연구성과로 돌아가기
2025 연구성과 (99 / 151)
※ 컨트롤 + 클릭으로 열별 다중 정렬 가능합니다.
Excel 다운로드
| WoS | SCOPUS | Document Type | Document Title | Abstract | Authors | Affiliation | ResearcherID (WoS) | AuthorsID (SCOPUS) | Author Email(s) | Journal Name | JCR Abbreviation | ISSN | eISSN | Volume | Issue | WoS Edition | WoS Category | JCR Year | IF | JCR (%) | FWCI | FWCI Update Date | WoS Citation | SCOPUS Citation | Keywords (WoS) | KeywordsPlus (WoS) | Keywords (SCOPUS) | KeywordsPlus (SCOPUS) | Language | Publication Stage | Publication Year | Publication Date | DOI | JCR Link | DOI Link | WOS Link | SCOPUS Link |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| ○ | ○ | Article | Integrated Spatiotemporal Hybrid Solar PV Generation Forecast Between Countries on Different Continents Using Transfer Learning Method | Solar photovoltaic (PV) generation is a cornerstone of sustainable energy production, but predicting its capacity across countries remains challenging due to factors like climate, terrain, and population density. To address this, a recent study proposed a novel approach using transfer learning, which is particularly valuable when historical data for newly established PV plants is limited. The study evaluated four PV plants in South Korea and Germany, selected for their diverse geographical and climatic conditions. The proposed CL-Transformer model outperformed established machine learning models such as LSTM, CNN-LSTM, and Transformer, consistently demonstrating superior predictive capabilities. Notably, when trained on Korean data and applied to both South Korea and Germany, the model achieved an average R (2) (adj) improvement of 23.5 %. When trained on German data, the improvement was even more pronounced at 67.3 %. Additionally, transfer learning experiments revealed up to a 50.6 % enhancement in R (2) (adj) across different plant scales. By integrating external weather variables and satellite data, this hybrid model provides valuable insights for accurate capacity prediction and strategic planning in deploying new PV plants, contributing to greater stability and efficiency in the power industry. | Kim, Bowoo; Belkilani, Kaouther; Heilscher, Gerd; Otto, Marc-Oliver; Huh, Jeung-Soo; Suh, Dongjun | Kyungpook Natl Univ, Dept Convergence & Fus Syst Engn, Sangju 37224, South Korea; Ulm Univ Appl Sci, Smart Grids Res Grp, D-89081 Ulm, Germany; Ulm Univ Appl Sci, Dept Math Nat & Econ Sci, D-89075 Ulm, Germany | 57219947521; 57200367618; 6506550047; 57217105369; 7102258915; 36613529600 | dongjunsuh@knu.ac.kr; | IEEE ACCESS | IEEE ACCESS | 2169-3536 | 13 | SCIE | ENGINEERING, ELECTRICAL & ELECTRONIC;TELECOMMUNICATIONS;COMPUTER SCIENCE, INFORMATION SYSTEMS | 2024 | 3.6 | 34.8 | 0 | 2025-05-07 | 0 | 0 | Predictive models; Meteorology; Transfer learning; Data models; Solar power generation; Accuracy; Forecasting; Autoregressive processes; Satellite images; Transformers; Geostationary satellite; photovoltaics; region of interest extraction; spatiotemporal; transfer learning | POWER-GENERATION; PREDICTION; NETWORKS; MODELS | Geostationary satellite; photovoltaics; region of interest extraction; spatiotemporal; transfer learning | Interest extractions; PhotoVoltaic plant; Photovoltaics; Region of interest extraction; Region-of-interest; Regions of interest; Solar photovoltaic generations; South Korea; Spatiotemporal; Transfer learning | English | 2025 | 2025 | 10.1109/access.2024.3514098 | 바로가기 | 바로가기 | 바로가기 | 바로가기 | ||||
| ○ | ○ | Article | Integrating Pull Request Comment Analysis and Developer Profiles for Expertise-Based Recommendations in Global Software Development | Determining a suitable software developer to match project needs within the Global software development (GSD) context requires detailed information. The complexity of this problem arises from the required combination of the developer's level of technical expertise, domain knowledge, and the extent to which they possess the collaborative skills necessary for a successful project. Typical developer recommendation systems do not consider the dynamics of expertise and cooperative nature of the tasks for assessing their correctness, often restricting themselves to extracting review comments only to measure their usefulness and suggest reviewers. This research intends to create a recommendation system using pull request review comments and selected data from developers' profiles to recommend better experts based on their dynamic expertise. Using advanced algorithm techniques, the proposed model Global Developer Expertise Recommendation System (GDERS) aims to improve the quality of captured data and substantially increase the accuracy of developer recommendations. Impressively, the proposed model significantly outperformed all other text-based classifiers TextCNN, TextRCNN, and Bilstm in this study, showing an accuracy of 91.85%. This research provides a significant achievement of recommendation systems in the global software development context that support more effective collaboration and increase the probability of project completion on time by allowing project managers to find easily accessible developers in the field with the right expertise. | Zamir, Sara; Rehman, Abdul; Mohsin, Hufsa; Zamir, Elif; Abbas, Assad; Al-Yarimi, Fuad A. M. | COMSATS Univ Islamabad, Dept Comp Sci, Islamabad 45550, Pakistan; Kyungpook Natl Univ, Sch Comp Sci & Engn, Daegu 41566, South Korea; Shifa Tameer e Millat Univ, Dept Comp, Islamabad, Pakistan; King Khalid Univ, Dept Comp Sci, Abha 61421, Saudi Arabia | Rehman, Abdul/D-5630-2019; Al Yarimi, Fuad/CAA-2602-2022 | 59534368400; 57200894071; 55364463800; 59534018400; 56349574400; 57203456925 | a.rehman.knu@knu.ac.kr; | IEEE ACCESS | IEEE ACCESS | 2169-3536 | 13 | SCIE | ENGINEERING, ELECTRICAL & ELECTRONIC;TELECOMMUNICATIONS;COMPUTER SCIENCE, INFORMATION SYSTEMS | 2024 | 3.6 | 34.8 | 0 | 2025-05-07 | 0 | 0 | Reviews; Codes; Software development management; Recommender systems; Software; Collaboration; Social networking (online); History; Analytical models; Accuracy; Expertise recommendation; global software development; pull request reviews; comments classification | comments classification; Expertise recommendation; global software development; pull request reviews | Comment analysis; Comment classification; Developer recommendations; Domain knowledge; Expertise recommendations; Global software development; Project completion; Pull request review; Software developer; Technical expertise; Data accuracy | English | 2025 | 2025 | 10.1109/access.2025.3532386 | 바로가기 | 바로가기 | 바로가기 | 바로가기 | ||||
