeling, simulation, prediction, optimization, and feedback, operated by data assimilation and LLMs.

Flexible research environment supported by NEC as a brand
Since I originally came from the field of astronomy, what I want to tell students who also have a basic science background such as physics, is that there is no need to hesitate to pursue a career as a corporate researcher. For both researchers in corporations and academia, the process to understand phenomena and to solve issues, analyzing observed data, referring basic principles and laws with your own knowledge is the same. You should be able to apply your abilities that you have developed. In fact, there are many active researchers in NEC who was specialized in astrophysics or other fields of basic science. If you are interested, I would like you to take on the challenge.
Regarding a characteristic feature of research at NEC, it would be the accessibility to fields that you cannot enter usually, while you can do during demonstration experiments and other opportunities. For example, in the collaborative research with an expressway service provider, I entered a service personnel-only passageway for a huge bridge at a port to install sensors while cars were passing through underneath. It was very thrilling but exciting experience.
This may be possible thanks to the branding of NEC. Fortunately, many people in Japan find NEC credible and the name is also known abroad at international conferences, which facilitates worldwide promotion of your technologies and developments. As goals, the laboratories also
promote submission and publication of research papers for international conferences and journals in addition to future business contributions.
NEC also has multiple overseas laboratories that offer opportunities for discussion with their members. In fact, I was on a two-week business trip to NEC Laboratories America, also to participate in an international conference, just last week before this interview. (Related link) NEC Laboratories have an environment open to the world, which can be a great advantage for your research.
The Laboratories also have this rule called the “20% time” (Note). It means that as long as it’s within 20% of your work time, you can work on research that you like. This attractive offering enables you to flexibly expand and deepen your research driven by your own interest.
- Note:The percentage varies by division or team.
- *The information posted on this website is the information at the time of publication.
A day at work


Message to my past self in my school days

Private column
I like watching stars. My hobby is taking photos of starry landscapes. During vacation, I drive to the suburbs or sometimes even make excursions to Hokkaido. When I was a student, my photo of a starry sky was selected for a university co-op calendar. Looking at a night sky with full of stars makes me realize the beauty of nature.
