GIST Scholar
Not a member yet
30271 research outputs found
Sort by
Multi-fidelity sub-label-guided transfer network with physically interpretable synthetic datasets for rotor fault diagnosis
No full textThe development of deep-learning-based diagnostic models for prognostics and health management relies on data from the target system. However, when the system does not exist (e.g., during the design phase), acquiring health data is impossible, making accurate fault diagnosis models difficult to develop. To resolve this challenge, this study proposes a fault diagnosis framework for rotating machinery that uses physically interpretable synthetic data generation and sub-label-guided transfer learning. An empirical model based on domain knowledge is used to generate large quantities of low-cost, low-fidelity data, although system configuration and expected operating condition information are incorporated into multi-body dynamics simulations to create a small amount of accurate, high-fidelity data. Low-fidelity data are labeled with a primary label that represents dominant fault modes and with sub-labels corresponding to detailed categories of faults, whereas high-fidelity data are assigned only a primary label for the dominant fault mode. Multi-fidelity data are preprocessed to create two-dimensional orbit plots and one-dimensional handcrafted features, which are then used as inputs for the transfer learning model. To effectively train the model, while considering the correlation between multi-fidelity data, this study designs a sub-label-guided transfer network (SlgTN) that combines sub-label guidance with transfer learning. In the pretraining process, feature learning is performed using low-fidelity data, with two parallel classification layers designed to effectively learn both the primary label and the sub-label. Center loss and cross entropy loss are incorporated. Experimental results from applying the method to a Bently-Nevada rotor kit demonstrate the effectiveness and validity of the proposed method. © 2025 Elsevier LtdFALSEsciescopu
Continuous-Flow Adsorption With Imidazolium-Based Porous Organic Polymers Capable of Selective Removal of Chromate Contaminants From Water
No full textN-rich triazine core functionalized imidazolium-based porous organic polymers (iPOPs) are developed with exceptional oxoanions sequestration capabilities. Two iPOPs, iPOP-1 and iPOP-2, are synthesized via a facile Radziszewski multicomponent reaction and designed with multiple interaction sites for the electrostatically driven adsorption of toxic oxoanion, chromate (CrO4 2-). The strategic incorporation of methylene spacers in iPOP-1 resulted in a structurally flexible framework, enabling ultrafast capture kinetics with 99% removal of CrO4 2- from a 40 ppm aqueous solution within just 15 sec. In contrast, the structurally rigid iPOP-2, lacking methylene spacer units, exhibits significantly reduced performance, achieving only 25% removal even after 300 seconds of contact time. Notably, iPOP-1 maintained remarkable selectivity, effectively capturing CrO4 2- even in the presence of a 200-fold excess of competing anions (Cl-, Br-, NO3 -, SO4 2-, CO3 2-, H2PO4 -). To evaluate the practical application potential of iPOP-1, continuous-flow CrO4 2- sequestration experiments are conducted using both glass-column chromatography and packed-bed reactor configurations, successfully reducing CrO4 2- concentrations to below 40 ppb, well within World Health Organization (WHO) drinking water safety thresholds. Additionally, iPOP-1 demonstrate excellent recyclability, maintaining high sequestration efficiency through five regeneration cycles. These results offer a robust, selective, and reusable approach for the efficient removal of hazardous oxoanions from industrial wastewater.FALSEsciescopu
Synergistic Effects of Seed Endophytic Bacterial Consortia from Lactuca serriola on Phosphorus Solubilization