| ○ | ○ | Article | Location-Aware Ambient Sound Model Adaptation for On-Device Human Activity Recognition in Living Spaces | Human activity recognition (HAR) using sound-based approaches provides a non-intrusive and practical solution for detecting and classifying human activities in real-world environments. However, deploying HAR systems across diverse environments presents challenges due to domain shift and variations in environmental acoustics. Traditional HAR models trained on pre-collected datasets often fail to generalize to new settings, leading to performance degradation and misclassification errors. This paper proposes an adaptive ambient sound model optimization system to enhance the robustness and adaptability of HAR in edge computing environments. The system can dynamically customize classification labels based on the installation environment, ensuring flexibility across different locations. To mitigate domain shift, edge-based transfer learning fine-tunes a pre-trained model using locally collected data, improving classification accuracy across varying acoustic conditions. Additionally, a pseudo-labeling mechanism continuously optimizes the model by iteratively refining predictions from high-confidence unlabeled data, enabling long term adaptability without extensive manual annotations. To validate the proposed approach, we implemented the system and conducted experiments using real-world sound data from multiple residential environments. Experimental results demonstrate that the system effectively adapts to different locations while maintaining high classification accuracy, real-time inference, and a short training time, making it suitable for practical edge deployment. | Lee, Cheolhwan; Kang, Homin; Ju Kang, Soon | Kyungpook Natl Univ, Sch Elect & Elect Engn, Daegu 41566, South Korea | 57216824872; 57952244800; 59913483100 | sjkang@knu.ac.kr; | IEEE ACCESS | IEEE ACCESS | 2169-3536 | 13 | SCIE | ENGINEERING, ELECTRICAL & ELECTRONIC;TELECOMMUNICATIONS;COMPUTER SCIENCE, INFORMATION SYSTEMS | 2024 | 3.6 | 34.8 | 0 | 2025-06-11 | 0 | 0 | Adaptation models; Computational modeling; Data models; Real-time systems; Accuracy; Noise; Acoustics; Atmospheric modeling; Transfer learning; Activity of daily living; ambient sound recognition; human activity recognition; Internet of Things; on-device AI; on-device AI | Activity of daily living; ambient sound recognition; human activity recognition; Internet of Things; on-device AI | Image coding; Activities of Daily Living; Ambient sound recognition; Ambient sounds; Classification accuracy; Human activity recognition; Location-aware; Model Adaptation; On-device AI; Sound modeling; Sound recognition; Transfer learning | English | 2025 | 2025 | 10.1109/access.2025.3568828 | 바로가기 | 바로가기 | 바로가기 | 바로가기 | |||||
| ○ | ○ | Article | MPFNet: Multiscale Prediction Network With Cross Fusion for Real-Time Semantic Segmentation | Semantic segmentation currently plays an important role in computer vision and is widely applied in both industry and human life. The self-driving car is one of the most trending applications, which assists humans in making informed decisions. The self-driving application has to interpret visual information from street scenes. However, how to effectively segment a long range of objective sizes is still a challenging problem. A feature pyramid network (FPN) builds up an architecture by processing four different features to contribute contextual and spatial information to the final map. Each feature can suitably process a specific range of objective sizes. Nevertheless, the final feature combination is not optimal when they raise the computation cost and reduce the semantic weights. We propose a multi-scale prediction network with cross-fusion in order to address the aforementioned drawbacks. The prediction module consists of three different predictions that allow the architecture to efficiently extract information of various sizes. Each prediction is generated from a pair of feature pyramids used to predict object classes. Furthermore, the cross-scale fusion is designed to enhance the weight aggregation of the final score map. The core component of the cross-fusion is the selective attention mechanism that determines uncertain weights of the lower prediction and then selects the complement from the adjacent prediction. By implementing this proposed scheme, we have achieved good results 78.3% mIoU and 45 FPS on Cityscapes and 45.9% mIoU on Mapillary Vistas datasets. Our method outperforms the baseline method with 7.0 mIoU improvement and a 27 FPS speedup on Cityscapes dataset. The experiment results demonstrate that the proposed model achieves a reasonable balance between performance and efficiency. | Toan Quyen, Van; Kim, Min Young | Kyungpook Natl Univ, IT Coll, Sch Elect & Elect Engn, Daegu 41566, South Korea; Kyungpook Natl Univ, IT Coll, Res Ctr Neurosurg Robot Syst, Daegu 41566, South Korea | 59593198200; 56739349100 | minykim@knu.ac.kr; | IEEE ACCESS | IEEE ACCESS | 2169-3536 | 13 | SCIE | ENGINEERING, ELECTRICAL & ELECTRONIC;TELECOMMUNICATIONS;COMPUTER SCIENCE, INFORMATION SYSTEMS | 2024 | 3.6 | 34.8 | 0 | 2025-05-07 | 0 | 0 | Real-time semantic segmentation; attention mechanism; multi-scale prediction; multi-scale prediction; context fusion; context fusion; feature pyramid network; feature pyramid network; feature pyramid network | AGGREGATION | attention mechanism; context fusion; feature pyramid network; multi-scale prediction; Real-time semantic segmentation | Data fusion; Prediction models; Scales (weighing instruments); Spatio-temporal data; Attention mechanisms; Context fusion; Feature pyramid; Feature pyramid network; Multi scale prediction; Pyramid network; Real-time semantic segmentation; Real-time semantics; Self drivings; Semantic segmentation; Semantic Segmentation | English | 2025 | 2025 | 10.1109/access.2025.3540454 | 바로가기 | 바로가기 | 바로가기 | 바로가기 | ||||
