Jonghyun Chae

Dr. Jonghyun Chae

orcid

Research Focus & Metrics

Academic Career Length

3Years (2024-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
26 August 2025 in Biosensors and Bioelectronics
Verified
Journal Article
Closed access
SCSunah Cheong
Sunah Cheong
JCJonghyun Chae
Jonghyun Chae
HLHwiwon Lee
Hwiwon Lee
SHSang Hoon Hong
Sang Hoon Hong
HLHyunmin Lim
Hyunmin Lim
Sei Kwang HahnSei Kwang Hahn
Sei Kwang Hahn

14 September 2024 in Fluids
Verified
Journal Article
Open access
PDF file
IKInnyoung Kim
Innyoung Kim
YJYoungmin Jeon
Youngmin Jeon
JCJonghyun Chae
Jonghyun Chae
Donghyun YouDonghyun You
Donghyun You
A comprehensive review of recent advancements in applying deep reinforcement learning (DRL) to fluid dynamics problems is presented. Applications in flow control and shape optimization, the primary fields where DRL is currently utilized, are thoroughly examined. Moreover, the review introduces emerging research trends in automation within computational fluid dynamics, a promising field for enhancing the efficiency and reliability of numerical analysis. Emphasis is placed on strategies developed to overcome challenges in applying DRL to complex, real-world engineering problems, such as data efficiency, turbulence, and partial observability. Specifically, the implementations of transfer learning, multi-agent reinforcement learning, and the partially observable Markov decision process are discussed, illustrating how these techniques can provide solutions to such issues. Finally, future research directions that could further advance the integration of DRL in fluid dynamics research are highlighted.

7 February 2026 in SSRN Electronic Journal
Verified
Preprint
Open access
JCJonghyun Chae
Jonghyun Chae
Donghyun YouDonghyun You
Donghyun You
A deep neural network (DNN) model is developed to predict the spatio-temporal distribution of PM\textsubscript{2.5} and PM\textsubscript{10} concentrations across East Asia. The proposed model combines convolutional neural networks, long short-term memory units, and discrete wavelet transforms to enhance multi-resolution spatio-temporal representation of PM fields. Unlike station-dependent approaches that rely on point-based observations, the framework is trained and evaluated using spatially continuous reanalysis fields, enabling domain-wide prediction on a consistent grid. A chemically informed input design is adopted by jointly integrating meteorological variables, chemically reactive species, and anthropogenic emissions to better represent PM variability under conditions in which secondary formation is expected to be important. To quantify the relative contribution of different predictor groups and to examine performance across pollution conditions, systematic ablation studies are conducted using three input configurations: (1) PM concentrations alone, (2) PM combined with meteorological data, and (3) a fully integrated set including chemical species and emissions. The fully integrated configuration yields the most consistent improvements, particularly for high-concentration episodes, indicating that chemical and emission information provides additional predictive value beyond meteorology-only inputs and improves peak prediction skill.