No full textPlants have evolved in association with plant-beneficial bacterial endophytes. Some of these bacteria live inside seed, known as seed endophytes, are vertically transmitted to the next generation and contribute to early seedling development. In this study, we aimed to assess the plant growth-promoting (PGP) potential of these seed endophytes. To better simulate natural conditions, we constructed bacterial consortia to evaluate their synergistic effects. From the seed endophytes of Lactuca serriola, designated as a harmful non-indigenous plant in South Korea, we selected seven bacterial strains. These strains and their consortia were screened for four key PGP traits. Among them, two strains that played a key role in consortium function were selected and co-inoculated with different isolates in L. serriola to assess their impact under phosphorus deficiency. The Kosakonia cowanii SD1 and Pantoea dispersa SD25 consortium, and Xanthomonas spp. SD2 and Stenotrophomonas maltophilia SD8 consortium enhanced soil phosphorus availability. Notably, these two consortia displayed contrasting results in in vitro assays, suggesting that these combinations utilize different mechanisms to enhance soil phosphorus availability. These findings provide valuable insights into the role of seed endophytes in the adaptability and ecological success of non-indigenous plants, contributing to a better understanding of invasive plant management
Light-Weight Vision Language Model Guided Gesture Recognition Based on Electromyography
No full textNeuromuscular dysfunction poses a critical health challenge, affecting both patients and caregivers. Current active rehabilitation devices relying on electromyography (EMG) face significant hurdles due to high data acquisition demands, inaccurate decoders, and model degradation from sensor distribution shifts across participants. We address these limitations with a vision language model (VLM) that pseudo-labels motor-intent predictions, enhancing decoding accuracy and reliability under sensor distribution shift across participants. Our proposed pipeline integrates visual information with EMG decoding by employing a parallel processing framework: vision data is processed through the VLM to determine user intent, object and grasp type, which in turn provides pseudo-labels for the EMG decoder’s output. The pseudo-label is then utilized to infer motor-intent, enabling precise external control of assistive devices. In our experimental analysis, our proposed pipeline demonstrated superior performance compared to an EMG-only decoder, achieving 100% decoding accuracy across four upper limb gestures in four participants despite sensor distribution shift across them. Our approach underscores the potential of integrating visual understanding with neural signal analysis to enhance the reliability and effectiveness of active rehabilitation devices, ultimately improving patient outcomes by providing more accurate and consistent assistive technologies. © 2001-2012 IEEE.sciescopu
Observer: An open-source framework for automating spectator for Real-time Strategy game of StarCraft
No full textSelecting engaging scenes is a critical component of esports broadcasting, traditionally performed by human observers. While recent research has explored AI-based automation, existing approaches often lack comprehensive frameworks for data extraction, human behavior modeling, and interface integration. We present Observer , an open-source framework that collects and preprocesses raw in-game data from StarCraft along with human observer viewport data to train AI-based automatic observers. The system transforms gameplay into multi-channel (hereafter, feature channels) representations and uses a modified Intersection over Union (IoU) metric to evaluate the overlap between predicted and aggregated human viewports. As reported in prior work, a learned observer achieves 56.9% similarity to human behavior, surpassing representative rule-based methods (52.4% and 49.1%) on standard benchmarks. In this software paper, we focus on a standardized, reproducible pipeline and system-level metrics. © 2025 Elsevier B.V., All rights reserved.TRUEsciescopu
Rapid urea decomposition via bromination for ultrapure water production: New insights into pH-dependent reaction kinetics and mechanisms compared to chlorination
No full textUrea, a persistent organic contaminant, poses significant challenges in ultrapure water production due to its resistance to conventional treatment processes. This study comprehensively investigated the reaction kinetics and mechanisms of urea bromination, with a comparative analysis of urea chlorination. Bromination proved significantly faster than chlorination, with optimal urea removal occurring at pH 10.5 for bromination and pH 6.0 for chlorination. The initial halogenation of urea to mono-halogenated urea was identified as a key step in facilitating total urea decomposition. Successive halogenation decreased the pKa of halogenated ureas, enhancing their deprotonation and promoting further transformation via hydrolysis. Acidic conditions favored the first halogenation step, while neutral to alkaline conditions facilitated multi-halogenation and