| ○ | ○ | Article | Multi-Sonar Fusion-Based Precision Underwater 3D Reconstruction for Optimal Scan Path Planning of AUV | This paper proposes a multi-sonar fusion-based underwater three-dimensional (3D) point cloud generation method for optimal scan path planning of an autonomous underwater vehicle (AUV). 3D reconstruction based on forward-looking sonar (FLS), a crucial sensor due to its robustness in underwater environments, is essential for underwater mapping and scan path planning of AUVs. A multi-sonar fusion method using FLS and profiling sonar (PS) in a previous study can eliminate the false slope points generated due to noise and the ambiguity in the vertical field-of-view (FoV) of the FLS. However, it shows limitations depending on "the encountered yaw angle", defined as the angle between AUV's scan path direction and the front edge line of object's upper area, which cannot be easily accounted for using a two-dimensional (2D) PS map. The proposed method detects the encountered yaw angle in the current scan path and estimates the next optimal scan path where the encountered yaw angle is suitable for fusion. Subsequently, using the fused 3D reconstruction result from the optimal scan path, we propose a 3D reconstruction of PS data for fusion with the FLS. We evaluated our method in the simulator and water tank experiments. The results show that the proposed method surpasses other methods, FLS only and the conventional fusion method with the FLS and 2D PS map in both environments. Finally, the proposed method will contribute to applications such as underwater exploration and AUV navigation, which require accurate mapping and scan path planning. | Rho, Sehwan; Joe, Hangil; Sung, Minsung; Kim, Jason; Kim, Seungmin; Yu, Son-Cheol | Pohang Univ Sci & Technol POSTECH, Dept Convergence IT Engn, Pohang 37673, South Korea; Kyungpook Natl Univ, Dept Robot & Smart Syst Engn, Daegu 41566, South Korea; Fraunhofer Inst Factory Operat & Automat, D-39106 Magdeburg, Germany | ; Joe, Hangil/LOR-9635-2024 | 57209683956; 55848385500; 57192577834; 59086663600; 59378684300; 8712522100 | sncyu@postech.ac.kr; | IEEE ACCESS | IEEE ACCESS | 2169-3536 | 13 | SCIE | ENGINEERING, ELECTRICAL & ELECTRONIC;TELECOMMUNICATIONS;COMPUTER SCIENCE, INFORMATION SYSTEMS | 2024 | 3.6 | 34.8 | 0 | 2025-05-07 | 0 | 0 | Three-dimensional displays; Sonar; Image edge detection; Point cloud compression; Imaging; Image reconstruction; Geometry; Feature extraction; Acoustic beams; Accuracy; Autonomous underwater vehicle (AUV); forward looking sonar; sensor fusion; underwater 3D reconstruction; underwater image sonar | 3-DIMENSIONAL RECONSTRUCTION | Autonomous underwater vehicle (AUV); forward looking sonar; sensor fusion; underwater 3D reconstruction; underwater image sonar | Autonomous underwater vehicles; Image fusion; Magnetic levitation vehicles; Motion planning; Photomapping; Sensor data fusion; Sonar; Underwater acoustics; Underwater imaging; Underwater photography; Autonomous underwater vehicle; Autonomous underwater vehicles]; Forward looking sonars; Profiling sonar; Scan path; Sensor fusion; Three-dimensional reconstruction; Underwater image sonar; Underwater three-dimensional reconstruction; Yaw angles; Water tanks | English | 2025 | 2025 | 10.1109/access.2025.3542084 | 바로가기 | 바로가기 | 바로가기 | 바로가기 | |||
| ○ | ○ | Article | Noise Suppression Method With Low-Complexity Noise Estimation Model and Heuristic Noise-Masking Algorithm for Real-Time Processing of Robot Vacuum Cleaners | Noise suppression in a high-level noise environment using a low-complexity method is challenging. This study proposes a low-complexity noise suppression algorithm for robot vacuum cleaner processors. We collected working noise from a robot vacuum cleaner along with speech signals and developed a method to extract the desired speech signal while estimating the noise. Our approach estimates the noise in the existing signal and converts it into the desired signal. In addition, we designed a low-complexity neural network capable of operating on mobile processors. The evaluation results demonstrate that our method achieves a performance comparable to that of highly computational methods. Notably, our method maintains superior performance when the intensity of the desired signal is low, and its performance is less degraded than that of other methods. It exhibits less degradation than existing methods, and in contrast to other neural networks, it avoids generating incorrect signals. Furthermore, we simplified the neural network architecture reducing its size by approximately 25% with minimal performance loss. | Shin, Seunghyeon; Kim, Minhan; Jeon, Inkoo; Song, Ju-Man; Park, Yongjin; Son, Jungkwan; Lee, Seokjin | Kyungpook Natl Univ, Sch Elect & Elect Engn, Daegu 41566, South Korea; LG Elect, Adv Robot Lab, Seoul 07336, South Korea; Kyungpook Natl Univ, Sch Elect Engn, Daegu 41566, South Korea | 57221769296; 57216617123; 59475546100; 59091026400; 59092152000; 59090481800; 36174416200 | sjlee6@knu.ac.kr; | IEEE ACCESS | IEEE ACCESS | 2169-3536 | 13 | SCIE | ENGINEERING, ELECTRICAL & ELECTRONIC;TELECOMMUNICATIONS;COMPUTER SCIENCE, INFORMATION SYSTEMS | 2024 | 3.6 | 34.8 | 0 | 2025-05-07 | 0 | 0 | Noise; Robots; Noise reduction; Neural networks; Complexity theory; Collision avoidance; Acoustics; Signal to noise ratio; Program processors; Spectrogram; Source separation; low-complexity; low-SNR; machine learning; mask estimation; mono channel; neural network; noise suppression; robot vacuum cleaner | low-complexity; low-SNR; machine learning; mask estimation; mono channel; neural network; noise suppression; robot vacuum cleaner; Source separation | Channel estimation; Echo suppression; Heuristic algorithms; Low SNR; Lower complexity; Machine-learning; Mask estimations; Mono channel; Neural-networks; Noise suppression; Performance; Robot vacuum cleaners; Speech signals; Blind source separation | English | 2025 | 2025 | 10.1109/access.2024.3522937 | 바로가기 | 바로가기 | 바로가기 | 바로가기 | |||||
