January 31, 2022

Synthetic Aperture Radar Opens the Window to New Frontiers

Data Science Research Laboratories
TAI Tsenjung

Tai Tsenjung obtained her bachelor’s degree in Computer Science from the University of Hong Kong and subsequently a master’s degree in Information Technology from the Hong Kong University of Science and Technology. She joined NEC in 2019. She is an expert in AI and machine learning.

Taking satellite image technology to the next level

I have been working for the satellite image analysis team of NEC Data Science Lab. Satellite imaging is an important technology for ground activity and natural disaster monitoring. The most common satellite images are optical ones which are similar to what we see via Google Earth. I am working on another type called Synthetic-aperture radar (SAR) images. SAR radar is employed to generate two- or three-dimensional reconstructions of objects. SAR is capable of weather-independent wide-range remote sensing, in addition to its day and night imaging capacity. Applications of SAR are numerous, and one of them is in case of emergencies such as earthquakes and floods. It enables us to clearly estimate the degree of damage and help promptly organize rescue operations and reconstruction plans. The ongoing challenge is that satellite data currently available are expensive and barely standardized. That is the reason why I work toward developing machine learning-based object classification algorithms which are robust to data taken by different satellite sensors. My latest research is focused on utilizing transfer learning techniques to adapt knowledge from one dataset to another. In this way, not only the machine learning models can be trained to work for images of different standards, but also the number of SAR images required for training can be reduced. These are cutting-edge technologies with a huge potential to help overcome numerous challenges to our national security. I have an exciting journey ahead of me and am totally committed to taking this early-stage research to the next level.

Respect for our expertise and trust in our research capability

Since I joined NEC in 2019, I have been feeling well-integrated and respected as a data scientist. The company has shown profound respect to our expertise and belief in our research capability while always trying to make sure we have the freedom to choose topics that we think are worthy of pursuing. I had a chance to speak to some senior managers during my job interview and heard them emphasize the company’s trust in researchers. Our 20% Rule^[i] derived from their respect and trust in us. Under the rule, each one of us dedicates roughly 80% of our working time to current assignments and spends the rest of the 20% for our future innovation. Besides the satellite image analysis project, I have been engaged in another collaborative project which is to develop the surveillance camera with a team from NEC Lab Singapore. Both the surveillance camera project and the satellite image analysis project aim for accurate object recognition. However, due to different data formats and application scenarios, the former is addressed by reinforcement learning while the latter is addressed by deep learning. The advantage is since the two projects are highly-related, sometimes when I explore the methods used in one topic, I get inspired by the solution to the other topic. These are just a few among the multitudes of innovative topics that NEC has offered us. We are continuously inspired by the wealth of opportunities that help make the best use of our expertise to contribute to society’s well-being.

Playing an instrumental role as an international liaison

I see myself pursuing a research career at NEC in the foreseeable future. Going forward, I would like to continue to work on building algorithms capable of extracting desired information from a reduced size of data. This is imperative, given that satellite imaging data is quite expensive. Working on advanced technologies as such, we explore external perspectives and collaborative opportunities as well. Going into my third year, the company has already entrusted me with a larger role which is to go out and get connected with top-tier research universities and professors by attending international conferences. I enjoy playing an instrumental role as an international liaison while bringing in new ideas for future innovation. High expertise coupled with fluency in English is a huge advantage as the company continues to increase its global presence. At our research center, we communicate in English. Although the company offers Japanese lessons, it is not mandatory. We feel the company sees cultural diversity as a driver to continued success. It is great to know they try to create environments that are helpful for us to thrive while encouraging us to contribute with our diverse perspectives and difference of opinions.

A day at work

Message to my past self in my school days