Article
Energy & Fuels
Soyeong Park, Seungwook Yoon, Byungtak Lee, Seokkap Ko, Euiseok Hwang
Summary: A new joint detection and imputation method is proposed to handle clustered bad data in residential electricity loads. The method investigates neighbors of invalid data points using probabilistic forecasting techniques and achieved a 95.5% accuracy in detecting clustered bad data.
Article
Engineering, Electrical & Electronic
Seungwook Yoon, Euiseok Hwang
Summary: This study proposes a novel joint multi-track multi-level (MTML) read channel signal processing scheme to enhance data transfer rate in array-reader-based interlaced magnetic recording (ARIMR). Numerical evaluations using a micro-pixelated magnetic channel model show improvements in the cross-track profile of bit error rate with the proposed MTML-ARIMR approach for dual-reader and triple-track readback configurations.
IEEE TRANSACTIONS ON MAGNETICS
(2021)
Article
Chemistry, Analytical
Jihoon Lee, Seungwook Yoon, Euiseok Hwang
Summary: This paper proposes an AE compression model based on a frequency selection method to improve reconstruction quality while maintaining compression ratio, reducing computational complexity involved in the learning process. The model shows significantly higher reconstruction performance than existing models and is able to adapt to varying spectral properties of power data over time.
Article
Spectroscopy
Jang-Hee Choi, Sungho Shin, Youngmin Moon, Jung Hyun Han, Euiseok Hwang, Sungho Jeong
Summary: This study reports the classification and imaging of melanoma in murine skin tissue using femtosecond laser-induced breakdown spectroscopy (fs-LIBS). The use of an ultraviolet (UV) fs-laser and support vector machine (SVM) classifier achieved an accurate classification accuracy of about 97.9%, making UV-fs-LIBS a potentially valuable tool for melanoma screening.
SPECTROCHIMICA ACTA PART B-ATOMIC SPECTROSCOPY
(2021)
Article
Engineering, Electrical & Electronic
Junho Song, Yonggu Lee, Euiseok Hwang
Summary: This paper presents a novel mask-based load disaggregation scheme using a deep neural network customized in a time-frequency domain to effectively extract the flexible portion of loads. The proposed scheme outperforms conventional disaggregation models in discriminating the contributions of the flexible load from the total power consumption, as shown in numerical evaluations.
IEEE TRANSACTIONS ON SMART GRID
(2021)
Article
Automation & Control Systems
Ekta Srivastava, Hyebin Kim, Jaepil Lee, Sungho Shin, Sungho Jeong, Euiseok Hwang
Summary: In this study, a domain adaptation-based approach is proposed for the compositional analysis of metal scraps using laser-induced breakdown spectroscopy (LIBS) measurements. The method improves the performance by transferring knowledge from a source domain to the target domain through domain adaptation. Experimental results demonstrate the effectiveness of the proposed approach in quantifying the elemental concentration of metal scraps.
CHEMOMETRICS AND INTELLIGENT LABORATORY SYSTEMS
(2022)
Article
Computer Science, Information Systems
Yongjune Kim, Euiseok Hwang, B. V. K. Vijaya Kumar
Summary: A channel coding scheme utilizing flash memory channel state information is proposed to combat inter-cell interference, improving data reliability by transforming the ICI problem into a memory model with defective cells. Simulation results show the effectiveness of the proposed scheme in reducing decoding failure probability.
JOURNAL OF COMMUNICATIONS AND NETWORKS
(2022)
Article
Instruments & Instrumentation
Ekta Srivastava, Hyebin Kim, Jaepil Lee, Sungho Shin, Sungho Jeong, Euiseok Hwang
Summary: In this study, a transfer learning-based classification model is proposed for scrap metal identification using laser-induced breakdown spectroscopy (LIBS) and machine learning methods. The model addresses the challenges of the diversity of scrap metal encountered in field measurements and differences in experimental configurations. It augments the training dataset and utilizes a convolutional neural network for real-time classification. The experimental results show high accuracy and improved performance compared to conventional models.
