Daegu Gyeongbuk Institute of Science and Technology
DGIST Library Institutional RepositoryNot a member yet
12664 research outputs found
Sort by
CaCO3-TiO2 전극이 염료감응형 베타전지(DSBC)의 성능에 미치는 영향에 관한 연구
No full textNuclear energy, Betavoltaic Cell, Dye Sensitized Solar Cell (DSSC), Dye Sensitized Betavoltaic Cell (DSBC)MasterdCollectio
해군 위성 SAR 레이더를 위한 잡음 제거 기반의 영상 트랜스포머(Vision Transformer) 인식 향상 방법
No full text본 논문에서는 해군 위성 SAR (Synthetic Aperture Radar)를 위한 잡음 제거 기반의 영상 트랜스포 머 (Vision Transformer) 인식 향상 방법을 제안한다. 최근 미 해군은 SAR 데이터를 활용하여 기상 조 건에 영향을 받지 않고 육지나 해상의 상태를 관측하고 추적하고 있다. 그러나 기존 SAR에 많이 활용되 는 인식 기법들은 인식률의 한계를 가지고 있다. 특히, 최근 ViT (Vision Transformer)나 스윈 스랜스 포머와 같은 영상 트랜스포머 기법은 최근 컴퓨터 비전 분야에서 널리 사용되고 있지만, SAR 결과를 활 용하여 타겟 인식을 하는 연구가 부족한 상황이다. 또한, 다양한 환경 조건으로 인해 잡음의 영향이 큰 경우에, SAR는 영상 트랜스포머 기법의 인식률을 높이기 위한 다양한 시도가 필요하다. 이를 해결하기 위해, 본 논문에서는 DnCNN (Denoising Convolutional Neural Networks) 기반의 영상 트랜스포머 인식 향상 방법에 대해 논의한다. 이 방법은 SAR 데이터에 transformer 기법을 적용하기 전에 잡음를 제거하여 인식률을 향상시킨다. 제안된 방법을 적용한 결과, DnCNN기반의 ViT의 경우 인식률이 기존 ViT 방식에 비해 약 119% 개선되었다. 또한, DnCNN기반의 스윈 트랜스포머의 경우에는 인식률이 기 존 스윈 트랜스포머 방식에 비해 약 18% 개선된 결과를 보여준다.
This paper proposes a noise removal-based image transformer enhancement method for naval satellite Synthetic-Aperture Radar (SAR). The recent U.S. Navy extensively utilizes SAR data to observe and track the conditions of land and sea regardless of weather conditions. However, existing recognition techniques widely used in SAR have limitations in recognition rates. Especially, recent image transformer techniques such as ViT and Swin Transformer are widely used in the computer vision field, but there is a lack of research utilizing SAR results for target recognition. Moreover, in cases where noise significantly affects due to various environmental conditions, SAR requires various attempts to improve the recognition rate of image transformer techniques. To address this issue, this paper discusses the denoising-based image transformer enhancement method using DnCNN. This method removes noise from SAR data before applying transformer techniques to improve the recognition rate. The results show that applying the proposed method improved the recognition rate by approximately 87% for ViT and approximately 18% for Swin Transformer compared to previous results.FALSEdomesti
Hopping the Hurdle: Strategies to Enhance the Molecular Delivery to the Brain through the Blood-Brain Barrier
Get PDFModern medicine has allowed for many advances in neurological and neurodegenerative disease (ND). However, the number of patients suffering from brain diseases is ever increasing and the treatment of brain diseases remains an issue, as drug efficacy is dramatically reduced due to the existence of the unique vascular structure, namely the blood–brain barrier (BBB). Several approaches to enhance drug delivery to the brain have been investigated but many have proven to be unsuccessful due to limited transport or damage induced in the BBB. Alternative approaches to enhance molecular delivery to the brain have been revealed in recent studies through the existence of molecular delivery pathways that regulate the passage of peripheral molecules. In this review, we present recent advancements of the basic research for these delivery pathways as well as examples of promising ventures to overcome the molecular hurdles that will enhance therapeutic interventions in the brain and potentially save the lives of millions of patients. © 2024 by the authors.TRUEsciescopu
Sustaining Surface Lithiophilicity of Ultrathin Li-Alloy Coating Layers on Current Collector for Zero-Excess Li-Metal Batteries
No full textZero-excess Li-metal batteries (ZE-LMBs) have emerged as the ultimate battery platform, offering an exceptionally high energy density. However, the absence of Li-hosting materials results in uncontrolled dendritic Li deposition on the Cu current collector, leading to chronic loss of Li inventory and severe electrolyte decomposition, limiting its full utilization upon cycling. This study presents the application of ultrathin (≈50nm) coatings comprising six metallic layers (Cu, Ag, Au, Pt, W, and Fe) on Cu substrates in order to provide insights into the design of Li-depositing current collectors for stable ZE-LMB operation. In contrast to non-alloy Cu, W, and Fe coatings, Ag, Au, and Pt coatings can enhance surface lithiophilicity, effectively suppressing Li dendrite growth, thereby improving Li reversibility. Considering the distinct Li-alloying behaviors, particularly solid-solution and/or intermetallic phase formation, Pt-coated Cu current collectors maintain surface lithiophilicity over repeated Li plating/stripping cycles by preserving the original coating layer, thereby attaining better cycling performance of ZE-LMBs. This highlights the importance of selecting suitable Li-alloy metals to sustain surface lithiophilicity throughout cycling to regulate dendrite-less Li plating and improve the electrochemical stability of ZE-LMBs. © 2024 Wiley-VCH GmbH.FALSEsciescopu
