📌print(“Hello World!”)

Profile Image

👋HI! I am Jaegun Lee. I am a M.S. student at KNU DataScience, majoring in Artificial Intelligence. I am working as an AI graduate researcher at Computer Vision at KNU, advised by Prof. Janghoon Choi.

I am deeply interested in AI research, particularly in Computer Vision. My work focuses on post-training methods such as reinforcement-based optimization, as well as Generative AI. Ultimately, I aim to create AI systems that deliver meaningful value by enabling users to imagine, build, and share their ideas freely.

Particularly, I am researching on the following topics:

1. Reinforcement

2. Generative AI

CV Portfolio E-mail Github

Email: leejken530@knu.ac.kr

📌News

Dec. 2025: Submitted Flow-Multi: A Flow-Matching Multi-Reward Reinforcement Learning Framework for Text-to-Image Generation to an SCI-indexed journal; currently under review.

Sep. 2025:SuperSmall-R1: A Lightweight Reinforcement Learning Model for Mathematical Reasoning was accepted to a KCI-indexed journal, validating reinforcement-learning-based reasoning in a compact model regime.

Jul. 2025:Small-Math-R1: A Lightweight Reinforcement Learning Model for Mathematical Reasoning was accepted to a KCC, showing the effectiveness of reinforcement learning on small-parameter models for math reasoning.

Apr. 2025:An End-to-End Framework for Automatic Captioning and Music Generation for Video Content was accepted to a ASK, showing the end-to-end framework for making music. if you want to see the project page, please go to the here

Jan. 2025: My MoE를 활용한 객체활용방법론 paper got accepted to IPIU!

Aug. 2024: Selected as a participant in Naver AI Tech Boostcamp (Cohort 7), engaging in industry-oriented projects on image classification, semantic segmentation, and model efficiency.Click_Here if you want to see projects

Mar. 2024: I got accepted to CV Lab as a M.S. graduate student in KNU DataScience.

📌Education

Kyungpook National University (KNU)

M.S. in Artificial Intelligence, Graduate School of Data Science; Mar. 2024 – Feb. 2026 (Expected)
GPA: 4.38 / 4.5 (Expected Graduation)

Korea National Open University (KNOU)

B.S. Statistics, Mar. 2023 ~ Feb.2025(Transfer university for Statistic Study)
GPA: 4.1/4.5

Gyeongguk National University (GNU)

B.S. in Environment Engineering and law, Mar. 2014 ~ Feb. 2022
GPA: 3.32 / 4.5

📌Publications

*: equal contribution. †: corresponding author. C: conference J: journal.</b> P: preprint.

[J2] Flow-Multi: A Flow-Matching Multi-Reward Reinforcement Learning Framework for Text-to-Image Generation

Flow-Multi

Jaegun Lee, Janghoon Choi†
Keywords: Text-to-Image Generation, Multi-Reward Reinforcement Learning, Flow Matching

Sensors (SCI_under review), Paper

[J1] SuperSmall-R1: A Lightweight Reinforcement Learning Model for Mathematical Reasoning

SuperSmall-R1

Jaegun Lee, Janghoon Choi†
Keywords: Mathematical Reasoning, Lightweight Models, Reinforcement Learning

KCI-indexed Journal, Paper

[C3] Small-Math-R1: A Lightweight Reinforcement Learning Model for Mathematical Reasoning

Small-Math-R1

Jaegun Lee, Janghoon Choi†
Keywords: Mathematical Reasoning, Small-Parameter Models, Reinforcement Learning

KCC 2025, Paper

[C2] An End-to-End Framework for Automatic Captioning and Music Generation for Video Content

V2M

Jaegun Lee, Taejun Kwon, Janghoon Choi†
Keywords: Video Captioning, Music Generation, Multimodal Learning

ASK 2025, Paper , Project Page

[C1] MoE-Based X-ray Object Classification Methodology

MoE

Jaegun Lee, Janghoon Choi†
Keywords: Medical Image Classification, Mixture-of-Experts

IPIU 2025, Paper

📌 Honors and Awards

If you want to see my awards Sep. 2025 — Future Research Award K-Data Science Conference, K-DS Consortium

Proposed SuperSmall-R1, a lightweight math reasoning model trained purely with reinforcement learning, achieving significant accuracy gains on Math-500.

Jan. 2024 — Excellence Prize Data Analysis Competition, Korea National Open University

Applied XAI techniques to improve interpretability of machine learning and deep learning models for practical deployment.

Sep. 2023 — Award Winner (Finance Division) 5th Daegu Big Data Analysis Competition, Daegu Digital Innovation Promotion Agency

Conducted time-series sales forecasting to identify high-performing but under-credited business sectors, suggesting risk-aware financial support strategies.

Aug. 2023 — Excellence Prize K-NU K-Digital Platform Hackathon, Kyungpook National University Data Convergence Research Institute

Designed and implemented a deep-learning-based sign language translation system, integrating video-to-text and STT pipelines to support communication for the hearing impaired.