APPLIED SPECTROSCOPY
(2023)
Article
Chemistry, Multidisciplinary
Giup Seo, Seungwook Yoon, Junyoung Song, Ekta Srivastava, Euiseok Hwang
Summary: Photovoltaic (PV) fault detection approaches can be categorized into two groups: end-to-end and threshold methods. While the former relies on labeled datasets to directly detect faults, the latter estimates power generation and uses thresholds to detect deviations from estimated values. This paper proposes a novel deep reinforcement learning (DRL)-based label-free fault detection scheme that dynamically assigns thresholds with suitable confidence intervals under varying environmental conditions.
APPLIED SCIENCES-BASEL
(2023)
Article
Computer Science, Information Systems
Yang-Hsi Su, Chouchang (Jack) Yang, Euiseok Hwang, Alanson P. Sample
Summary: Cost-effective and accurate localization of radio transmitters has great potential in various fields. However, existing methods have limitations such as high cost and complexity. This paper proposes a new approach using a high-speed RF mux and sub-packet switching, which eliminates the need for synchronization and reduces cost and system complexity. The results show high accuracy and the ability to track multiple users.
PROCEEDINGS OF THE ACM ON INTERACTIVE MOBILE WEARABLE AND UBIQUITOUS TECHNOLOGIES-IMWUT
(2023)
Article
Computer Science, Information Systems
Seungwook Yoon, Seungnam Han, Euiseok Hwang
Summary: This study proposes a novel authentication scheme that provides enhanced security for Internet of Things (IoT) applications by integrating a radio frequency physical unclonable function (PUF) with a device PUF. The scheme effectively defends against diverse attacks by using the physical layer features of wireless channels and the device PUF's characteristics to generate the cryptographic key.
IEEE INTERNET OF THINGS JOURNAL
(2023)
Proceedings Paper
Green & Sustainable Science & Technology
Myungsun Kim, Dongju Kim, Euiseok Hwang, Eden Kim, Seok-gap Ko, Byung-tak Lee
Summary: This paper proposes a framework for inferring socio-demographic information using smart meter data. It uses transfer learning methodology with datasets from different countries to train a deep learning model, and improves the model performance by instance selection and feature removal. The proposed method enhances the accuracy and performance of information inference.
2022 6TH INTERNATIONAL CONFERENCE ON GREEN ENERGY AND APPLICATIONS (ICGEA 2022)
(2022)
Proceedings Paper
Computer Science, Artificial Intelligence
Seungwook Yoon, Euiseok Hwang
Summary: This paper introduces a new decentralized electric vehicle (EV) charging coordination scheme for smart buildings, aiming to minimize overall operating costs for EV owners while protecting privacy. Experimental results demonstrate that this price control-based decentralized EV coordination can reduce overall electricity costs for owners and arbitration type aggregators of buildings.
2021 IEEE INTERNATIONAL SMART CITIES CONFERENCE (ISC2)
(2021)
Proceedings Paper
Computer Science, Information Systems
Seungnam Han, Euiseok Hwang
Summary: This paper explores the implicit detection of object movement in an indoor environment using RF attributes and CSI, tracking changes in CSI with simple statistical measures to detect movements of objects effectively, even for small objects less than half the RF wavelength.
2021 WORKSHOP ON COMMUNICATION NETWORKS AND POWER SYSTEMS (WCNPS)
(2021)
Article
Computer Science, Information Systems
Giup Seo, Seungwook Yoon, Myungsun Kim, Changho Mun, Euiseok Hwang
Summary: This research introduces a predictive control scheme based on deep reinforcement learning to reduce energy consumption and costs in wastewater treatment plant pump systems. Utilizing a deep neural network model and DRL agent, the scheme considers energy consumption, electricity prices, and pump wear prevention factors for efficient control.