Hello, I’m Dong-Hee

👋 About Me

Hello, I'm Dong-Hee Shin, an Assistant Professor in the Department of Software at Yonsei University Mirae Campus.
My main research interests are in AI for Science, enhancing the efficiency and scalability of scientific discovery.
I am especially interested in applications such as drug discovery, material science, and semiconductor design.
On the AI side, I work on reinforcement learning and robotics algorithms to support scientific innovation.

Current topics of interest include:

• Physical AI and optimization algorithms for autonomous self-driving chemistry labs.
• Reinforcement learning for molecular engineering and semiconductor chip design.
• AI-driven discovery and design of molecular systems and advanced materials.

📚 Selected Publications

Optical Diffraction-based Convolution for Semiconductor Lithography

Young-Han Son, Dong-Hee Shin, Deok-Joong Lee, Hyun Jung Lee, Tae-Eui Kam

Conference on Computer Vision and Pattern Recognition (CVPR), 2026

Acceptance rate 25.4% | 4,090 of 16,092 submissions

• We introduce a physics-inspired neural network for semiconductor lithography that explicitly incorporates optical diffraction to model phase variations arising from light propagation, enabling accurate representation of the underlying physics.

Treatment Stitching with Schrödinger Bridge for Enhancing Offline Reinforcement Learning in Adaptive Treatment Strategies

Dong-Hee Shin, Deok-Joong Lee, Young-Han Son, Tae-Eui Kam

Association for the Advancement of Artificial Intelligence (AAAI), 2026

Oral presentation

• We introduce a novel trajectory augmentation framework for offline reinforcement learning in adaptive treatment strategies, leveraging the Schrödinger Bridge (SB) to generate smooth and energy-efficient transitions that address sparsity in EHR data.

Offline Model-based Optimization for Real-World Molecular Discovery

Dong-Hee Shin, Young-Han Son, Hyun Jung Lee, Deok-Joong Lee, Tae-Eui Kam

International Conference on Machine Learning (ICML), 2025

Acceptance rate 26.9% | 3,260 of 12,107 submissions

• We introduce a novel framework that stitches molecules from an offline dataset to generate new samples for fine-tuning the generative model even in offline settings.

Sharpness-Aware Minimization with Physics-Informed Regularizations for Predicting Semiconductor Material Properties in Molecular Dynamics

Dong-Hee Shin, Young-Han Son, Tae-Eui Kam

Chemometrics and Intelligent Laboratory Systems (CHEMOLAB), 2025

JCR Top 3.9% (IF: 3.8)

• We introduce a novel framework that incorporates physics-informed regularizations with sharpness-aware minimization for semiconductor material property prediction.

Population-based Evolutionary Search for Joint Hyperparameter and Architecture Optimization in Brain-Computer Interface

Dong-Hee Shin, Deok-Joong Lee, Ji-Wung Han, Young-Han Son, Tae-Eui Kam

Expert Systems with Applications (ESWA), 2025

JCR Top 6.1% (IF: 7.5)

• We propose a joint optimization framework that leverages a population-based evolutionary search to optimize both hyperparameters and architectures in BCI.

Dynamic Many-Objective Molecular Optimization: Unfolding Complexity with Objective Decomposition and Progressive Optimization

Dong-Hee Shin, Young-Han Son, Deok-Joong Lee, Ji-Wung Han, Tae-Eui Kam

International Joint Conference on Artificial Intelligence (IJCAI), 2024

Lead oral presentation (12 min) | Top 2.2% (128 of 5,651 submissions)

• We propose a method for tackling dynamic many-objective molecular optimization problem by utilizing (1) objective decomposition and (2) progressive optimization.
  (1) Objective Decomposition: our method decomposes many-objective sets into more manageable sub-problems facilitated by our decomposition module.
  (2) Progressive Optimization: The optimization process begins with a single objective, then progressively adds subsequent objective in decomposition order.

Sparse Graph Representation Learning Based on Reinforcement Learning for Personalized Mild Cognitive Impairment (MCI) Diagnosis

Chang-Hoon Ji, Dong-Hee Shin, Young-Han Son, Tae-Eui Kam

IEEE Journal of Biomedical and Health Informatics (JBHI), 2024

JCR Top 3.7% (IF: 6.8)

FTMMR: Fusion Transformer for Integrating Multiple Molecular Representations

Young-Han Son, Dong-Hee Shin, Tae-Eui Kam

IEEE Journal of Biomedical and Health Informatics (JBHI), 2024

JCR Top 3.7% (IF: 6.8)

MARS: multiagent reinforcement learning for spatial—spectral and temporal feature selection in EEG-based BCI

Dong-Hee Shin, Young-Han Son, Jun-Mo Kim, Hee-Jun Ahn, Jun-Ho Seo, Chang-Hoon Ji, Ji-Wung Han, Byung-Jun Lee, Dong-Ok Won, Tae-Eui Kam

IEEE Transactions on Systems, Man, and Cybernetics: Systems (TSMC), 2024

JCR Top 6.2% (IF: 8.7)

• We propose a cooperative multi-agent reinforcement learning (MARL) algorithm that performs feature selection in both spatial-spectral and temporal domains simultaneously for a motor imagery (MI)-EEG classification task.

🏆 Grants & Awards

• Ph.D. Fellowship Awardee, National Research Foundation of Korea (2024)
  Description: awarded new research project funding through a government-supported doctoral research grant administered by the National Research Foundation of Korea (NRF)
• Korea University Research Challenger Program: RCP (2022)
  Description: selected for interdisciplinary research funding as part of Korea University’s RCP initiative to support collaborative and innovative graduate research projects
• Graduate Research Fellowship (BK21 FOUR), National Research Foundation of Korea (2021)
  Description: awarded competitive graduate research fellowship under the BK21 FOUR program, funded by the NRF to support advanced academic training through structured research

🏫 Teaching & Mentoring

• Robot Mechatronics – Professor, Spring 2026
  Yonsei University Mirae Campus
• Introduction to Generative Model – Instructor, Feb 2023
  Tech University of Korea
• Introduction to Artificial Intelligence – Tutor, 2021, 2022, 2023
  LG CNS
• Reinforcement Learning – Teaching Assistant, Fall 2022
  Korea University

🎤 Talks & Presentations

• Offline Reinforcment Learning for Optimizing Treatment Strategies
  AAAI, 2026 – Singapore
• Cepstrum Analysis for Molecular Structure Elucidation from Mass Spectra
  ICEIC, 2026 – Macau
• Machine Learning Force Fields for Semiconductor Design
  ISOCC, 2025 – Busan, Korea
• Designing Selective Drugs for Mitigating Off-Target Effects
  ICEIC, 2025 – Osaka, Japan
• Dynamic Many-Objective Molecular Optimization
  IJCAI, 2024 – Jeju, Korea
• Progressive Optimization for Molecular Discovery
  ICLR Workshop on Generative and Experimental Perspectives for Biomolecular Design, 2024 – Vienna, Austria
• Reinforcement Learning for Brain-Computer Interface
  International Winter Conference on Brain-Computer Interface, 2022 – Gangwon, Korea