Hongjoong Kim
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Hongjoong Kim has a total of 25 co-authors. Here are the top co-authors by number of shared publications:
- Kyoung-Sook Moon
Department of Mathematical Finance, Gachon University, Seongnam-si, Korea
10 shared publications Kyoung-Sook Moon
3 shared publicationsSungwon Park
3 shared publicationsSookyung Jun
Department of Mathematics, Korea University, 145 Anam-ro, Seongbuk-gu, Seoul 02841, Korea
2 shared publicationsYein Kwon
2 shared publications- Alexander G. Tartakovsky
Space Informatics Laboratory, Moscow Institute of Physics and Technology, Moscow, Russia
1 shared publication
Latest Publications
- 8 June 2026
- Published by Elsevier in Finance Research Letters
- Vol. 106
Cluster-based Adaptive Generation for imbalanced financial data
- 8 June 2026
- Published by Elsevier in Finance Research Letters
- Vol. 106
Cluster-based Adaptive Generation for imbalanced financial data
- 6 June 2026
- Published by Springer Nature in Computational Economics
- p. 1-22
Vocabulary Optimization Method of FinBERT for Financial Market Forecasting
- 6 June 2026
- Published by Springer Nature in Computational Economics
- p. 1-22
Vocabulary Optimization Method of FinBERT for Financial Market Forecasting
- Open Access
- 30 March 2026
- Published by Elsevier in SSRN Electronic Journal
Cluster-based Adaptive Generation for Imbalanced Financial Data
Financial institutions routinely predict rare but economically consequential events such as credit defaults and client responses. In such environments, severe class imbalance and asymmetric misclassification costs distort predictive performance and lead to suboptimal capital allocation. We propose a cluster-based adaptive generation (CAGE) framework that addresses unobserved heterogeneity within minority events. The framework partitions minority observations into homogeneous sub-populations and dynamically calibrates generative capacity according to structural complexity.Across financial settings such as risk-averse lending, cost-sensitive marketing, and income-based client targeting, we show that adaptive synthetic calibration consistently improves cost-relevant performance relative to global oversampling approaches. In credit risk prediction, CAGE increases \textit{Recall} and reduces economically costly false negatives. In marketing environments, it preserves \textit{Precision} while maintaining competitive overall balance, thereby protecting the efficiency frontier.Our findings demonstrate that aligning synthetic data generation with asymmetric economic loss functions yields measurable reductions in total expected cost. The proposed framework provides a cost-sensitive empirical strategy for financial forecasting under structural imbalance.
- 30 March 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessCluster-based Adaptive Generation for Imbalanced Financial Data
Financial institutions routinely predict rare but economically consequential events such as credit defaults and client responses. In such environments, severe class imbalance and asymmetric misclassification costs distort predictive performance and lead to suboptimal capital allocation. We propose a cluster-based adaptive generation (CAGE) framework that addresses unobserved heterogeneity within minority events. The framework partitions minority observations into homogeneous sub-populations and dynamically calibrates generative capacity according to structural complexity.Across financial settings such as risk-averse lending, cost-sensitive marketing, and income-based client targeting, we show that adaptive synthetic calibration consistently improves cost-relevant performance relative to global oversampling approaches. In credit risk prediction, CAGE increases \textit{Recall} and reduces economically costly false negatives. In marketing environments, it preserves \textit{Precision} while maintaining competitive overall balance, thereby protecting the efficiency frontier.Our findings demonstrate that aligning synthetic data generation with asymmetric economic loss functions yields measurable reductions in total expected cost. The proposed framework provides a cost-sensitive empirical strategy for financial forecasting under structural imbalance.
- Open Access
Generative Neural Operators through Diffusion Last Layer
Neural operators have emerged as a powerful paradigm for learning discretization-invariant function-to-function mappings in scientific computing. However, many practical systems are inherently stochastic, making principled uncertainty quantification essential for reliable deployment. To address this, we introduce a simple add-on, the diffusion last layer (DLL), a lightweight probabilistic head that can be attached to arbitrary neural operator backbones to model predictive uncertainty. Motivated by the relative smoothness and low-dimensional structure often exhibited by PDE solution distributions, DLL parameterizes the conditional output distribution directly in function space through a low-rank Karhunen-Loève expansion, enabling efficient and expressive uncertainty modeling. Across stochastic PDE operator learning benchmarks, DLL improves generalization and uncertainty-aware prediction. Moreover, even in deterministic long-horizon rollout settings, DLL enhances rollout stability and provides meaningful estimates of epistemic uncertainty for backbone neural operators.