| ○ | ○ | Article | Prototypical Few-Shot Learning for Histopathology Classification: Leveraging Foundation Models With Adapter Architectures | Histopathology is a critical tool for disease diagnosis and identifying cancer via microscopic tissue analysis. Traditional deep learning methods for histopathology often require extensive labeled data, which can be scarce and expensive. This study introduces a framework for the few-shot adaptation of self-supervised histopathology pretrained foundation models using multilayer perception adapters and convolutional adapters. An adapter comprising two linear or convolutional layers with nonlinear activation and residual connections transforms embeddings from foundation models for histopathology classification tasks. This study employs prototypical networks, SimpleShot, and bias-diminishing cosine similarity-based prototypical networks as few-shot learning algorithms. Comprehensive experiments are conducted across benchmark histopathology datasets: NCT, LC25000, Kather, and Camelyon17 Wilds. The results demonstrate that both adapter architectures consistently outperform the linear probe method, whereas multilayer perception adapters have an overall higher accuracy, especially when fine-tuned with five or more samples. The iBOT model with multilayer perception adapters fine-tuned using the bias-diminishing cosine similarity-based prototypical network algorithm achieved remarkable accuracy, reaching 95.04% on Camelyon17 Wilds and 96.55% on the NCT dataset with 20 images per class while using less than 0.002% of the dataset. These findings underscore the effectiveness of the proposed approach in addressing challenges posed by low-data regimes in the computer-aided histopathology domain and the potential for optimizing foundation models with minimal labeled data using prototypical few-shot algorithms. | Hasan, Kazi Rakib; Kim, Sijin; Cho, Junghwan; Han, Hyung Soo | Kyungpook Natl Univ, Dept Biomed Sci, Daegu 41944, South Korea; Kyungpook Natl Univ, Clin Om Inst, Daegu 41405, South Korea; Kyungpook Natl Univ, Dept Physiol, Daegu 41944, South Korea | RAKIB HASAN, KAZI/GNP-8208-2022 | 57825087200; 59121832700; 59899936300; 7401969388 | joshua@knu.ac.kr; | IEEE ACCESS | IEEE ACCESS | 2169-3536 | 13 | SCIE | ENGINEERING, ELECTRICAL & ELECTRONIC;TELECOMMUNICATIONS;COMPUTER SCIENCE, INFORMATION SYSTEMS | 2024 | 3.6 | 34.8 | 0 | 2025-06-11 | 0 | 0 | Histopathology; Adaptation models; Foundation models; Feature extraction; Biomedical imaging; Computational modeling; Nonhomogeneous media; Few shot learning; Data models; Benchmark testing; Adapter; bias-diminishing cosine similarity-based prototypical network (BDCSPN); breast cancer; Camelyon17; Camelyon17 wilds; colorectal cancer; convolutional (Conv) adapter; few-shot learning (FSL); foundation models; histopathology classification; Kather; LC25000; multilayer perception (MLP) adapter; NCT; PatchCamelyon (PCAM); prototypical network (ProtoNet); self-supervised learning (SSL); SimpleShot; multi-organ histopathology classification | Adapter; bias-diminishing cosine similarity-based prototypical network (BDCSPN); breast cancer; Camelyon17; Camelyon17 wilds; colorectal cancer; convolutional (Conv) adapter; few-shot learning (FSL); foundation models; histopathology classification; Kather; LC25000; multi-organ histopathology classification; multilayer perception (MLP) adapter; NCT; PatchCamelyon (PCAM); prototypical network (ProtoNet); self-supervised learning (SSL); SimpleShot | Convolution; Cosine transforms; Deep learning; Adapter; Bias-diminishing cosine similarity-based prototypical network; Breast Cancer; Camelyon17; Camelyon17 wild; Colorectal cancer; Convolutional adapter; Cosine similarity; Few-shot learning; Foundation models; Histopathology classification; Kather; Lc25000; Multi-layer perception; Multi-organ histopathology classification; Multilayer perception adapter; NCT; Patchcamelya; Prototypical network; Self-supervised learning; Simpleshot; Labeled data | English | 2025 | 2025 | 10.1109/access.2025.3570673 | 바로가기 | 바로가기 | 바로가기 | 바로가기 | ||||
| ○ | ○ | Article | Radar-Based Hand Gesture Recognition With Feature Fusion Using Robust CNN-LSTM and Attention Architecture | In Human-Computer Interaction (HCI), seamless hand gesture recognition is essential for intuitive and natural interactions. Gestures act as a universal language, bridging the gap between humans and machines. Radar-based recognition surpasses traditional optical methods, offering robust interaction capabilities in diverse environments. This article introduces a novel deep learning approach for hand gesture recognition, leveraging convolutional neural networks (CNNs), long short-term memory networks (LSTMs), and attention mechanisms. CNNs extract spatial features from radar signals, while LSTMs model the temporal dependencies crucial for dynamic gestures. Additionally, attention mechanisms enhance feature selection, ultimately improving recognition performance. We evaluate our method on the UWB-Gestures dataset, which has 12 gestures from eight people that were recorded using three X-Thru X4 UWB impulse radar sensors. Our processing pipeline integrates feature extraction, LSTM-attention blocks, and dense layers for final classification. Early fusion techniques, which combine spatial and temporal features in the initial stages, yield superior results, achieving an overall accuracy of 98.33% and outperforming intermediate fusion methods across gesture classes. To enhance the model's robustness, we evaluated its performance under common contributors of radar-specific noise scenarios in practical applications, including Gaussian noise, signal inversion, and multipath interference. Our model demonstrates high resilience, maintaining performance despite adverse conditions. As compared to state-of-the-art approaches, our approach delivers competitive accuracy and enhanced robustness, offering a reliable solution for noise-resilient radar-based hand gesture recognition in real-world applications. | Khan, Irshad; Kwon, Young-Woo | Kyungpook Natl Univ, Sch