Dong Hoon Lee

donghoonlee [at] kaist.ac.kr

I am interested in expanding the applicability of AI/ML models to resource-constrained settings. My Ph.D. work has focused on (1) data efficiency—enabling learning with minimal labeled data through few-shot and self-supervised learning—and (2) computational efficiency—reducing inference and training costs through efficient transformer architectures. Recently, I am extending these interests to generative models, exploring efficient inference and training of diffusion transformers.

Google Scholar / Github / CV

Publications

Variable-Length Tokenization via Learnable Global Merging for Diffusion Transformers
Dong Hoon Lee and Seunghoon Hong
Proceedings of the 43rd International Conference on Machine Learning (ICML), 2026.
pdf / code

Disentangled Representation Learning via Modular Compositional Bias
Whie Jung, Dong Hoon Lee, Seunghoon Hong
Advances in Neural Information Processing Systems (NeurIPS), 2025.
pdf / code

Universal Few-shot Spatial Control for Diffusion Models
Kiet T Nguyen, Chanhyuk Lee, Donggyun Kim, Dong Hoon Lee, Seunghoon Hong
Advances in Neural Information Processing Systems (NeurIPS), 2025.
pdf / code

Learning to Merge Tokens via Decoupled Embedding for Efficient Vision Transformers
Dong Hoon Lee and Seunghoon Hong
Advances in Neural Information Processing Systems (NeurIPS), 2024.
pdf / code

Unsupervised Visual Representation Learning via Mutual Information Regularized Assignment
Dong Hoon Lee, Sungik Choi, Hyunwoo Kim and Sae-Young Chung
Advances in Neural Information Processing Systems (NeurIPS), 2022.
pdf / code

Unsupervised Embedding Adaptation via Early-Stage Feature Reconstruction for Few-Shot Classification
Dong Hoon Lee and Sae-Young Chung
Proceedings of the 38th International Conference on Machine Learning (ICML), 2021.
pdf / code

Education

2026: Ph.D. in Kim Jaechul Graduate School of AI, KAIST, Daejeon, Korea

2018: M.S. in Electrical Engineering, KAIST, Daejeon, Korea

2016: B.S. in Electrical Engineering, KAIST, Daejeon, Korea

2012: Korea Science Academy, Busan, Korea

Experience

Research Intern, LG AI Research, Seoul, Korea, 2022

Awards

NeurIPS 2022 Scholar Award, 2022

Qualcomm Innovation Fellowship, 2021 South Korea Finalist, 2021

Korea Government Fellowship, March 2021 to present

Teaching

2019 fall: TA, EE807 Special Topics in Electrical Engineering. Deep Reinforcement Learning and AlphaGo, KAIST.

2019 spring: TA, EE405 Electronics Design Lab. Network of Smart Things, KAIST.

2018 fall: TA, EE405 Electronics Design Lab. Robocam, KAIST.

2018 spring: TA, EE807 Special Topics in Electrical Engineering. Mathematical Foundation of Reinforcement Learning, KAIST.

2017 spring: TA, EE210 Probability and Introductory Random Processes, KAIST.

Website template from here.