About me
I am an Assistant Professor in the Department of Artificial Intelligence at Kookmin University, where I lead the Probabilistic Machine Learning Lab (PML). Previously, I was a CIFAR AI Safety Postdoctoral Fellow and a postdoctoral research fellow in the Department of Computer Science at the University of British Columbia (UBC). I received my Ph.D. in Artificial Intelligence from KAIST under the supervision of Prof. Se-Young Yun. Before starting my Ph.D., I worked as a research engineer at Samsung Heavy Industries. I also received an M.S. in Ocean Systems Engineering from KAIST under the supervision of Prof. Hyun Chung, and a B.S. in Industrial Engineering from Konkuk University. My research interests include safe AI, generative models, probabilistic machine learning, generalization, few-shot learning, meta-learning, Bayesian models, and variational inference. Here is my CV.
Contact
Email : mgyukim [at] kookmin.ac.kr
Research Experience
[2026-] Assistant Professor in Department of Artificial Intelligence at Kookmin University.
[2025-2026] CIFAR AI Safety Postdoctoral Fellowship by CIFAR AI Safety.
[2024-2026] Postdoctoral Fellowship by UBC Data Science Institute.
[2024-2026] Postdoctoral Research Fellow at UBC CS.
[2023] Visiting PhD Student at DTU Congnitive System hosted by Prof. Søren Hauberg.
[2023] Research Intern at NAVER Cloud AI Lab.
[2014-2019] Research engineer at Samsung Heavy Industries.
[2021-2024] AI Advisor at Actnova
Awards
[2025] Top Reviewer at NeurIPS2025.
[2025] Best Reviewer at AISTATS2025.
[2021] 2nd Winner in Student track at Machine Learning for Combinatorial Optimization Challenge (ML4CO) in NeurIPS2021 (Team : KAIST_OSI, Team leader).
[2016] Outstanding Project aimed at Increasing Productivity at Samsung Heavy Industires.
Conference Publications (Peer reviewed)
Safety-Guided Flow (SGF): A Unified Framework for Negative Guidance in Safe Generation
[paper][code]
Mingyu Kim, Young-Heon Kim and Mijung Park
International Conference on Learning Representations (ICLR) 2026, Rio de Janeiro
Oral Presentation (Top 1.2% of ~19,000 submissions)
Training-Free Safe Denoisers For Safe Use of Diffusion Models
[paper][code]
Mingyu Kim*, Dongjun Kim*, Amman Yusuf, Stefano Ermon and Mijung Park
Advances in Neural Information Processing Systems (NeurIPS) 2025, San Diego
Bayesian Principles Improve Prompt Learning In Vision-Language Models
[paper][code]
Mingyu Kim*, Jongwoo Ko* and Mijung Park
International Conference on Artificial Intelligence and Statistics (AISTATS) 2025, Mai Khao
Factorized Multi-Resolution HashGrid for Efficient Neural Radiance Fields: Execution on Edge-Devices and Large Scale Environments
[paper][projectpage][conference page]
Jun-Seong Kim*, Mingyu Kim*, Geon-U Kim, Tae-Hyun Oh† and Jin-Hwa Kim† (* : Co-first authors, † : Co-corresponding authors)
International Conference on Robotics & Automation (ICRA), 2025, Atlanta
IEEE Robotics and Automation Letters, 2024
Synergistic Integration of Coordinate Network and Tensorial Feature for Improving Neural Radiance Fields from Sparse Inputs
[paper][code][projectpage]
Mingyu Kim, Jun-Seong Kim, Se-Young Yun† and Jin-Hwa Kim† († : Co-corresponding authors)
International Conference on Machine Learning (ICML) 2024, Vienna
Neural Processes with Stochastic Attention: Paying more attention to the context dataset
[paper][code][video]
Mingyu Kim, Kyeong Ryeol Go and Se-Young Yun
International Conference on Learning Representations (ICLR) 2022, Virtual
Journal Publications (Peer reviewed)
PolarGAN: Creating Realistic Arctic Sea Ice Concentration Images with User-Defined Geometric Preferences
[paper]
Mingyu Kim, Jaekyeong Lee, Leechan Choi and Minjoo Choi
Engineering Applications of Artificial Intelligence, Vol. 126, 2023