Hysteresis Compensation of Flexible Continuum Manipulator using RGBD Sensing and Temporal Convolutional Network
No full textFlexible continuum manipulators are valued for minimally invasive surgery, offering access to confined spaces through nonlinear paths. However, cable-driven manipulators face control difficulties due to hysteresis from cabling effects such as friction, elongation, and coupling. These effects are difficult to model due to nonlinearity and the difficulties become even more evident when dealing with long and coupled, multi-segmented manipulator. This paper proposes a data-driven approach based on Deep Neural Networks (DNN) to capture these nonlinear and previous states-dependent characteristics of cable actuation. We collect physical joint configurations according to command joint configurations using RGBD sensing and 7 fiducial markers to model the hysteresis of the proposed manipulator. Result on a study comparing the estimation performance of four DNN models show that the Temporal Convolution Network (TCN) demonstrates the highest predictive capability. Leveraging trained TCNs, we build a control algorithm to compensate for hysteresis. Tracking tests in task space using unseen trajectories show that the proposed control algorithm reduces the average position and orientation error by 61.39% (from to ) and 64.04% (from 31.17 to 11.21), respectively. This result implies that the proposed calibrated controller effectively reaches the desired configurations by estimating the hysteresis of the manipulator. Applying this method in real surgical scenarios has the potential to enhance control precision and improve surgical performance. IEEEFALSEsciescopu
Phosphorylation-mediated disassembly of C-terminal binding protein 2 tetramer impedes epigenetic silencing of pluripotency in mouse embryonic stem cells
Get PDFCells need to overcome both intrinsic and extrinsic threats. Although pluripotency is associated with damage responses, how stem cells respond to DNA damage remains controversial. Here, we elucidate that DNA damage activates Chk2, leading to the phosphorylation of serine 164 on C-terminal binding protein 2 (Ctbp2). The phosphorylation of Ctbp2 induces the disruption of Ctbp2 tetramer, weakening interactions with zinc finger proteins, leading to the dissociation of phosphorylated Ctbp2 from chromatin. This transition to a monomeric state results in the separation of histone deacetylase 1 from Ctbp2, consequently slowing the rate of H3K27 deacetylation. In contrast to the nucleosome remodeling and deacetylase complex, phosphorylated Ctbp2 increased binding affinity to polycomb repressive complex (PRC)2, interacting through the N-terminal domain of Suz12. Through this domain, Ctbp2 competes with Jarid2, inhibiting the function of PRC2. Thus, the phosphorylation of Ctbp2 under stress conditions represents a precise mechanism aimed at preserving stemness traits by inhibiting permanent transcriptional shutdown. © The Author(s) 2024. Published by Oxford University Press on behalf of Nucleic Acids Research.TRUEsciescopu
BurstM: Deep Burst Multi-scale SR Using Fourier Space with Optical Flow
No full textMulti frame super-resolution (MFSR) achieves higher performance than single image super-resolution (SISR), because MFSR leverages abundant information from multiple frames. Recent MFSR approaches adapt the deformable convolution network (DCN) to align the frames. However, the existing MFSR suffers from misalignments between the reference and source frames due to the limitations of DCN, such as small receptive fields and the predefined number of kernels. From these problems, existing MFSR approaches struggle to represent high-frequency information. To this end, we propose Deep Burst Multi-scale SR using Fourier Space with Optical Flow (BurstM). The proposed method estimates the optical flow offset for accurate alignment and predicts the continuous Fourier coefficient of each frame for representing high-frequency textures. In addition, we have enhanced the network’s flexibility by supporting various super-resolution (SR) scale factors with the unimodel. We demonstrate that our method has the highest performance and flexibility than the existing MFSR methods. Our source code is available at https://github.com/Egkang-Luis/burstm. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025
