Donghee Lee
Graduate Student
Donghee Lee
Graduate Student
Nature was never designed to be easy for humans to understand
Therefore, throughout history, physicists have sought mathematical lenses through which to interpret nature effectively
Even phrases that emphasize AI’s success have started to feel clichéd
Academic Career
Study Interest – Formulating a physics-based theoretical framework to investigate the fundamental properties of deep learning
Current position - Department of Physics in Korea Advanced Institute of Science and Technology
02/2023 - current Integrated master's/doctoral course in KAIST, Particle Theory Group
02/2018 - 02/2023 B.S in Physics at KAIST (Second major in Mathematical science)
Research Interests
• Physics of deep learning: field-theoretic approaches and collective phenomena
• Inherent properties of training dynamics: locality and homogeneity
• Beyond scaling laws in deep learning: efficient modeling strategies
Publications
GlueNN: gluing patchwise analytic solutions with neural networks (arXiv:2601.05889)
Bulk Boundary Decomposition of neural networks (arXiv:2511.02003)
Synaptic Field Theory for Neural Networks - Phys. Rev. D (arXiv:2503.08827)
Dynamic neuron approach to deep neural networks - Phys. Rev. Research (arXiv:2410.00396)
Conference
Presentation files can be found here
(oral) SigmaPhi 7/06/2026 - 7/10/2026
- Bulk-boundary decomposition of neural networks
(oral) SigmaPhi 7/06/2026 - 7/10/2026
- Dynamic neuron approach to deep neural networks
(oral) KPS spring meeting 4/22/2026 - 4/24/2026
- Bulk-boundary decomposition of neural networks
(poster) StatPhys29 7/12/2025 - 7/19/2025
- Dynamic Neuron Approach
(poster) KPS meeting - Daejeon, Chungnam, and Sejong Branch 7/24/2025
- Dynamic Neuron Approach
(oral) KPS spring meeting 4/23/2025 - 4/25/2025
- Dynamic neurons: A statistical physics approach for analyzing deep neural networks
Seminar
Presented lab internal seminar : presented on 01/18-01/19 and 02/15-02/16 in 2023
Review article : Ising model and Its Critical Phenomena
Teaching Assistance
Quantum Information with Atoms 2026 spring
Classical Mechanics 2 2024 fall
Quantum Mechanics 2 2023 fall Homework & Exam Solutions
General Physics 1 2023 spring
Academic Activity - Attended Conference
Symmetries in Quantum Field Theory and Particle Physics 6/30/2025 - 7/11/2025
Attended Light Dark World 2024 12/08/2024 - 15/08/2024
Attended KPS spring meeting 4/24/2024 - 4/26/2024
Attended KPS fall meeting 10/25/2023 - 10/27/2023
Academic Activity - Study Group & Journal Club
AI+HEP journal club 05/2025 -present
Study group of free topics in physics 05/20/2024 - present
Cosmology study group 09/21/2023 - 11/9/2023
Quantum Field Theory study group 03/05/2025 - present
Lead Particle Physics study group 12/22/2023 - 05/10/2024
Quantum Field Theory study group 01/18/2023 - 08/07/2024
Academic Activity - Seninar and Presentations
Attended in Graduate School of AI for Math Opening Workshop at KAIST on 03/06 2026
Attended in invited talk about our researches at SNU on 02/26 2026
Attended in invited talk about our researches at KAIST on 02/25 2026
Feature learning from realistic inputs: from toy models to deep neural networks (SISSA) on 11/25 2025
Attended in invited talk about Synaptic Field Theory at KAIST on 10/15 2025
Attended in Commentary Lectures on Nobel Prize Winning Achievement by College of Natural Sciences at KAIST on 11/13 2024
Presented in Quantum Gravity Class : presented on 05/01 2024
Review article : Viscosity in Strongly Interacting Quanntum Field Theories from Black Hole Physics
Contact
E-mail : dhlee641@kaist.ac.kr
Group: Particle Theory Group (Room 4320 in E6-2, +82-42-350-7357)