The StarCraft Multi-Agent Exploration Challenges : Learning Multi-Stage Tasks and Environmental Factors without Precise Reward Functions
[paper][code][projectpage]
Mingyu Kim*, Jihwan Oh*, Yongsik Lee, Joonkee Kim, SeongHwan Kim, Song Chong and Se-Young Yun
IEEE Access, Vol. 11, 2023
Simplified welding distortion analysis for fillet welding using composite shell elements
[paper]
Mingyu Kim, Minseok Kang and Hyun Chung
International Journal of Naval Architecture and Ocean Engineering, Vol. 7(3), 2015
Meta-learning Amidst Heterogeneity and Ambiguity
[paper]
KyeongRyeol Go, Mingyu Kim, and Se-young Yun
IEEE Access, Vol. 11, 2023
Optimization of P.E area division and arrangement based on product mix
[paper]
Sanghwan Kim, Mingyu Kim and Hyun Chung
Journal of Marine Science and Technology, Vol. 19(4), 2014
Workshop Publications (Peer reviewed)
SteeringTTA: Guiding Diffusion Trajectories for Robust Test-Time-Adaptation
[paper]
Jihyun Yu, Yoojin Oh, Wonho Bae, Mingyu Kim† and Junhyug Noh† († : Corresponding authors)
Putting Updates to the Test Workshop in International Conference on Machine Learning (ICML) 2025, Vancouver
Training-Free Safe Denoisers For Safe Use of Diffusion Models
[paper]
Mingyu Kim*, Dongjun Kim*, Amman Yusuf, Stefano Ermon and Mi Jung Park
SynthData Workshop in International Conference on Learning Representations (ICLR) 2025, Singapore
Refined Tensorial Radiance Field: Harnessing Coordinate-Based Networks for Novel View Synthesis from Sparse Inputs
[paper]
Mingyu Kim, Jun-Seong Kim, Se-Young Yun†, Jin-Hwa Kim†
Learning-Based Solutions for Inverse Problems Workshop in Neural Information Processing Systems (NeurIPS) 2023, New Orleans
The StarCraft Multi-Agent Challenges+ : Learning Multi-Stage Tasks and Environmental Factors without Precise Reward Functions
[paper][code][projectpage]
Mingyu Kim*, Jihwan Oh*, Yongsik Lee, Joonkee Kim, SeongHwan Kim, Song Chong and Se-Young Yun
AI for Agent Based Modeling Workshop(Spotlight) in International Conference on Machine Learning (ICML) 2022, Baltimore
AVATAR: AI Vision Analysis for Three-dimensional Action in Real-time
Daegun Kim*, Jineun Kim*, Wongyo Jung*, Jungoon Park*, Mingyu Kim*, Anna Shin, Yong-Cheol Jeong, Seahyung Park, Gwanhoo Shin, Yewon Lee, Jea Kwon, Daesoo Kim
CV4Animals Workshop in Computer Vision and Pattern Recognition Conference (CVPR) 2022, New Orleans
Neural Processes with Stochastic Attention: Paying more attention to the context dataset
[paper][code]
Mingyu Kim, Kyeong Ryeol Go and Se-Young Yun
5th Meta-learn Workshop in Neural Information Processing Systems (NeurIPS) 2021, Virtual
Domestic Journal Publications (Peer reviewed)
Optimization of Quantity Allocation using Integer Linear Programming in Shipbuilding Industry
[paper]
Junggoo Park, Mingyu Kim
Journal of the Society of Naval Architecture of Korea (Domestic), Vol. 57(1), 2020
A Study on Process Management Method of Offshore Plant Piping Material using Process Mining Technique
[paper]
Junggoo Park, Mingyu Kim, Jichan Park
Journal of the Society of Naval Architecture of Korea (Domestic), Vol. 56(2), 2019
A Study on the Installation Readiness Management Method of Offshore Plant using CAD Information
[paper]
Junggoo Park, Hojung Kim, Mingyu Kim, Jichan Park
Journal of the Society of Naval Architecture of Korea (Domestic), Vol. 56(2), 2019
A Study on the Utilization Strategy of Social Commerce in Small Restaurants
[paper]
Sungseok Ko, Mingyu Kim, Yungjin Yoo, Seoksul Kim
Journal of the Society of Korea Industrial and Systems Engineering (Domestic), Vol. 35(1), 2012
Academic Services
[Conference Reviewer] NeurIPS{2022,2024,2025}, ICML{2025,2026}, ICLR{2023,2025,2026}, CVPR{2024,2025,2026}, AISTATS{2025,2026}, ICCV·ECCV{2025,2026}, AAAI{2025,2026}, WACV2026
[Workshop Reviewer] MetaLearn in NeurIPS2021, GRaM in ICML2024
[PhD Application Reviewer] ELLIS-PhD Program 2024