ON CAMPUS x OFF CAMPUS

We Support Students’ Activities and Introduce Them to You

Landslide Assessment and Disaster Prevention by Weakly Supervised Machine Learning

Most of Japan’s terrain is mountainous, and people often reside close to the mountains. Consequently, landslides and mudslides are some factors that cause substantial damage. There has been an increase in local downpours and sudden downpours caused landslides and mudslides in recent years.

Chenzuo Ye, a student in the Department of Natural Environment, recognized the difficulty of forecasting natural disasters during the 2018 West Japan Torrential Rain. This experience motivated him to start a project that aims to provide new disaster prevention measures. The method involves using weakly supervised machine learning to accurately predict the risk of landslides.

Although deep learning is commonly used for predictions of sediment-related disasters owing to its high accuracy, extensive amounts of data, time, and money are required. Furthermore, the factors that cause landslides are diverse, and accumulated observation data are not necessarily accurate.

Ye proposed using weakly supervised machine learning, despite the incompleteness of data collection and pre-processing, to train prediction models. He aims to achieve more accurate prediction methods by considering a wide range of information, different from the conventional manners.

Ye’s goal is to protect people’s lives and property and, ultimately, to make the entire society resilient to natural disasters by providing natural disaster risk predictions and solutions to communities.

“I aspire to create a society where technology contributes to social welfare and provides effective solutions for natural disaster prevention,” he enthusiastically told us about his dream.

*This research activity is partially supported by the GSFS Student Project fund.

(Original Japanese text by Mayuko Araragi)

Kashiwa Campus Science Camp

https://ksc.edu.k.u-tokyo.ac.jp/

The 8th UTokyo Kashiwa Campus Science Camp was successfully held in collaboration with 36 laboratories of the GSFS and the affiliated centers in the 2022 academic year. The program welcomed 135 first-year and second-year undergraduate students from liberal arts courses at the University of Tokyo for four days of face-to-face training. Student feedback indicated their positive opinions of the program.

Students’ voices:

“The program broadened my views and improved my understanding of research and graduate school life.”
“The program helped me solidify my ideas for my future career.”

Professors’ voices:

“The trainings are meaningful for students.”
“We received students from both science and humanity courses at the same time. I was glad that they were very eager about research.”

　

Souiki-kai Student Department Circles

http://souiki-kai.net/

For the first time in three years, the GSFS alumni and students’ association, Souiki-kai’s Student Department, held an in-person guidance session for circles, with eight unique circles showcasing their activities. As there are now more chances to remove face masks than in the past three years, offline events and exchanges will increase this year. The Student Department is also planning to hold regular events, which students are eagerly anticipating. To learn more about the active groups and circles at Kashiwa Campus, visit the Soiki-kai website.