Giung Nam
Curriculum Vitae
Postdoctoral Researcher @ KAIST AI
Information & Electronics Research Institute
- giung [at] kaist [dot] ac [dot] kr
- cs-giung.github.io
- Seoul, South Korea
Positions
Postdoctoral Researcher
Information & Electronics Research Institute, KAIST
Mar 2026 – Feb 2027 (expected)
Fulfilling mandatory military service; mentoring graduate students in research.
Research Intern
Visual Intelligence Laboratory, ETRI
Summer 2019; Winter & Summer 2020
Developed AI-based software for detection, segmentation, and face recognition.
Research Intern
Research and Engineering Team, VisualCamp
Spring 2019
Developed AI-based software for eye tracking.
Education
KAIST
Ph.D., Kim Jaechul Graduate School of AI
Sep 2022 – Feb 2026
ThesisExploiting prior knowledge in pre-trained models
CommitteeJuho Lee (Advisor), Eunho Yang, Jong Chul Ye, Sehoon Kim, Se-Young Yun
KAIST
M.S., Kim Jaechul Graduate School of AI
Sep 2020 – Aug 2022
ThesisDiversity matters when learning from ensembles
CommitteeJuho Lee (Advisor), Eunho Yang, Sung Ju Hwang
Korea University
B.S., Department of Computer Science and Engineering
Mar 2017 – Aug 2020
Honors & Awards
Kim Younghan Global Leader Fellowship · KAIST
2023
For outstanding academic excellence, research capabilities, and leadership qualities.
Director General's Award · NIMS
2021
For contributions to an AI-based emulator for physics processes in numerical models.
Publications
* Co-first author; † Co-corresponding author
Conference Papers
- [C16]
SKIM: pruning large language model agents via selective knowledge informed masking
Conference on Neural Information Processing Systems, Dec 2026 (To appear).
- [C15]
SAVE: sparsity-aware influence estimation for vocabulary-expanded LLMs
Conference on Neural Information Processing Systems, Dec 2026 (To appear).
- [C14]
Example-based spatial guidance for training-free concept erasure in diffusion models
Conference on Neural Information Processing Systems, Dec 2026 (To appear).
- [C13]
OCNR: stabilizing self-play by mitigating iteration-collapse with one-class novelty rewards
International Conference on Machine Learning, Jul 2026.
- [C12]
PANGEA: projection-based augmentation with non-relevant general data for enhanced domain adaptation in LLMs
Conference on Neural Information Processing Systems, Dec 2025.
- [C11]
Ensemble distribution distillation via flow matching
International Conference on Machine Learning, Jul 2025.
- [C10]
Parameter expanded stochastic gradient Markov chain Monte Carlo
International Conference on Learning Representations, Apr 2025.
- [C9]
Ex uno pluria: insights on ensembling in low precision number systems
Conference on Neural Information Processing Systems, Dec 2024.
- [C8]
Sparse weight averaging with multiple particles for iterative magnitude pruning
International Conference on Learning Representations, May 2024.
- [C7]
Enhancing transfer learning with flexible nonparametric posterior sampling
International Conference on Learning Representations, May 2024.
- [C6]
Lipsum-FT: robust fine-tuning of zero-shot models using random text guidance
International Conference on Learning Representations, May 2024.
- [C5]
Traversing between modes in function space for fast ensembling
International Conference on Machine Learning, Jul 2023.
- [C4]
Martingale posterior neural processes
International Conference on Learning Representations, May 2023.
- [C3]
Decoupled training for long-tailed classification with stochastic representations
International Conference on Learning Representations, May 2023.
- [C2]
Improving ensemble distillation with weight averaging and diversifying perturbation
International Conference on Machine Learning, Jul 2022.
- [C1]
Diversity matters when learning from ensembles
Conference on Neural Information Processing Systems, Dec 2021.
Journal Articles
- [J1]
Benefits of stochastic weight averaging in developing neural network radiation scheme for numerical weather prediction
Journal of Advances in Modeling Earth Systems, Oct 2022.
Workshop Contributions
- [W2]
ASCG: adaptive spatial classifier guidance for surgical concept suppression in diffusion models
Second Edition of Workshop on Synthetic & Adversarial ForEnsics, June 2026.
- [W1]
Improving constrained language generation via self-distilled twisted sequential Monte Carlo
Second Edition of Frontiers in Probabilistic Inference: Learning meets Sampling, Dec 2025.
Preprints
- [P3]
Rationale distribution distillation via flow matching for distributional continuous latent reasoning
arXiv:2610.19871, Oct 2026.
- [P2]
Harness-aware distillation for small language model agents
arXiv:2610.02858, Oct 2026.
- [P1]
Proper scoring rule-based diffusion for probabilistic weather forecasting
arXiv:2609.38632, Sep 2026.
Academic Service
Area Chair
- ICLR2027
- ICML2027
Reviewer
- AAAI2025, 2027
- AISTATS2025, 2026, 2027
- TMLR2025, 2026
- ICLR2024, 2025, 2026
- IJCAI2024
- ICML2022, 2023, 2024, 2025, 2026
- NeurIPS2022, 2023, 2024, 2025, 2026
Recognition
- Gold Reviewer, ICML2026
- Top Reviewer, NeurIPS2022, 2023, 2024
Teaching
- Bayesian Machine Learning · Teaching Assistant · KAIST AI (Instructor: Juho Lee)2021
- Machine Learning for AI · Teaching Assistant · KAIST AI (Instructor: Juho Lee)2021
- Deep Learning · Teaching Assistant · KAIST AI (Instructor: Jaesik Choi)2020
Talks & Presentations
- Institute of Statistics at Korea University · Seoul, South Korea2026
- 2026 Global AI Frontier Lab Workshop · Seoul, South Korea2026
- 2026 Global AI Frontier Lab Workshop at KAIST · Seoul, South Korea2026
- AI SEOUL 2024 · Seoul, South Korea2024
- Samsung Electronics DS Division · Hwaseong, South Korea2023
- KAIST AI Workshop 21/22 · Daejeon, South Korea2022