Yeongmin Kim (김영민)
Yeongmin Kim (김영민)
Contact: alsdudrla10@kaist.ac.kr Homepages: Google Scholar, Github
📢 I am actively seeking a research internship in Generative AI for the first half of 2026.
I am a 4th-year Ph.D. candidate at the Korea Advanced Institute of Science and Technology (KAIST), advised by Prof. Il-Chul Moon. I received my B.S. in ISysE & CS in February 2022 from KAIST. I interned at Meta GenAI in 2024.
Research Interests
My research interests lie in Generative AI grounded in probabilistic principles. My work has focused on improving the robustness, efficiency, and steerability of diffusion models. Building on these insights, I am currently exploring how probabilistic modeling can enhance the training and inference of large language models.
Diffusion Models
Robust Training: [C3, C4, C5]
Efficient Inference: [C9, C10]
Controllable Generation: [W1, C6, C8]
Language Models
RL-based Fine-Tuning: [C7, W2, P1]
Others
Discriminative Tasks: [C1, C2]
[P1] Preference Optimization by Estimating the Ratio of the Data Distribution [pdf]
Yeongmin Kim, Heesun Bae, Byeonghu Na, Il-Chul Moon
[C10] Autoregressive Distillation of Diffusion Transformers [pdf][code]
Yeongmin Kim, Sotiris Anagnostidis, Yuming Du, Edgar Schönfeld, Jonas K Kohler, Markos Georgopoulos, Albert Pumarola, Ali Thabet, Artsiom Sanakoyeu
Conference on Computer Vision and Pattern Recognition (CVPR 2025) - Oral (Top 0.7%)
[C8] Diffusion Bridge AutoEncoders for Unsupervised Representation Learning [pdf][code]
Yeongmin Kim, Kwanghyeon Lee, Minsang Park, Byeonghu Na, Il-Chul Moon
International Conference on Learning Representations (ICLR 2025) - Spotlight (Top 5.1%)
[C5] Training Unbiased Diffusion Models From Biased Dataset [pdf][code]
Yeongmin Kim, Byeonghu Na, Minsang Park, JoonHo Jang, Dongjun Kim, Wanmo Kang, Il-Chul Moon
International Conference on Learning Representations (ICLR 2024)
[C3] Refining Generative Process with Discriminator Guidance in Score-based Diffusion Models [pdf][code][media1][media2]
Dongjun Kim*, Yeongmin Kim*, Se Jung Kwon, Wanmo Kang, Il-Chul Moon
International Conference on Machine Learning (ICML 2023) - Oral (Top 2.3%)
Graduate School of Data Science (GSDS)
Transferred from an MS program in 2023.
Topic: Financial AI
Topic: Diffusion Model, Efficient Inference
Manager: Artsiom Sanakoyeu
Preference Optimization by Estimating the Ratio of Data Distribution, Tech Talk at Upstage AI, July 18, 2025.
International Conference on Machine Learning (ICML) 2024, 2025
Neural Information Processing System (NeurIPS) 2024, 2025
International Conference on Learning Representations (ICLR) 2025
Conference on Computer Vision and Pattern Recognition (CVPR) 2025
International Conference on Artificial Intelligence and Statistics (AISTATS) 2025
FPI@ICLR 2025
SPIGM@ICML 2023, 2024, NeurIPS 2025
NCW@ICML 2023