Comp Sci & Engn, Daegu 41566, South Korea | Kwon, Young-Woo/HGE-6607-2022 | 36166674500; 57208480210 | ywkwon@knu.ac.kr; | IEEE ACCESS | IEEE ACCESS | 2169-3536 | 13 | SCIE | ENGINEERING, ELECTRICAL & ELECTRONIC;TELECOMMUNICATIONS;COMPUTER SCIENCE, INFORMATION SYSTEMS | 2024 | 3.6 | 34.8 | 0 | 2025-05-07 | 0 | 0 | Hands; Gesture recognition; Radar; Feature extraction; Convolutional neural networks; Ultra wideband radar; Sensors; Frequency modulation; Deep learning; Machine learning; Hand gesture recognition; UWB-sensors; deep learning; fusion; noise robust | deep learning; fusion; Hand gesture recognition; noise robust; UWB-sensors | Convolutional neural networks; Deep neural networks; Feature Selection; Gaussian noise (electronic); Image coding; Image segmentation; Long short-term memory; Multilayer neural networks; Optical character recognition; Palmprint recognition; Radar interference; Attention mechanisms; Convolutional neural network; Deep learning; Hand-gesture recognition; Memory network; Noise robust; Performance; Short term memory; Spatial features; UWB-sensor; Gesture recognition | English | 2025 | 2025 | 10.1109/access.2025.3558293 | 바로가기 | 바로가기 | 바로가기 | 바로가기 | ||||
| ○ | ○ | Article | Reconfigurable Ultra-Miniaturized MIMO Antenna for Tissue-Independent Communication in Injectable Medical Implants | This article introduces a compact, reconfigurable multiple-input-multiple-output (MIMO) antenna tailored for injectable in-body implants, designed to operate efficiently in the 2.45 GHz industrial, scientific, and medical (ISM) band, ensuring consistent performance regardless of the host tissue type, such as muscle or fat. The antenna design achieves miniaturization using meandered lines and shorting pins, while reconfigurability is obtained by adjusting the meandered length with an ON-OFF switch (diode). When the diode is ON, the antenna's meandered length is increased, thereby extending its effective length and enabling resonance in the 2.42-2.48 GHz range for fat tissues. Conversely, when the diode is OFF, the effective length is reduced, allowing the same antenna to operate within the 2.38-2.48 GHz range in muscle tissues. The proposed antenna, fabricated on a flexible polyimide substrate (epsilon(r) =3.5), with overall dimensions of 5 mm x 2 mm x 0.05 mm, is enclosed within a biocompatible polyimide casing. The antenna's performance is evaluated through simulations in homogeneous single-layer (fat/muscle) and three-layer (skin, fat, and muscle) phantoms, along with a realistic heterogeneous hand model, and validated through an ex-vivo setup, which includes fabricating the antenna prototype and measuring its performance in pork meat and a phantom solution. This innovative, miniaturized, tissue-independent MIMO antenna achieves isolation greater than 15 dB via orthogonal arrangement and 1 mm element spacing, with impedance bandwidths of 60 MHz in fat and 100 MHz in muscle. Furthermore, MIMO characteristics are evaluated and antenna robustness and safety are verified through specific absorption rate measurements using voxel models, highlighting the antenna's suitability for injectable applications. | Harlan, L.; Susila, M.; Kumar, Sachin; Chul Choi, Hyun; Wook Kim, Kang | SRM Inst Sci & Technol, Fac Engn & Technol, Dept Elect & Commun Engn, Kattankulathur 603203, Tamil Nadu, India; Galgotias Coll Engn & Technol, Dept Elect & Commun Engn, Greater Noida 201310, Uttar Pradesh, India; Kyungpook Natl Univ, Sch Elect & Elect Engn, Daegu 41566, South Korea | L, HARLAN/MGV-5395-2025; M, Susila/AAS-9566-2020; Kumar, Sachin/W-2211-2019 | 58726329100; 55842129700; 56907994000; 55248359100; 59552277500 | gupta.sachin0708@gmail.com; kang_kim@ee.knu.ac.kr; | IEEE ACCESS | IEEE ACCESS | 2169-3536 | 13 | SCIE | ENGINEERING, ELECTRICAL & ELECTRONIC;TELECOMMUNICATIONS;COMPUTER SCIENCE, INFORMATION SYSTEMS | 2024 | 3.6 | 34.8 | 0 | 2025-05-07 | 0 | 0 | Antennas; Fats; Implants; Resonant frequency; Muscles; Antenna measurements; Polyimides; Permittivity; Design methodology; Substrates; Biomedical device; flexible antenna; injectable antenna; MIMO; reconfigurable antenna | CIRCUIT | Biomedical device; flexible antenna; injectable antenna; MIMO; reconfigurable antenna | Breath controlled devices; Electrotherapeutics; Microstrip antennas; Negative resistance; Photon correlation spectroscopy; Tissue engineering; Biomedical devices; Flexible antennas; Injectable antenna; Injectables; Multiple input multiple output antennas; Multiple inputs; Multiple outputs; Multiple-input-multiple-output; Reconfigurable; Reconfigurable antenna; Diodes | English | 2025 | 2025 | 10.1109/access.2025.3535780 | 바로가기 | 바로가기 | 바로가기 | 바로가기 | |||
| ○ | ○ | Article | ResTreeNet: A Height-Aware LiDAR Tree Classification Model With Explainable AI for Forestry Applications | Tree species classification plays a crucial role in forest management, biodiversity conservation, and ecological monitoring. Light detection and ranging (LiDAR) technology, capturing highly detailed 3D structural information of vegetation, has become a powerful tool for automated tree classification. Among LiDAR techniques, terrestrial LiDAR provides high-resolution point-cloud data by scanning trees from the ground level, enabling precise species identification. However, applying deep learning models to LiDAR-based tree classification remains challenging due to the computational complexity of existing 3D architectures, which often struggle with scalability and practical large-scale implementation. To address these critical limitations, we propose ResTreeNet, an efficient and lightweight deep learning model designed explicitly for tree classification using terrestrial LiDAR point clouds. Our innovative approach combines residual networks for hierarchical feature extraction, a height-based grouping strategy to enhance structural representation, and a parameterized geometric transformation module to improve translation invariance and model adaptability. This work integrates explainable artificial intelligence (XAI) techniques, including gradient-weighted class action mapping (Grad-CAM) visualizations, to provide transparent