The Department of Computer Science & Engineering and Engineering’s Optimization Lab won the Best Paper Award at the Korea Artificial Intelligence Society Fall Conference.

The Department of Computer Science & Engineering and Engineering’s Optimization Research Lab (advisor: Professor Namhoon Lee ) won the Korea South-East Power Best Paper Award at the ‘2025 Korean Artificial Intelligence Society Fall Conference’ held at COEX in Seoul on Nov. 20th. This year, two Best Paper Awards and two Excellent Paper Awards were selected, and the Optimization Research Lab’s Research Affairs and Project Management was Name as one of the two Best Paper Awards.

The award-winning paper, ‘The Unseen Frontier: Pushing the Limits of LLM Sparsity with Surrogate-Free ADMM,’ was highly praised for implementing a new optimization technique to achieve a 90% high-sparsity region that previous Research Affairs and Project Management had not been able to reach in the field of large-scale language model (LLM) sparsification.

The research team proposed a new constrained optimization sparsification technique based on the objective function, and demonstrated through theoretical analysis and experiments that this method achieves superior performance compared to existing approaches. In particular, in a 90% sparsification experiment targeting the LLaMA-2-7B model, they confirmed 7.8 times lower perplexity, approximately 2.5 times faster inference speed, and 4.6 times higher memory compression ratio compared to existing approaches, suggesting the possibility of reducing LLM operating costs and improving efficiency.

In addition to this achievement, the Optimization Research Lab led by Professor Namhoon Lee has also won awards such as the Outstanding Paper Award at the 2025 Summer Conference of the Korean Artificial Intelligence Society and the Best Paper Award at the 2023 Fall Conference, proving its competitiveness at the highest level in Korea in LLM optimization and theory-based model efficiency Research Affairs and Project Management.

[Laboratory Information]

1) Optimization Lab Introduction Video: YouTube Link (Click)
2) [Recruiting] Participate in the 2025–26 Winter Internship Program: Check (Click)
※ The program will be announced regularly through the Computer Science & Engineering Internship Bulletin Board.