PeopleofSuperintelligence

Researcher

Il-Chul Moon

Also known as Ilchul Moon, Il Chul Moon

Il-Chul Moon is a KAIST professor whose research on generative and probabilistic models spans diffusion models, representation learning, and the alignment and distillation of large language models. He earned an engineering degree at Seoul National University in 2004 and a doctorate at Carnegie Mellon University in 2008, then worked at KAIST in Daejeon as a postdoctoral researcher and faculty member. With Kathleen Carley and Tag Gon Kim, he published a 2013 book on modeling and simulating command and control for organizations under extreme situations. His 2023 ICML oral paper introduced discriminator guidance for score-based diffusion models, followed by 2024 and 2025 work on unbiased diffusion models, diffusion bridge autoencoders and Bregman preference optimization, and 2026 ICLR papers on Wasserstein policy regularization and concrete score matching distillation. Promoted to full professor in 2025, he is a key faculty member of KAIST's AI College.

Now at
KAIST

Timeline

  1. 2004–2008

    Seoul National and Carnegie Mellon

    He earns a B.E. from Seoul National University in 2004, then completes a Ph.D. at Carnegie Mellon University in 2008.

  2. 2008–2017

    Joining KAIST in Daejeon

    From 2008 to 2011 he is a postdoctoral researcher at KAIST in Daejeon, and from 2011 to 2017 an assistant professor there. By 2015, he is an assistant professor in KAIST's Department of Industrial and Systems Engineering, where his theoretical research focuses on the overlap of computer science, management, sociology and operations research.

  3. 2013

    A book on command and control

    Springer publishes his 2013 book Modeling and Simulating Command and Control: For Organizations Under Extreme Situations, written with Kathleen M. Carley and Tag Gon Kim. The book shows how to design and run virtual experiments on resilient organizational structures, with examples from naval-air defense, counterterrorism operations and disaster response.

  4. 2023

    Discriminator guidance for diffusion

    Refining Generative Process with Discriminator Guidance in Score-based Diffusion Models, coauthored with Dongjun Kim, Yeongmin Kim, Se Jung Kwon and Wanmo Kang, is an oral at ICML 2023. A discriminator trained after score training adds a correcting term to a pre-trained diffusion model's score and reaches an FID of 1.83 on ImageNet 256x256.

  5. 2024

    Unbiased diffusion models at ICLR

    In 2024, Training Unbiased Diffusion Models From Biased Dataset proposes time-dependent importance reweighting to reduce bias in diffusion models and appears at ICLR 2024.

  6. 2025

    Bridge autoencoders and preference optimization

    In 2025, Diffusion Bridge AutoEncoders, coauthored with Yeongmin Kim, Kwanghyeon Lee, Minsang Park and Byeonghu Na, makes the diffusion endpoint depend on the latent code and is an ICLR 2025 Spotlight. For NeurIPS 2025, Preference Optimization by Estimating the Ratio of the Data Distribution, coauthored with Yeongmin Kim, HeeSun Bae and Byeonghu Na, introduces Bregman preference optimization, which includes DPO as a special case.

  7. 2026

    Language model alignment and distillation

    In 2026, Semantic-aware Wasserstein Policy Regularization for Large Language Model Alignment, accepted at ICLR 2026, proposes a semantic-aware regularization for RLHF based on the entropy-regularized Wasserstein distance. At ICLR 2026, he also coauthors Distillation of Large Language Models via Concrete Score Matching with Yeongmin Kim, Donghyeok Shin, Mina Kang and Byeonghu Na.

  8. Since 2025

    Full professor at KAIST

    In 2025 he becomes a full professor at KAIST in Daejeon, after serving as an associate professor from 2017 to 2025. As a key faculty member of KAIST's AI College, he works on generative and probabilistic models, with core competencies in generative AI and inverse design, data-centric AI, foundation models and multi-agent systems.

Career

  1. Since 2008KAISTPostdoctoral Researcher, Assistant Professor

Education

  1. 2008Carnegie Mellon UniversityPh.D.
  2. 2004Seoul National UniversityB.E.

From the biography.

  1. The United States employs an estimated 48% of people shaping AIIncludes KAIST. We estimate where the people shaping AI work today, how universities, companies and research institutes split across countries, and how country shares have changed since the 1960s.
  2. More people shaping AI studied at Massachusetts Institute of Technology than at any other institutionIncludes KAIST. We estimate where the people shaping AI studied and where their careers led. Massachusetts Institute of Technology leads with 467 former students, including 291 recorded doctorates.
  3. The company share of AI researchers’ top papers was 31% in 2020–2024 and 8% in 1970–1974Includes KAIST. We estimate where the people shaping AI worked when their most-cited papers appeared: the company share was 31% in 2020–2024 and 8% in 1970–1974, and Google led all organizations in 2020–2024.
  4. 44% of AI career moves go from one university to anotherIncludes KAIST. We estimate how 26,439 AI career moves link universities, companies, labs, nonprofits and government research centers, with decade counts, net flows and the top paths.

Colleagues

People with dated positions at the same organization in the same years.

Sources and credits

Photo: We have no published photo of Il-Chul Moon yet. Send us a free-licensed photo.

  1. Il-Chul Moon - Korea Advanced Institute of Science and Technology pure.kaist.ac.kr. Checked 22 October 2026.
  2. Woo-Seop Yun, Il-Chul Moon and Tae-Eog Lee jasss.org. Checked 22 October 2026.
  3. Modeling and Simulating Command and Control: For Organizations Under Extreme Situations - Il-Chul Moon, Kathleen M. Carley, Tag Gon Kim - Google Books books.google.com. Checked 22 October 2026.
  4. [2211.17091] Refining Generative Process with Discriminator Guidance in Score-based Diffusion Models arxiv.org. Checked 22 October 2026.
  5. ICML Oral Refining Generative Process with Discriminator Guidance in Score-based Diffusion Models icml.cc. Checked 22 October 2026.
  6. [2403.01189] Training Unbiased Diffusion Models From Biased Dataset arxiv.org. Checked 22 October 2026.
  7. [2405.17111] Diffusion Bridge AutoEncoders for Unsupervised Representation Learning arxiv.org. Checked 22 October 2026.
  8. [2505.19601] Preference Optimization by Estimating the Ratio of the Data Distribution arxiv.org. Checked 22 October 2026.
  9. [2602.01685] Semantic-aware Wasserstein Policy Regularization for Large Language Model Alignment arxiv.org. Checked 22 October 2026.
  10. Moon, Il-Chul | ML Anthology mlanthology.org. Checked 22 October 2026.
  11. Faculty | AX aicollege.kaist.ac.kr. Checked 22 October 2026.