and interpretable insight into the classification reasoning of the model, addressing the critical need for understanding automated decision-making processes. The comprehensive evaluation on a terrestrial LiDAR dataset demonstrates the superior performance of ResTreeNet, achieving a state-of-the-art accuracy of 94.02% on samples with 1024-points, surpassing the existing models by 2.03%. The robust capabilities of the model are further validated by outstanding classification metrics, including precision (94.24%), recall (93.63%), and the F1-score (93.54%), ensuring a balanced and reliable approach to tree species classification. With its lightweight architecture (requiring only 0.47 million parameters) and computational efficiency, ResTreeNet is a practical solution for large-scale ecological research, offering an innovative approach to automated forest monitoring and sustainable resource management. | Taye, Asrat Kaleab; Park, Jeong-Mook; Cho, Hyung-Ju; Kang, Jin-Taek; Seo, Yeon-Ok | Kyungpook Natl Univ, Dept Software, Sangju 37224, Gyeongsangbuk D, South Korea; Natl Inst Forest Sci, Div Forest Management Res, Seoul 02455, South Korea | 59903016000; 57211901305; 55177091600; 59800097000; 53867163500 | pjm7@korea.kr; hyungju@knu.ac.kr; | IEEE ACCESS | IEEE ACCESS | 2169-3536 | 13 | SCIE | ENGINEERING, ELECTRICAL & ELECTRONIC;TELECOMMUNICATIONS;COMPUTER SCIENCE, INFORMATION SYSTEMS | 2024 | 3.6 | 34.8 | N/A | 0 | 0 | Vegetation; Laser radar; Forestry; Three-dimensional displays; Biological system modeling; Random forests; Computational modeling; Point cloud compression; Feature extraction; Vegetation mapping; Explainable artificial intelligence; residual network; terrestrial LiDAR; tree classification | SPECIES CLASSIFICATION | Explainable artificial intelligence; residual network; terrestrial LiDAR; tree classification | Conformal mapping; Decision trees; Linear transformations; Logging (forestry); Resource allocation; Risk perception; Vegetation; Explainable artificial intelligence; Innovative approaches; Large-scales; Learning models; Light detection and ranging; Residual network; Species classification; Terrestrial light detection and ranging; Tree classification; Tree species; Deep learning | English | 2025 | 2025 | 10.1109/access.2025.3567042 | 바로가기 | 바로가기 | 바로가기 | 바로가기 | |||||
| ○ | ○ | Article | Scalable Context-Based Facial Emotion Recognition Using Facial Landmarks and Attention Mechanism | Deciphering emotions from a person's perspective is critical for meaningful human relationships. Enabling computers to interpret emotional cues similarly could significantly improve human-machine interaction. Accurate emotion recognition involves more than just analyzing facial expressions; it requires situational context and facial landmarks, which together reveal a broader range of emotional states. Existing emotion recognition frameworks primarily focus on facial imaging, often overlooking the contextual elements and the subtle significance of facial landmarks. This paper proposes a scalable approach to emotion recognition that combines situational context comprehension, accurate facial landmark detection, and facial feature analysis. Due to its scalability, our model can be applied across diverse computational platforms and operational circumstances while maintaining high performance. The model's robustness and utility were validated against the EMOTIC benchmark, achieving an impressive overall accuracy of 84%. The findings underscore the importance of incorporating contextual information and facial landmarks to enhance emotion recognition accuracy. This advancement is expected to contribute substantially to fields such as augmented reality, medical imaging, and sophisticated human-computer interaction systems. | Colaco, Savina Jassica; Han, Dong Seog | Kyungpook Natl Univ, Ctr ICT & Automot Convergence, Daegu 41566, South Korea; Kyungpook Natl Univ, Sch Elect & Elect Engn, Daegu 41566, South Korea | Han, Dong Seog/N-8949-2018 | 59543674400; 7403219442 | dshan@knu.ac.kr; | IEEE ACCESS | IEEE ACCESS | 2169-3536 | 13 | SCIE | ENGINEERING, ELECTRICAL & ELECTRONIC;TELECOMMUNICATIONS;COMPUTER SCIENCE, INFORMATION SYSTEMS | 2024 | 3.6 | 34.8 | 0 | 2025-05-07 | 0 | 0 | Emotion recognition; Face recognition; Feature extraction; Accuracy; Representation learning; Attention mechanisms; Visualization; Scalability; Kernel; Image recognition; Contextual cues; deep learning; emotion recognition; facial landmarks; scalable models | EXPRESSION RECOGNITION | Contextual cues; deep learning; emotion recognition; facial landmarks; scalable models | Contrastive Learning; Emotion Recognition; Face recognition; Human computer interaction; Attention mechanisms; Context-based; Contextual cue; Deep learning; Emotion recognition; Facial emotions; Facial landmark; Human relationships; Scalable Modelling; Situational context; Medical imaging | English | 2025 | 2025 | 10.1109/access.2025.3534328 | 바로가기 | 바로가기 | 바로가기 | 바로가기 | |||
| ○ | Article | Ternary Toward Binary: Circuit-Level Implementation of Ternary Logic Using Depletion-Mode and Conventional MOSFETs | The application of artificial intelligence (AI) requires advanced computation to address complex problems. However, the improvement of binary computing systems supporting these applications is approaching their limits due to atomic-level scaling. Regarding this challenging situation, ternary computing is gaining more attention due to its better data saving/computing/moving capability. Thus, ternary logic based on various devices was proposed, but these circuits are still encountering issues of high-power consumption, low operating speed, and challenges in manufacturing compared to silicon-based circuits. Therefore, this paper presents a methodology for designing ternary logic based on Depletion-mode metal-oxide-semiconductor field-effect transistor (DEPFET) and multi-threshold voltage complementary metal-oxide-semiconductor (MTCMOS). Our silicon-based devices are easier to manufacture and support high-speed/low-power operations through our complementary ternary logic. Our balanced ternary full adder (BTFA) is 9.70 × better energy