Korea has become one of Theta EdgeCloud’s most active regions. From research collaborations with KAIST, Seoul National University and Yonsei University to AI agent deployments for K League clubs FC Seoul and Ulsan HD FC, our footprint in the country continues to grow.
It is with great pride that we welcome the AI Vision Lab (AIV) and Professor Youngwook Kim at Kookmin University in Seoul to our global academic network.
Professor Kim is an Assistant Professor in the School of Artificial Intelligence at Kookmin University’s College of Computer Science, where he has led the AI Vision Lab since 2024. He earned his PhD in Electrical and Computer Engineering from Seoul National University through its integrated M.S./Ph.D. program, with a dissertation on training strategies and benchmarks for weakly supervised multi-label classification, followed by postdoctoral research at SNU’s Institute of New Media and Communications. He serves as a reviewer for leading venues in the field, including IEEE TPAMI, IEEE Transactions on Multimedia, CVPR, ICCV and ECCV.
Robust Vision from Imperfect Data
The AI Vision Lab’s research rests on two pillars:
Applied Vision
Puts image data to work across real-world domains.
Intelligent Vision
Builds models that learn robustly even from incomplete or noisy information.
That work spans geospatial, autonomous driving, medical, multimodal and data-robust AI, with current projects including cloud removal for satellite imagery. In 2025, the lab was selected for both a national Satellite Image Utilization Research Program and the AI Star Fellowship Program.
Professor Kim’s CVPR 2022 paper, Large Loss Matters in Weakly Supervised Multi-Label Classification (arXiv, code), showed that models trained on partially labeled images first learn the clean signal and then start memorizing the noise, and proposed simple methods to prevent it that outperformed prior state-of-the-art approaches across major benchmarks. Research like this is compute-hungry by nature, demanding sustained GPU capacity for large datasets, repeated experiments and high-resolution imagery.
We are excited to welcome Kookmin University and the AI Vision Lab to Theta’s global academic partner network. Professor Kim’s work on robust learning from imperfect data addresses one of the most practical challenges in applied AI, from satellite imaging to autonomous driving and healthcare. We look forward to supporting his researchers and students with flexible, scalable GPU infrastructure.
Why Theta EdgeCloud
Theta EdgeCloud provides distributed GPU capacity through a global network of more than 30,000 edge nodes, delivering on-demand, high-performance computing resources at a fraction of the cost of traditional cloud infrastructure.
Through this collaboration, the AI Vision Lab will be able to access decentralized computing resources for model training, large-scale benchmarking, and experimentation across its geospatial, autonomous driving and medical imaging research.
Computer vision research increasingly depends on the ability to run many experiments at scale. Through this collaboration with Theta Network, our lab will be able to train and evaluate models on larger datasets, iterate faster, and give our students hands-on experience with distributed AI infrastructure. We look forward to exploring how Theta EdgeCloud can accelerate our work in applied and intelligent vision.
Looking Ahead
Kookmin University adds another leading Seoul research institution to Theta EdgeCloud’s growing academic network in Korea.
Theta Network looks forward to supporting Professor Kim and the AI Vision Lab team as they advance research in geospatial AI, autonomous driving, medical imaging and data-robust learning.
To access affordable compute for your research institution, startup or enterprise, please visit thetaedgecloud.com.
