Generative Adversarial Networks for Fire Data Augmentation
Kwangsue Chung, Hyungtak Lee, Seongju Kang, and Jaegi Hwang. "Generative Adversarial Networks for Fire Data Augmentation." Korean Patent KR 10-2931826, registered Feb. 2026.
CV
Education
- Ph.D. in Computer Science, Iowa State University, Ames, IA, Aug. 2023 – May 2029 (expected)
- Advisor: Christopher J. Quinn
- GPA: 3.96/4.0
- B.S. in Computer Engineering, Kwangwoon University, Seoul, Korea, Mar. 2017 – Feb. 2023
- GPA: 4.29/4.5 (3.81/4.0), Dean’s List (3 semesters)
Research experience
- Ph.D. Researcher, Iowa State University, Aug. 2024 – Present
- Developing approximate inference methods for preferential Gaussian processes, which learn latent utility functions from human choice feedback, with applications to preference-based Bayesian optimization
- Reimplemented and extended preferential Bayesian optimization methods in GPyTorch/PyTorch, including a migration of prior GPy/NumPy code
- Research Project: Uncertainty Quantification for LLMs, 2026 – Present
- Developing methods for uncertainty quantification–based fine-tuning for large language models; co-first author on the resulting manuscript
- Undergraduate Research Assistant, Computer Communications Lab, Kwangwoon University, Jul. 2021 – Aug. 2023
- Proposed a GAN architecture that generates training images together with bounding-box labels to augment scarce object-detection datasets (fire detection); first-author journal paper and registered Korean patent
- Contributed to government-funded projects on generative data augmentation and lightweight edge object detection
Publications
Uncertainty Quantification-Based Fine-Tuning for Large Language Models
Minoo Hong*, Hyungtak Lee*, et al. (2026). "Uncertainty Quantification-Based Fine-Tuning for Large Language Models." Manuscript in preparation. (*equal contribution)
Robust Data Augmentation Generative Adversarial Network for Object Detection
Hyungtak Lee, Seongju Kang, and Kwangsue Chung. (2023). "Robust Data Augmentation Generative Adversarial Network for Object Detection." Sensors, 23(1), 157. https://doi.org/10.3390/s23010157
Object Detection with Dataset Augmentation for Fire Images Based on GAN
Hyungtak Lee, Seongju Kang, and Kwangsue Chung. (2022). "Object Detection with Dataset Augmentation for Fire Images Based on GAN." 2022 13th International Conference on Information and Communication Technology Convergence (ICTC), pp. 2118-2123.
Object Detection Dataset Generation Framework using GAN
Hyungtak Lee, Seongju Kang, Jaegi Hwang, and Kwangsue Chung. (2022). "Object Detection Dataset Generation Framework using GAN." Korea Computer Congress (KCC).
IoT Collaboration System Based on Edge Computing for Smart Livestock System
Chi-Hyun Ahn, Hyungtak Lee, and Kwangsue Chung. (2022). "IoT Collaboration System Based on Edge Computing for Smart Livestock System." Journal of the Korea Institute of Information and Communication Engineering.
Teaching
Other experience
- Korean Military Service (KATUSA), assigned to the Eighth U.S. Army, Dec. 2019 – Jul. 2021
- Mandatory service as a network specialist in a U.S. Army signal brigade; awarded the Army Commendation Medal
Skills
- Methods: Bayesian Optimization, Gaussian Processes, Preference Learning, Bandits, Uncertainty Quantification, PEFT
- Languages: Python, C/C++, Rust, ARMv8 Assembly
- Frameworks & Tools: PyTorch, BoTorch, GPyTorch, NumPy, CUDA, Git, Linux, LaTeX