efficiency than the latestcarbon nanotube field-effect transistor (CNTFET) based BTFA. We also propose the first methodology to design a ternary cell layout in multi-height standard cell design. We propose an algorithm for the best ternary cell layout and a concept of integrated layout that reduces area when required cells are close to each other. © 2013 IEEE. | Lee, Hyundong; Kim, Seonghoon; Kim, Jongbeom; Jeong, Jaehoon; Yang, Jeonggyu; Song, Taigon | Kyungpook National University (KNU), School of Electronic and Electrical Engineering, Daegu, 41566, South Korea; Kyungpook National University (KNU), School of Electronic and Electrical Engineering, Daegu, 41566, South Korea; Kyungpook National University (KNU), School of Electronic and Electrical Engineering, Daegu, 41566, South Korea; Samsung Electronics, Gyeonggi-do, 18448, South Korea; Samsung Electronics, Gyeonggi-do, 18448, South Korea; Kyungpook National University (KNU), School of Electronic and Electrical Engineering, Daegu, 41566, South Korea | 57226892881; 59540375800; 57782068500; 57226881576; 57221952581; 36005021000 | tsong@knu.ac.kr; | IEEE Access | IEEE ACCESS | 2169-3536 | 2169-3536 | 13 | SCIE | ENGINEERING, ELECTRICAL & ELECTRONIC;TELECOMMUNICATIONS;COMPUTER SCIENCE, INFORMATION SYSTEMS | 2024 | 3.6 | 34.8 | 0 | 2025-05-07 | 0 | depletion-mode MOSFET; full-adder; layout; Multi-valued logic; ternary logic | Chemically sensitive field effect transistors; Electron beam lithography; Integrated circuit layout; Integrated circuit manufacture; Many valued logics; MOS devices; MOSFET devices; Process control; Semiconductor device manufacture; Threshold logic; Binary circuits; Cell layout; Depletion modes; Depletion-mode MOSFET; Full adders; Layout; MOSFETs; Multi-valued; Multi-valued logic; Ternary logic; Oxide semiconductors | English | Final | 2025 | 10.1109/access.2024.3523344 | 바로가기 | 바로가기 | 바로가기 | ||||||||
| ○ | ○ | Article | Transfer Learning-Based Ensemble Approach for Rainfall Class Amount Prediction | Predicting short-term precipitation amounts is challenging, especially due to meteorological data scarcity. While deep learning-based models have been shown to be more effective in predicting precipitation amounts, their performance heavily relies on the size of the training datasets. This paper presents a multi-station-based transfer learning ensemble approach to mitigate the data scarcity problem by transferring knowledge learned from multiple meteorological station datasets to a single target station. To achieve this, multi-layer perceptron, convolutional neural networks, and long-short-term memory (LSTM) systems were trained on weather station datasets from the Lake Victoria Basin (LVB). From the experiments, the LSTM model outperformed other state-of-the-art models achieving high F1 scores across individual stations. Fine-tuning pre-trained models for the target station demonstrated improved accuracy, with performance gains of up to 5%. Additionally, the ensemble of these models further enhanced performance, delivering highly accurate classification results. Summarily, the proposed ensemble approach demonstrates significant improvements in predicting rainfall class amounts, offering a robust solution for precipitation forecasting in data-scarce regions like the LVB. | Gahwera, Tumusiime Andrew; Eyobu, Odongo Steven; Isaac, Mugume; Kakuba, Samuel; Han, Dong Seog | Makerere Univ, Sch Comp & Informat Technol, Dept Informat Syst, Kampala, Uganda; Makerere Univ, Sch Comp & Informat Technol, Dept Networks, Kampala, Uganda; Makerere Univ, Sch Forestry Environm & Geog Sci, Dept Geog Geoinformat & Climat Sci, Kampala, Uganda; Kyungpook Natl Univ, Sch Elect & Elect Engn, Daegu 41566, South Korea; Kabale Univ, Fac Engn Technol Appl Design & Fine Art, Kabale, Uganda | Han, Dong Seog/N-8949-2018; Kakuba, Samuel/HLX-4856-2023; Eyobu, Odongo/ABD-9473-2020 | 59709041600; 57190814517; 59032163100; 57988218000; 59307662300 | dshan@knu.ac.kr; | IEEE ACCESS | IEEE ACCESS | 2169-3536 | 13 | SCIE | ENGINEERING, ELECTRICAL & ELECTRONIC;TELECOMMUNICATIONS;COMPUTER SCIENCE, INFORMATION SYSTEMS | 2024 | 3.6 | 34.8 | 0 | 2025-05-07 | 0 | 0 | Predictive models; Ensemble learning; Rain; Data models; Weather forecasting; Accuracy; Transfer learning; Biological system modeling; Atmospheric modeling; Long short term memory; Deep learning; ensemble learning; fine-tuning; rainfall class amount prediction; transfer learning | Deep learning; ensemble learning; fine-tuning; rainfall class amount prediction; transfer learning | Adversarial machine learning; Contrastive Learning; Federated learning; Long short-term memory; Multilayer neural networks; Prediction models; Rain; Transfer learning; Weather forecasting; Data scarcity; Deep learning; Ensemble approaches; Ensemble learning; Fine tuning; Lake Victoria; Performance; Rainfall class amount prediction; Short term memory; Transfer learning; Convolutional neural networks | English | 2025 | 2025 | 10.1109/access.2025.3551737 | 바로가기 | 바로가기 | 바로가기 | 바로가기 | ||||
| ○ | ○ | Article | Comparative analysis of methods for calculating Hubbard parameters using cRPA | In this study, we present a systematic comparison of various approaches within the constrained random-phase approximation (cRPA) for calculating the Coulomb interaction parameter U. While defining the correlated space is straightforward for disentangled bands, the situation is more complex for entangled bands, where different projection schemes from hybridized bands to the target space can yield varying sizes of interaction parameters. We systematically evaluated different methods for calculating the polarizability functions within the correlated space. Furthermore, we analyze how different definitions of the correlated space, often constructed through Wannierization from Kohn-Sham orbitals, defines the orbital localization and play a crucial role in determining the interaction parameter. To illustrate these effects, we consider two sets of representative correlated d-orbital oxides: LiMO2 (M = V - Ni) as examples of isolated d-electron systems and SrMO3 (M = Mn, Fe, and Co) as cases of entangled d-electron