Jun-Ha Park

Mr. Jun-Ha Park

orcid

Research Focus & Metrics

Academic Career Length

6Years (2021-Present)

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

Representative Publications

View All Publications
27 September 2021
Verified
Conference Paper
Open access
PWPeng Wu
Peng Wu
TITales Cesar de Oliveira Imbiriba
Tales Cesar de Oliveira Imbiriba
JPJun-Ha Park
Jun-Ha Park
SKSunwoo Kim
Sunwoo Kim
Pau ClosasPau Closas
Pau Closas
Localization and tracking of objects using data-driven methods is a popular topic due to the complexity in characterizing the physics of wireless channel propagation models. In these modeling approaches, data needs to be gathered to accurately train models, at the same time that user’s privacy is maintained. An appealing scheme to cooperatively achieve these goals is known as Federated Learning (FL). A challenge in FL schemes is the presence of non-independent and identically distributed (non-IID) data, caused by unevenly exploration of different areas. In this paper, we consider the use of recent FL schemes to train a set of personalized models that are then optimally fused through Bayesian rules, which makes it appropriate in the context of indoor localization.

2 May 2025 in Sensors
Verified
Journal Article
Open access
PDF file
YHYoung Jae Huh
Young Jae Huh
JPJun-Ha Park
Jun-Ha Park
YKYoung Jae Kim
Young Jae Kim
KKKwang Gi Kim
Kwang Gi Kim
Ensemble learning (EL), a machine learning technique that combines the results of multiple learning algorithms to obtain predicted values, aims to achieve better predictive performance than a single learning algorithm alone. Machine learning techniques, including EL, have been applied in the field of medicine to assist in the clinical interpretation of specific diseases. Although neurodegenerative diseases, especially Alzheimer’s disease (AD), are of interest to clinicians and researchers due to their rapid increase in clinical cases, the application of EL in AD diagnosis has been relatively less attempted. In this research, we demonstrate that three machine learning algorithms, trained on an ensemble of electroencephalogram (EEG) and clock drawing test (CDT) feature data for an AD classification task, show improved AD detection accuracy compared to when either the EEG or CDT dataset is used independently. We also explore which feature contributes most to decision-making in AD and healthy control (HC) classification. In conclusion, the current study suggests that EL can be a novel clinical application of machine learning (ML) in the automated AD screening process.

8 July 2025 in Sensors
Verified
Journal Article
Open access
PDF file
KYKicheol Yoon
Kicheol Yoon
SLSangyun Lee
Sangyun Lee
JPJun-Ha Park
Jun-Ha Park
KKKwang Gi Kim
Kwang Gi Kim
This paper proposes a drug classification system using convolutional neural network (CNN) training and rotational pill dropping technology. Images of 40 pills for each of 102 types (total 4080 images) were captured, achieving a CNN classification accuracy of 88.8%. The system uses a bowl feeder with optimized operating parameters—voltage, torque, PWM, tilt angle, vibration amplitude (0.2–1.5 mm), and frequency (4–40 Hz)—to ensure stable, sequential pill movement without loss or clumping. Performance tests were conducted at 5 V, 20 rpm, 20% PWM (@40 Hz), and 1.5 mm vibration amplitude. The bowl feeder structure tolerates oblique angles up to 75°, enabling precise pill alignment and classification. The CNN model plays a key role in accurate pill detection and classification.

24 March 2024 in Proceedings of the AAAI Conference on Artificial Intelligence
Verified
Journal Article
Open access
PDF file
JKJuyeop Kim
Juyeop Kim
JPJun-Ha Park
Jun-Ha Park
SKSongkuk Kim
Songkuk Kim
JLJong-Seok Lee
Jong-Seok Lee
Neural networks with self-attention (a.k.a. Transformers) like ViT and Swin have emerged as a better alternative to traditional convolutional neural networks (CNNs). However, our understanding of how the new architecture works is still limited. In this paper, we focus on the phenomenon that Transformers show higher robustness against corruptions than CNNs, while not being overconfident. This is contrary to the intuition that robustness increases with confidence. We resolve this contradiction by empirically investigating how the output of the penultimate layer moves in the representation space as the input data moves linearly within a small area. In particular, we show the following. (1) While CNNs exhibit fairly linear relationship between the input and output movements, Transformers show nonlinear relationship for some data. For those data, the output of Transformers moves in a curved trajectory as the input moves linearly. (2) When a data is located in a curved region, it is hard to move it out of the decision region since the output moves along a curved trajectory instead of a straight line to the decision boundary, resulting in high robustness of Transformers. (3) If a data is slightly modified to jump out of the curved region, the movements afterwards become linear and the output goes to the decision boundary directly. In other words, there does exist a decision boundary near the data, which is hard to find only because of the curved representation space. This explains the underconfident prediction of Transformers. Also, we examine mathematical properties of the attention operation that induce nonlinear response to linear perturbation. Finally, we share our additional findings, regarding what contributes to the curved representation space of Transformers, and how the curvedness evolves during training.

30 May 2024 in Journal of Innovative Medical Technology
Verified
Journal Article
Open access
PDF file
JPJun-Ha Park
Jun-Ha Park
YKYoung Jae Kim
Young Jae Kim
KKKwang Gi Kim
Kwang Gi Kim
Jun-Ha Park, Young Jae Kim, Kwang Gi Kim. J Innov Med Technol 2024;2:11-9. https://doi.org/10.61940/jimt.240002

1 October 2023
Verified
Conference Paper
Closed access
SJSoobeom Jang
Soobeom Jang
JPJun-Ha Park
Jun-Ha Park
JLJong-Seok Lee
Jong-Seok Lee
Recently, graph neural networks (GNNs) have become a popular approach to deal with machine learning tasks for graph-structured data. To achieve reliable performance with a GNN-based approach, obtaining high-quality graph structures is crucial. However, the graph data in the real-world often contain noise from data themselves or during the collecting procedure, which leads to the performance degradation of GNNs. In this paper, we propose a novel approach to enhance graph structures for performance improvement of GNNs by reversely applying the concept of adversarial attacks on graph data. Experimental results demonstrate the effectiveness of our method in improving performance of GNNs. Furthermore, we investigate the changes in the graph structure induced by our method, taking into account the connectivity of both interclass and intra-class edges and measuring the extent of over-smoothing.