Information Design and Animal to Animal Information Network utilized outer urban infrastructure.(Prof. Kobayashi)
To solve environmental problems that occur in spaces where information communication technologies are difficult to be introduced, we are researching systems to realize holistic advanced information processing, in which the information space and the ecosystem are inextricably integrated.
This research includes information design of animal IoT for wild animals living in the exclusion zone around the Fukushima Daiichi Nuclear Power Plant.
Large-scale graph neural networks and its applications to various domains.(Prof. Suzumura)
Graph is a powerful data structure that can represent relationships of various entities such as things and humans in the digital and physical world. We work on representation learning on graph-structure data that can be leveraged for downstream machine learning tasks for real-world problems including recommender system, anomaly detection, material discovery, and so forth.
Adapting software vulnerability testing for data center hardware.(Assoc. Prof. Kuga)
Datacenter computing environments use modern high-speed interconnects and accelerator devices to improve computing performance. We’re developing a method for early detection of hardware defects and vulnerabilities to safely use data center hardware such as GPU, SSD, and NIC.