페이스북 인스타그램 유튜브 검색 버튼

About Us

Trending

Prof. Lee Sangmin’s Research Team Wins the ICIEA 2026 “Best Oral Presentation Award”

  • 국제교류팀
  • 2026-05-14
  • 57

Professor Lee Sang-min’s Research Team (Division of Information Convergence) Wins the ICIEA 2026 “Best Oral Presentation Award”

 

 


 

The research team led by Professor Lee Sangmin of the Division of Information Convergence — including doctoral student Jung Hwa-yong and researcher Jang Yu-na from the Artificial Intelligence Service Laboratory — received the Best Oral Presentation Award at the International Conference on Industrial Engineering and Applications (ICIEA) 2026, held in Kyoto, Japan, from April 9 to 12.

 

Jung Hwa-yong presented a Multiple Heterogeneous Teacher-based Knowledge Distillation (MHTKD) methodology designed to achieve robust object detection performance across diverse environments. The proposed framework generates highly reliable soft supervision by integrating prediction results from heterogeneous teacher models with different inductive biases.

 

Through this approach, MHTKD demonstrated superior object detection performance compared to conventional single-teacher-based knowledge distillation methods, both in settings where teacher and student models were similar in size and in settings with significant size differences. The study also verified the effectiveness of integrating complementary knowledge from heterogeneous teacher models.

 

Researcher Jang Yu-na presented the ZAAS methodology, which improves the performance of SAM-adapter optimization for adapting the object segmentation foundation model SAM (Segment Anything Model). The study demonstrated enhanced prediction performance of the model in specialized domains.

 

In particular, the proposed methodology enabled rapid convergence while ensuring a broad exploration range through the construction of an efficient search space and the application of a perturbation-based search technique. By improving the fine-tuning performance of existing SAM models in fields such as medical imaging, the research proved that exploring effective adapter architectures tailored to each domain can lead to substantial performance improvements.

 

At ICIEA, an international conference that promotes collaboration among global researchers by sharing the latest theories, technologies, and practical applications in industrial engineering, the research team gained recognition for its outstanding achievements through active presentations and academic exchanges. This award once again demonstrated the team’s strong research competitiveness and contributions in the field.