subsequent mineralization. Complete urea decomposition required three bromine molecules at pH ≥ 6, whereas four to eight bromine molecules at pH ≤ 5, with bromine consumption increasing under acidic conditions. Treatment efficiency improved with higher oxidant-to-urea molar ratios but declined as initial urea concentrations decreased. Elevated oxidant dosages were necessary for rapid urea removal at ppb levels, typical for ultrapure water production. Urea reactions with bromine and chlorine produced halogenated ureas, which subsequently underwent hydrolysis and/or further halogen attack, forming haloamines, NH4+, NO2−, NO3−, N2, N2O and CO2. This study presents the first detailed comparison of pH-dependent kinetic and mechanistic differences between urea bromination and chlorination, revealing the distinct and superior efficiency of bromination at µg/L levels and new insights into mineralization pathways. The findings highlight the potential of free bromine as a more reactive and cost-effective alternative for water treatment. © 2025 Elsevier B.V.FALSEsciescopu
Thermochromic Gires-Tournois Resonators with Tellurium for Battery Thermal Runaway Warning
No full textEffective temperature monitoring is crucial for preventing battery fires caused by thermal runaway, ensuring human safety, and providing timely warnings. While thermochromic materials offer intuitive, real-time temperature visualization, their slow response times remain them unsuitable for battery monitoring. A thermochromic Gires-Tournois (GT) resonator specifically designed for rapid and accurate battery temperature detection in the critical range below 80 degrees C is introduced, where thermal runaway risks can be effectively mitigated. Central to this design is an ultrathin (10 nm) thermo-responsive tellurium film, paired with a protective glass layer and an underlying metallic mirror. This thermochromic GT resonator exhibits reversible temperature detection over multiple cycles, actively responding to temperature changes through partial melting of tellurium, which alters its complex refractive index-a property discovered in the 1960s but now harnessed for this novel application. Notably, the resonator monitors both specific temperature points and overall heat transfer across the battery surface, achieving sub-second response times in an untethered manner. These findings position the thermochromic GT resonator as a promising platform for direct, intuitive, and compact temperature monitoring in energy storage systems.TRUEsciescopu
Integrating reductants and organic ligands for improved Mn recovery and CO₂ sequestration from electric arc furnace slag
No full textElectric arc furnace (EAF) slag exhibits great potential in carbon mineralization and resource recovery due to its high calcium (Ca) and manganese (Mn) content. In particular, a pH swing-assisted carbonation process has been recognized as a plausible option to simultaneously recover valuable resource (Mn) and sequester CO2 via reactions with Ca forming calcium carbonate (CaCO3). However, the presence of insoluble Mn3 + often poses a significant challenge, leading to a low leaching efficiency for Mn. This study investigates the effects of reductants, including hydrogen peroxide (H₂O₂), ascorbic acid (C₆H₈O₆), and sodium sulfite (Na₂SO₃), in enhancing Mn leaching efficiency by effectively converting Mn3+ into its more leachable Mn²⁺ form. Additionally, the ligand effects of biogenic volatile organic acids, including acetic acid (C₂H₄O₂), propionic acid (C₃H₆O₂), butyric acid (C₄H₈O₂), and valeric acid (C₅H₁₀O₂), were evaluated to improve the overall leaching efficiency of Ca and other major elements, as well as Mn. The synergistic effect of reductants and organic acids resulted in superior leaching efficiencies (∼100 %) for magnesium (Mg), aluminum (Al), Mn, and Ca. The pH swing process utilizing this synergistic effect achieved great Mn recovery and obtained solid residue included an Mn concentration of up to 43.2 wt% at pH 9. Following Mn recovery, carbonation was conducted using the residual leachate, resulting in the formation of high-purity CaCO₃ (93 %), with a CO₂ storage capacity of 118 kg/ton slag. These findings highlight the potential of EAF slag as a sustainable Mn source while simultaneously contributing to CO2 storage.FALSEsciescopu
Regeneration of dialysis solution by dual-layer hollow fiber mixed matrix membrane (DLHF-MMM) incorporated with amine-functionalized mesoporous silica nanoparticles
