Suwan Yoon

Suwan Yoon

Suwan Yoon

Graduate Student (M.S./Ph.D. Integrated) at the Actionable Intelligence Lab, Korea University.
Former undergraduate at the Department of AI, Chung-Ang University (CAU).

Google Scholar GitHub LinkedIn Email CV

About Me

I’m Suwan Yoon, a graduate student in the M.S./Ph.D. integrated course at Korea University, working at the Actionable Intelligence Laboratory led by Prof. Changhee Lee.

My work centers on:

  • AI4Healthcare
  • AI4Science
  • LLMs

I also enjoy exploring theoretical areas, such as Optimal Transport & Generative Modeling. My previous work has centered on time-series analysis, visual language models, multimodal VAEs, and knowledge distillation.

🎓 Education

Korea University — M.S./Ph.D. Integrated Course in Artificial Intelligence
📆 Mar 2026 – Present
(Actionable Intelligence Lab, led by Prof. Changhee Lee)

Chung-Ang University — B.S. in Artificial Intelligence
📆 Mar 2022 – Feb 2026
GPA: 4.19 / 4.5 (Major: 4.29 / 4.5)
Coursework: Linear Algebra, Algorithm, Pattern Recognition, Machine Learning, Neural Network, Image/NLP/Speech Processing, etc.

💼 Experience

Research Intern — Actionable Intelligence Lab @ Korea University (Seoul, Korea)
📆 Sep 2024 – Feb 2026 (Led by Prof. Changhee Lee)

  • In collaboration with ETRI, built Time Series Forecasting models for Proactive Handover.
  • In collaboration with Severance Hospital, analyzed Lifelog data from insomnia patients.
  • Followed up on ML/DL papers and studied Stochastic Differential Equations.

Research Intern — Decision Intelligence Lab @ Chung-Ang University (Seoul, Korea)
📆 Jan 2024 – Aug 2024 (Led by Prof. Changhee Lee)

  • Studied machine learning and deep learning (CS4780).
  • Researched a deep learning-based time-series forecasting model in the medical field.

🧾 Publications

Top-Tier Conference

ToDi: Token-wise Distillation via Fine-Grained Divergence Control
📆 Oct 2025 — EMNLP 2025 (Main, 🔴 Oral Presentation)
Authors: Seongryong Jung, Suwan Yoon, DongGeon Kim, Hwanhee Lee

  • 🏅 Outstanding Paper Award Finalist — only 75 / 1,811 papers nominated
  • ⭐ Overall Rating: 4/5  |  SAC: 9/10 (top 15% of accepted papers)
    🔗 arXiv:2505.16297

Preprints

Agentic Molecular Recovery via Molecule-Aware Exploration
📆 Jun 2026 Authors: Suwan Yoon, Changhee Lee

🔗 arXiv:2606.05847

Localizing Input Uncertainty Quantification for Large Language Models via Shapley Values
📆 May 2026 Authors: Seongjun Lee*, Suwan Yoon*, Changhee Lee

🔗 arXiv:2605.28170

Conference

Towards Speed-Agnostic Time-Series Forecasting for Proactive Handover
📆 Oct 2025 — ICTC 2025 (🔴 Oral Presentation)
Authors: Junseo Lee*, Suwan Yoon*, Changhee Lee

🔗 IEEE Xplore

IMMU: IMage to MUsic Sequence for Social Network Service
📆 Oct 2024 — DCSC Fall 2024 (🏅 Student Paper Award, Silver Prize)
Authors: Minsung Kim*, Taehwan Kim*, Suwan Yoon*, Kisung Lee

* Equal contribution.

🚀 Projects

[Research] Biochemical Agentic AI (2025.11 – Present)

Developing LLM-driven agentic AI systems for biochemical discovery. This research explores LLMs as interfaces and reasoning engines that leverage intrinsic scientific knowledge to guide molecule-aware exploration, molecular generation, molecular recovery, and automated drug-discovery workflows.

Supported by NIPA’s Advanced GPU Utilization Support Program, the IITP AI Star Fellowship Support Program, and the 2026 AI Seoul Tech Research Support Program.

Agentic AI LLMs Molecular Generation Molecular Recovery Drug Discovery

[Research] Collaboration w/ ETRI (2025.01 – 2025.12)
Time-series forecasting for proactive mobile-device handover. Developing a domain-agnostic model robust to speed prediction under changing environments, cell-deployment-agnostic approaches invariant to feature permutation and resilient to missing features.
This project was presented at ICTC 2025.
Time Series Forecasting Wireless Communications ETRI

[Research] Embedding Space of VLM (2025.03 – 2025.07)
Collaborated with Prof. Junhyuk Kim (CAU) on VLM for semantic segmentation. Proposed two novel losses (HNS, UAI) for fine-tuning CLIP and one metric (Unmix Rate) for evaluating CLIP Zero-shot semantic segmentation. Achieved State-of-the-Art on CLIP ZS3 task.
This project was presented at CAU Engineering Conference 2025.
VLM Zero-Shot Segmentation Computer Vision

KYS — Keep Your Smile!
Facial Emotion Recognition on Jetson Nano, optimized via pruning and quantization. Reports performance degradation trends caused by model compression on edge devices.
This project was presented at DCSC Fall 2025.
Embedded AI Model Pruning Quantization

RepHeat
A routine timer app that visualizes daily repetitions as a heat map to motivate users. Built with Flutter.
Flutter

IMMU: SNS with Image-to-Music Sequence
Transforms images into music using Riffusion (diffusion-based), BLIP (image captioning), and GPT-3 (sentence generation). This project was presented at DCSC Fall 2024.
🏆 CCSSAA Idea-thon 2023 3rd Prize.
Image Captioning Generative AI Flutter

🏅 Scholarship & Awards

YearAward
2026AI Seoul Tech Scholarship — Seoul Future Foundation (20,000,000 KRW)
2025CAU Engineering Conference — Software University President’s Award (400,000 KRW)
2025Junior Department Honor Scholarship — Full-Funded (1st Semester)
2024DCSC Student Paper Award — Silver Prize
2023CCSSAA Idea-thon — Third Prize (700,000 KRW)

🎸 Leadership

LAON (Live at Our Night) — Student Band, Dept. of AI, Chung-Ang University
Founder & Leader · Nov 2023 – Feb 2026
Founded the student band and led the planning and execution of four performances.

🛠️ Technologies

Languages: Python, C, Java, Dart (Flutter)
ML/DL: PyTorch, NumPy, Pandas, Scikit-Learn, Matplotlib, Seaborn
Tools: Git, Docker, Kubectl

📬 Contact