systems. Through this systematic comparison, we provide a detailed analysis of different cRPA methodologies for computing the Hubbard parameters. | Reddy, Indukuru Ramesh; Kaltak, M.; Kim, Bongjae | Kyungpook Natl Univ, Dept Phys, Daegu 41566, South Korea; VASP Software GmbH, Berggasse 21-14, A-1090 Vienna, Austria | Reddy, Indukuru/AAF-2670-2019 | 59910600200; 55337824700; 55650566000 | merzuk.kaltak@vasp.at; bongjae@knu.ac.kr; | PHYSICAL REVIEW B | PHYS REV B | 2469-9950 | 2469-9969 | 111 | 19 | SCIE | MATERIALS SCIENCE, MULTIDISCIPLINARY;PHYSICS, APPLIED;PHYSICS, CONDENSED MATTER | 2024 | 3.7 | 35.0 | 0 | 2025-06-11 | 1 | 0 | COULOMB INTERACTIONS; DIELECTRIC-CONSTANT; LA2CUO4 | Comparative analyzes; Coulomb interaction parameters; D electrons; Electron systems; Hubbard; Hybridized bands; Interaction parameters; Projection schemes; Random phase approximations; Target space | English | 2025 | 2025-05-21 | 10.1103/physrevb.111.195144 | 바로가기 | 바로가기 | 바로가기 | 바로가기 | |||
| ○ | ○ | Article | Stylopization by Xenos spp. (Xenidae, Strepsiptera) in invasive alien hornet, Vespa velutina, in South Korea | The invasive hornet Vespa velutina Lepeletier, which first invaded South Korea in 2003, has spread throughout the country, significantly affecting apiaries, ecosystems, and human health. Xenos spp. (Xenidae, Strepsiptera) are primarily parasitic to social wasps, with V. analis being the only known host in Korea. Until recently, no parasites or parasitoids on V. velutina had been discovered. In 2020, strepsipteran parasites were discovered on 11 hornet workers in Andong City, South Korea. These parasites, comprising four larvae and seven pupae, were all male, except for one individual of an undetermined sex. Molecular analysis and morphological examination identified the parasites as Xenos moutoni (du Buysson, 1903) and X. oxyodontes Nakase & Kato, 2013. This marks the first recorded instance of strepsipteran parasites on V. velutina in regions invaded by this hornet. Although the exact infection rate of these parasites could not be determined, it appears that native strepsipteran parasites have adapted to a non-native Vespa species. Stylopization, the condition caused by these parasites, is known to negatively affect hornet colonies: infected workers do not contribute to nest activities, hindering nest development, and infected reproductive individuals (males and new queens) do not mate, which impedes the establishment of new colonies. However, due to the hornet's high reproductive rate and compensatory mechanisms, the overall control effect of the parasites is likely to be minor. Le frelon invasif Vespa velutina Lepeletier, qui a envahi la Cor & eacute;e du Sud pour la premi & egrave;re fois en 2003, s'est r & eacute;pandu dans tout le pays, affectant consid & eacute;rablement les ruchers, les & eacute;cosyst & egrave;mes et la sant & eacute; humaine. Les esp & egrave;ces de Xenos (Xenidae, Strepsiptera) sont principalement des parasites des gu & ecirc;pes sociales, V. analis & eacute;tant le seul h & ocirc;te connu en Cor & eacute;e. Jusqu'& agrave; r & eacute;cemment, aucun parasite ou parasito & iuml;de sur V. velutina n'avait & eacute;t & eacute; d & eacute;couvert. En 2020, des strepsipt & egrave;res parasites ont & eacute;t & eacute; d & eacute;couverts sur 11 ouvri & egrave;res de frelons dans la ville d'Andong, en Cor & eacute;e du Sud. Ces parasites, comprenant quatre larves et sept pupes, & eacute;taient tous m & acirc;les, & agrave; l'exception d'un individu de sexe ind & eacute;termin & eacute;. L'analyse mol & eacute;culaire et l'examen morphologique ont permis d'identifier les parasites comme & eacute;tant Xenos moutoni (du Buysson, 1903) et X. oxyodontes Nakase & Kato, 2013. Il s'agit du premier cas enregistr & eacute; de strepsipt & egrave;res parasites chez V. velutina dans les r & eacute;gions envahies par ce frelon. Bien que le taux d'infection exact de ces parasites n'ait pas pu & ecirc;tre d & eacute;termin & eacute;, il semble que les strepsipt & egrave;res parasites indig & egrave;nes se soient adapt & eacute;s & agrave; une esp & egrave;ce de Vespa non indig & egrave;ne. La stylopisation, la condition caus & eacute;e par ces parasites, est connue pour affecter n & eacute;gativement les colonies de frelons : les ouvri & egrave;res infect & eacute;es ne contribuent pas aux activit & eacute;s de nidification, ce qui entrave le d & eacute;veloppement du nid, et les individus reproducteurs infect & eacute;s (m & acirc;les et nouvelles reines) ne s'accouplent pas, ce qui emp & ecirc;che l'& eacute;tablissement de nouvelles colonies. Cependant, en raison du taux de reproduction & eacute;lev & eacute; du frelon et de ses m & eacute;canismes compensatoires, il est probable que l'effet global de contr & ocirc;le par ces parasites ne sera que mineur. | Kim, Il-Kwon; Kim, Chang-Jun; Choi, Jeong-Hwan; Kang, Hyun Jun; Choi, Moon Bo | Korea Natl Arboretum, Div Forest Biodivers, Pochon 11186, South Korea; Korea Natl Arboretum, Div Gardens & Educ, Pochon 11186, South Korea; Haesol Ecofriendly Res Inst, Busan 46720, South Korea; Kyungpook Natl Univ, Inst Agr Sci & Technol, Daegu 41566, South Korea; Wild Beei, Dept R&D, Chilgok 39864, South Korea | KIm, Changjun/GZM-7308-2022 | 55477687300; 55286588300; 57844023000; 59650807600; 51863232400 | kosinchoi@hanmail.net; | PARASITE | PARASITE | 1252-607X | 1776-1042 | 32 | SCIE | PARASITOLOGY | 2024 | 2.4 | 35.1 | 0 | 2025-05-07 | 0 | 0 | Strepsiptera; Vespa velutina; Invasive species; DNA barcodes; Xenos moutoni; X. oxyodontes | HYMENOPTERA VESPIDAE; GAOLIGONG MOUNTAINS; GROUP-SIZE; HOST; NIGRITHORAX; WASP; PREVALENCE; LEPELETIER; EVOLUTION; PARASITES | DNA barcodes; Invasive species; Strepsiptera; Vespa velutina; X. oxyodontes; Xenos moutoni | Animals; Female; Host-Parasite Interactions; Introduced Species; Larva; Male; Pupa; Republic of Korea; Wasps; animal; classification; female; growth, development and aging; host parasite interaction; introduced species; larva; male; parasitology; physiology; pupa; South Korea; wasp | English | 2025 | 2025-02-17 | 10.1051/parasite/2025004 | 바로가기 | 바로가기 | 바로가기 | 바로가기 |
페이지 이동: