Seunghoon Hong

Seunghoon Hong is an Associate Professor at the School of Computing, Korea Advanced Institute of Science and Technology (KAIST), where he leads the KAIST Vision and Learning Lab (VLLab@KAIST). His research focuses on advancing machine learning and computer vision with emphasis on reducing human supervision requirements and enhancing model interpretability. Prior to joining KAIST, he was a visiting faculty researcher at Google Brain and completed a postdoctoral fellowship at the University of Michigan working with Professor Honglak Lee.

Education:

• PhD in Computer Science and Engineering, 2017, POSTECH, Korea (Supervised by Professor Bohyung Han)
• BS in Computer Science and Engineering, 2011, POSTECH, Korea

Professor Hong's research primarily explores scaling up machine learning algorithms for visual perception while minimizing human supervision requirements. He investigates techniques to make machine learning algorithms more interpretable, enabling users to better understand and participate in the decision-making process. His work spans few-shot learning, generative models, vision transformers, and equivariant neural networks, with applications across various computer vision tasks.

His recent publications reveal a strong focus on advancing foundation models for computer vision with emphasis on data efficiency, interpretability, and generalization. Key themes include developing novel approaches for few-shot learning, improving model efficiency through token merging and compositionality, and creating robust vision systems capable of handling real-world challenges like weather degradation. His work frequently appears in top-tier conferences including NeurIPS, ICLR, CVPR, and ECCV.

Scientific Awards:

• Outstanding Paper Award at ICLR 2023

Professor Hong actively mentors a substantial group of students, including multiple Ph.D. candidates, master's students, and undergraduates. He serves in leadership roles for major conferences, including as Senior Area Chair for NeurIPS 2025 and Workshop Chair for ICCV 2025. His research has significant practical implications for developing more efficient, interpretable, and adaptable computer vision systems.

He leads the KAIST Vision and Learning Lab, which focuses on cutting-edge research at the intersection of machine learning and computer vision, with particular emphasis on developing algorithms that require minimal human supervision while maintaining high performance and interpretability.