Open AccessGenerative Neural Operators through Diffusion Last Layer
Neural operators have emerged as a powerful paradigm for learning discretization-invariant function-to-function mappings in scientific computing. However, many practical systems are inherently stochastic, making principled uncertainty quantification essential for reliable deployment. To address this, we introduce a simple add-on, the diffusion last layer (DLL), a lightweight probabilistic head that can be attached to arbitrary neural operator backbones to model predictive uncertainty. Motivated by the relative smoothness and low-dimensional structure often exhibited by PDE solution distributions, DLL parameterizes the conditional output distribution directly in function space through a low-rank Karhunen-Loève expansion, enabling efficient and expressive uncertainty modeling. Across stochastic PDE operator learning benchmarks, DLL improves generalization and uncertainty-aware prediction. Moreover, even in deterministic long-horizon rollout settings, DLL enhances rollout stability and provides meaningful estimates of epistemic uncertainty for backbone neural operators.
- 28 August 2025
- Published by Elsevier in Finance Research Letters
- Vol. 85
Robust deep hedging of equity-linked securities under covariance uncertainty
- 28 August 2025
- Published by Elsevier in Finance Research Letters
- Vol. 85
Robust deep hedging of equity-linked securities under covariance uncertainty
- Open Access
Adaptive Sampling Framework for Imbalanced DDoS Traffic Classification
Imbalanced data is a major challenge in network security applications, particularly in DDoS (Distributed Denial of Service) traffic classification, where detecting minority classes is critical for timely and cost-effective defense. Existing machine learning and deep learning models often fail to accurately classify such underrepresented attack types, leading to significant degradation in performance. In this study, we propose an adaptive sampling strategy that combines oversampling and undersampling techniques to address the class imbalance problem at the data level. We evaluated our approach using benchmark DDoS traffic datasets, where it demonstrated improved classification performance across key metrics, including accuracy, recall, and F1-score, compared to baseline models and conventional sampling methods. The results indicate that the proposed adaptive sampling approach improved minority class detection performance under the tested conditions, thereby improving the reliability of sensor-driven security systems. This work contributes a robust and adaptable method for imbalanced data classification, with potential applications across simulated sensor environments where anomaly detection is essential.
Open AccessAdaptive Sampling Framework for Imbalanced DDoS Traffic Classification
Imbalanced data is a major challenge in network security applications, particularly in DDoS (Distributed Denial of Service) traffic classification, where detecting minority classes is critical for timely and cost-effective defense. Existing machine learning and deep learning models often fail to accurately classify such underrepresented attack types, leading to significant degradation in performance. In this study, we propose an adaptive sampling strategy that combines oversampling and undersampling techniques to address the class imbalance problem at the data level. We evaluated our approach using benchmark DDoS traffic datasets, where it demonstrated improved classification performance across key metrics, including accuracy, recall, and F1-score, compared to baseline models and conventional sampling methods. The results indicate that the proposed adaptive sampling approach improved minority class detection performance under the tested conditions, thereby improving the reliability of sensor-driven security systems. This work contributes a robust and adaptable method for imbalanced data classification, with potential applications across simulated sensor environments where anomaly detection is essential.
- Open Access
- 5 December 2024
- Published by The Korea Institute of Military Science and Technology in Journal of the Korea Institute of Military Science and Technology
- Vol. 26 (5) , 422-430
Generating Synthetic Raman Spectra of DMMP and 2-CEES by Mathematical Transforms and Deep Generative Models
To build an automated system detecting toxic chemicals from Raman spectra, we have to obtain sufficient data of toxic chemicals. However, it usually costs high to gather Raman spectra of toxic chemicals in diverse situations. Tackling this problem, we develop methods to generate synthetic Raman spectra of DMMP and 2-CEES without actual experiments. First, we propose certain mathematical transforms to augment few original Raman spectra. Then, we train deep generative models to generate more realistic and diverse data. Analyzing synthetic Raman spectra of toxic chemicals generated by our methods through visualization, we qualitatively verify that the data are sufficiently similar to original data and diverse. For conclusion, we obtain a synthetic dataset of DMMP and 2-CEES with the proposed algorithm.