No full textA large amount of purified water is used in conventional hemodialysis (HD) for treating end-stage kidney disease (ESKD). To minimize the water demand and waste generation, the regeneration of dialysis solution is considered the most efficient control strategy. In this study, an innovative dual-layer hollow fiber (DLHF) mixed matrix membrane (MMM) incorporated with amine-functionalized mesoporous silica nanoparticles (MPS-NPs) was developed to regenerate spent dialysis solution. The fabricated DLHF-MMM configuration enabled the continuous removal of small, medium, and large weight uremic toxins (UTs) through dual mechanisms. The inner layer composed of polyethersulfone (PES) and polyethylene glycol (PEG) rejected medium-large weight UTs (i.e., MW > 500 Da) via the molecular sieving. Meanwhile, the outer layer containing amine-functionalized MPS-NPs effectively removed small weight UTs, such as urea and creatinine. The DLHF-MMM with 6 wt% of amine-functionalized MPS-NPs demonstrated the most favorable characteristics, i.e., high water permeability (298.6 ± 3.2 mL/m2.h.mmHg) and adsorption capacity of urea (523.5 mg/g) and creatinine (28.1 mg/g). Notably, the optimal membrane (DLHF-4) also achieved favorable removal rates from the spent dialysis solution of actual patient, i.e., urea (74.4 %), creatinine (56 %), hippuric acid (16.1 %), and lysozyme (58.7 %, additionally spiked as a mimicking for β-2 microglobulin). These results indicate that the fabricated DLHF-MMM in this study can effectively overcome the challenges posed by the complex matrix components. Overall, the results of this study demonstrate that the DLHF-MMM incorporated with amine-functionalized MPS-NPs is a promising and potential tool for the regeneration of dialysis solution. Furthermore, this approach can contribute to water conservation and reduce the burden on wastewater treatment processes associated with wastewater generated from conventional HD. © 2025 Elsevier LtdFALSEsciescopu
Junsuk M oon Grid Defined Lane Detection
No full textLane lines contain a lot of information beyond informing the direction of the vehicle's progress when driving on the road. The direction of driving of other vehicles and furthermore, traffic rules have meaning. So it is very important to detect it. In autonomous vehicles, lane detection is a task of providing essential information for vehicle control. Lane detection is used for vehicles to locate themselves on the road and maintain the correct path. However, lanes occupy a very small part of the overall image and have a wide variety of forms depending on the road environment. Detection is often difficult depending on weather conditions and the surrounding environment. In this paper, we propose a method to detect the overall lane by overcoming these challenges and restoring even the distorted part. This pipeline, which creates multiple lane candidates and allows each lane candidate to find the actual shape of the lane, consists of four modules: lane point regressor, lane point classifier, lane set suppressor, and lane set classifier. Through this pipeline, we introduce a novel methodology in which lane candidates correspond to all lanes in the image by selecting a label lane to correspond to by L1 distance calculation and having a shape that matches the label lane.MasterList of contents
Abstract i
List of contents ii
List of tables iv
List of figures v
I. INTRODUCTION 1
1. 1. Research background 1
1. 2. Previous research 1
1. 3. Proposed model 2
1. 4. Related works 3
1. 4. 1 Visual geometry approch 3
1. 4. 2. Lane Segmentation approch 3
1. 4. 3. Anchor based approch 3
II. DATASET 4
2. 1. TuSimple 4
III. Approch 5
3. 1. Model Pipleline 5
3. 1. 1. Lane representation 5
3. 1. 2. Points and set 6
3. 1. 3. Pipeline abstract 6
3. 2. Lane Point Regressor 8
3. 2. 1. Model 8
3. 2. 2. Label Lane Assignment 8
3. 2. 3. Training details 9
3. 3. Lane Point Classifier 11
3. 3. 1. Model 11
3. 3. 2. Classification of set points and labeling 11
3. 3. 3. Training details 12
3. 4. Lane Point Suppressor 13
3. 4. 1. Model 13
3. 4. 2. Overlapping lane candidate 14
3. 4. 3. Best lane selection 15
3. 5. Lane Set Classifier 15
3. 5. 1. False lane 15
3. 5. 2. False lane dataset 16
3. 5. 3. Model 17
3. 5. 4. Training details 18
IV. Experiment 19
4. 1. TuSimple 19
4. 1. 1. Data augmentation 19
4. 1. 2. Pipleline output 20
4. 1. 3. Effect of data augmentation 21
4. 1. 4. Inference 22
4. 1. 5. TuSimple Metric 24
V. Ablation Study 25
5. 1. Ablation for data augmentation 25
5. 2. Ablation for number of lane candidate 25
5. 3. Ablation for size of path 26
VI. Future Work & Conclusion 27
Summary 28
References 2