Molecular and cellular basis of sodium sensing in Drosophila labellum
No full textAppropriate ingestion of salt is essential for physiological processes such as ionic homeostasis and neuronal activity. Generally, low concentrations of salt elicit attraction, while high concentrations elicit aversive responses. Here, we observed that sugar neurons in the L sensilla of the Drosophila labellum cf. responses to NaCl, while sugar neurons in the S-c sensilla do not respond to NaCl, suggesting that gustatory receptor neurons involved in NaCl sensing may employ diverse molecular mechanisms. Through an RNAi screen of the entire Ir and ppk gene families and molecular genetic approaches, we identified IR76b, IR25a, and IR56b as necessary components for NaCl sensing in the Drosophila labellum. Co-expression of these three IRs in heterologous systems such as S2 cells or Xenopus oocytes resulted in a current in response to sodium stimulation, suggesting formation of a sodium-sensing complex. Our results should provide insights for research on the diverse combinations constituting salt receptor complexes. © 2024 The Author(s)TRUEsciescopu
Progresses and Perspectives of 1D Soft Sensing Devices for Healthcare Applications
No full textHealthcare sensing devices enable continuous monitoring of diverse biosignals, such as respiration, heartbeat, temperature, inflammation, Electroencephalogram (EEG), and biomechanical movement, contributing to the management of health conditions and early diagnosis of diseases. However, clinically available tools have several limitations for real-time measurement of biosignals in non-hospital environments owing to their cumbersome, complex designs, and rigidity. To address these limitations, there has been a growing body of research that explores flexible and soft electronics for healthcare applications, which feature high mechanical compliance and adaptability. Especially, sensing devices based on fiber structures have attracted significant attention due to high flexibility, lightweight design, effective workspace and unique structural adaptability. Moreover, 1D sensing devices can be seamlessly integrated into garments and the human body with complex structures without any unconformity. In this perspective, The fabrication and electrical functionalization of fiber substrates and focus on recent advances in various fiber-based sensing systems for strain, pressure, temperature, pH, biomarkers, and neural activity is explored. Additionally, biodegradable fiber-based sensing devices as future healthcare technologies are briefly discussed. Finally, this article provides a summary and outlook on the remaining challenges for current fiber-based sensing devices. © 2024 Wiley-VCH GmbH.FALSEsciescopu
Subject-Adaptive Transfer Learning Using Resting State EEG Signals for Cross-Subject EEG Motor Imagery Classification
No full textElectroencephalography (EEG) motor imagery (MI) classification is a fundamental, yet challenging task due to the variation of signals between individuals i.e., inter-subject variability. Previous approaches try to mitigate this using task-specific (TS) EEG signals from the target subject in training. However, recording TS EEG signals requires time and limits its applicability in various fields. In contrast, resting state (RS) EEG signals are a viable alternative due to ease of acquisition with rich subject information. In this paper, we propose a novel subject-adaptive transfer learning strategy that utilizes RS EEG signals to adapt models on unseen subject data. Specifically, we disentangle extracted features into task- and subject-dependent features and use them to calibrate RS EEG signals for obtaining task information while preserving subject characteristics. The calibrated signals are then used to adapt the model to the target subject, enabling the model to simulate processing TS EEG signals of the target subject. The proposed method achieves state-of-the-art accuracy on three public benchmarks, demonstrating the effectiveness of our method in cross-subject EEG MI classification. Our findings highlight the potential of leveraging RS EEG signals to advance practical brain-computer interface systems. The code is available at https://github.com/SionAn/MICCAI2024-ResTL