- 5 December 2024
- Published by The Korea Institute of Military Science and Technology in Journal of the Korea Institute of Military Science and Technology
- Vol. 26 (5) , 422-430
Open AccessGenerating Synthetic Raman Spectra of DMMP and 2-CEES by Mathematical Transforms and Deep Generative Models
To build an automated system detecting toxic chemicals from Raman spectra, we have to obtain sufficient data of toxic chemicals. However, it usually costs high to gather Raman spectra of toxic chemicals in diverse situations. Tackling this problem, we develop methods to generate synthetic Raman spectra of DMMP and 2-CEES without actual experiments. First, we propose certain mathematical transforms to augment few original Raman spectra. Then, we train deep generative models to generate more realistic and diverse data. Analyzing synthetic Raman spectra of toxic chemicals generated by our methods through visualization, we qualitatively verify that the data are sufficiently similar to original data and diverse. For conclusion, we obtain a synthetic dataset of DMMP and 2-CEES with the proposed algorithm.
- 23 August 2024
- Published by Elsevier in Chemometrics and Intelligent Laboratory Systems
- Vol. 253
Robust baseline correction for Raman spectra by constrained Gaussian radial basis function fitting
- 23 August 2024
- Published by Elsevier in Chemometrics and Intelligent Laboratory Systems
- Vol. 253
Robust baseline correction for Raman spectra by constrained Gaussian radial basis function fitting
- Open Access
Adaptive Data Selection-Based Machine Learning Algorithm for Prediction of Component Obsolescence
Product obsolescence occurs in the manufacturing industry as new products with better performance or improved cost-effectiveness are developed. A proactive strategy for predicting component obsolescence can reduce manufacturing losses and lead to customer satisfaction. In this study, we propose a machine learning algorithm for a proactive strategy based on an adaptive data selection method to forecast the obsolescence of electronic diodes. Typical machine learning algorithms construct a single model for a dataset. By contrast, the proposed algorithm first determines a mathematical cover of the dataset via unsupervised clustering and subsequently constructs multiple models, each of which is trained with the data in one cover. For each data point in the test dataset, an optimal model is selected for regression. Results of empirical experiments show that the proposed method improves the obsolescence prediction accuracy and accelerates the training procedure. A novelty of this study is that it demonstrates the effectiveness of unsupervised clustering methods for improving supervised regression algorithms.
Open AccessAdaptive Data Selection-Based Machine Learning Algorithm for Prediction of Component Obsolescence
Product obsolescence occurs in the manufacturing industry as new products with better performance or improved cost-effectiveness are developed. A proactive strategy for predicting component obsolescence can reduce manufacturing losses and lead to customer satisfaction. In this study, we propose a machine learning algorithm for a proactive strategy based on an adaptive data selection method to forecast the obsolescence of electronic diodes. Typical machine learning algorithms construct a single model for a dataset. By contrast, the proposed algorithm first determines a mathematical cover of the dataset via unsupervised clustering and subsequently constructs multiple models, each of which is trained with the data in one cover. For each data point in the test dataset, an optimal model is selected for regression. Results of empirical experiments show that the proposed method improves the obsolescence prediction accuracy and accelerates the training procedure. A novelty of this study is that it demonstrates the effectiveness of unsupervised clustering methods for improving supervised regression algorithms.
- Open Access
Forecasting Obsolescence of Components by Using a Clustering-Based Hybrid Machine-Learning Algorithm
Product obsolescence occurs in every production line in the industry as better-performance or cost-effective products become available. A proactive strategy for obsolescence allows firms to prepare for such events and reduces the manufacturing loss, which eventually leads to positive customer satisfaction. We propose a machine learning-based algorithm to forecast the obsolescence date of electronic diodes, which has a limitation on the amount of data available. The proposed algorithm overcomes these limitations in two ways. First, an unsupervised clustering algorithm is applied to group the data based on their similarity and build independent machine-learning models specialized for each group. Second, a hybrid method including several reliable techniques is constructed to improve the prediction accuracy and overcome the limitation of the lack of data. It is empirically confirmed that the prediction accuracy of the obsolescence date for the electrical component data is improved through the proposed clustering-based hybrid method.
Open AccessForecasting Obsolescence of Components by Using a Clustering-Based Hybrid Machine-Learning Algorithm
Product obsolescence occurs in every production line in the industry as better-performance or cost-effective products become available. A proactive strategy for obsolescence allows firms to prepare for such events and reduces the manufacturing loss, which eventually leads to positive customer satisfaction. We propose a machine learning-based algorithm to forecast the obsolescence date of electronic diodes, which has a limitation on the amount of data available. The proposed algorithm overcomes these limitations in two ways. First, an unsupervised clustering algorithm is applied to group the data based on their similarity and build independent machine-learning models specialized for each group. Second, a hybrid method including several reliable techniques is constructed to improve the prediction accuracy and overcome the limitation of the lack of data. It is empirically confirmed that the prediction accuracy of the obsolescence date for the electrical component data is improved through the proposed clustering-based hybrid method.