34254 research outputs found
Sort by
Dynamics of Carbonate System and Net Community Production in the East China Sea Upwelling Regions
No full textAlthough the upwelling region covers less than 3% of the global ocean surface, these regions contributes 10\ue220% of global oceanic production. Owing to the complex interplay of physical and biogeochemical processes, accurately estimating the factors that control the partial pressure of carbon dioxide (pCO2) and the variability of net community production (NCP) in these upwelling regions remains challenging,. To enhance our understanding of these mechanisms, we conducted two research cruises on the southern East China Sea (sECS) to survey sea surface pCO2, NCP, inorganic carbonate parameters, and relevant hydrological variables.
During the decline phase of upwelling (before the typhoon, July 6-9), the stratified surface layer exhibited high sea surface temperatures (SST) and nutrient depletion. In these upwelling regions, the average pCO2 ranged from 322.9 to 325.1 \uce\ubcatm, and NCP averaged 26.6\uc2\ub116.1 mmol C m-2 d-1. During the upwelling development stage (after the typhoon, July 13-17), cold and nutrient-rich Kuroshio subsurface waters were transported into the mixed layer. Simultaneously, surface pCO2 slightly decreased to 305.7-315.5 \uce\ubcatm and NCP increased to 65.8\uc2\ub123.5 mmol C m-2 d-1. Based on our analysis, during the upwelling process, thermal effects and NCP contributed to an increase of 38-40% and a decrease of 33%-36%, respectively, playing crucial roles in modulating pCO2 levels. Additionally, the mixing effect still contributes to 24% of pCO2 variation. During the upwelling process, the entrainment of low-oxygen subsurface water into the mixed layer led to underestimating oxygen concentration and NCP. To address this issue, we utilized the relationship between N2O and Apparent Oxygen Utilization to correct the effects on mixed-layer oxygen levels, refining our estimates of NCP (28%-47%).
This study incorporated in-situ data, model simulations, and a strong linear correlation between dissolved inorganic carbon (DIC) and temperature in the upper ocean of the Kuroshio to predict the variations of surface pCO2 in upwelling regions at under two global warming scenarios (RCP2.6 and RCP8.5). These results suggest that the upwelling regions in the western North Pacific Ocean could potentially act as a global atmospheric CO2 sink under global warming
A particle picking pipeline for cryo-EM using semantic segmentation and conditional random field
No full textOver the past decade, semantic segmentation technology has considerably advanced across various computer vision fields. A crucial application in cryo-electron microscopy (cryo-EM) is particle-picking, where semantic segmentation models have the potential to effectively distinguish particles from the background. However, the low signal-to-noise ratio (SNR) in cryo-EM images poses challenges in establishing an objective, automated process for generating accurate pixel-level protein annotations necessary for training these models.
This study presents a procedural framework for generating binary images from cryo-EM data, serving as ground truth labels for model training. Fully Convolutional Networks (FCN) and DeepLabv3 were utilized for semantic segmentation, trained on our generated labeled dataset. Model performance was evaluated using accuracy, recall, precision, and Intersection over Union (IoU) metrics, and F1-score.
Additionally, we investigated the application of post-processing techniques, including Fully Connected Conditional Random Fields (Fully Connected CRFs or Dense CRFs) and Convolutional Conditional Random Fields (Convolutional CRFs or ConvCRFs), to enhance semantic segmentation results in cryo-electron microscopy (cryo-EM) images. Our experiments revealed that Dense CRF had limited effectiveness in extracting features from cryo-EM images and did not significantly improve model classification. On the other hand, integrating ConvCRF into the models facilitated a more accurate classification of particles labeled in the training set but also introduced anomalies in some boundaries.
As a result, the inclusion of post-processing techniques in current models presents trade-offs, emphasizing the need for further investigation to optimize semantic segmentation techniques in cryo-EM image analysis
Multi-Core Task Scheduling Method Based on Multi-Hop and Preemption Mechanism in Distributed Architecture
No full textAs technology advances, the scope of IoT (Internet of Things) applications expands rapidly, accompanied by increasing task demands. Sending massive data to the cloud for processing entails significant computational time. Moreover, IoT tasks often demand ultra-low latency, posing a challenge for cloud computing due to transmission delays. To enhance task completion rates, edge computing circumvents these cloud computing limitations by processing data closer to the source device.
This study aims to develop offloading strategies and scheduling mechanisms tailored to the distinctive attributes of tasks within the edge computing paradigm, ensuring timely task completion and maximizing rewards. Our framework supports multi-hop technology, facilitating collaboration among edge servers to optimize task execution. Additionally, we have implemented a preemption mechanism to assess the time cost of each task based on its specific characteristics, such as deadline, partitioning, and workloads. These costs are factored into our scheduling considerations as part of the preemption strategy. Given our edge server's multi-core architecture, we must carefully allocate tasks across multiple cores to efficiently meet each task's requirements and prevent unnecessary consumption of computing resources
Factors Influencing Tourist Demand in Thailand: An Analysis Focused on the Role of Corruption
No full textThis study investigates the factors affecting tourist demand in Thailand, with a specific focus on the role of corruption. It employs a panel data approach, encompassing data from 20 tourism source countries over the period 2000-2019. The results show that a 1-point increase in the Corruption Perception Index (indicating a decrease in the level of corruption) results in a 0.8% to 1.1% increase in tourist demand, indicating that corruption plays a very important role in explaining the behavior of inbound tourist demand in Thailand. When evaluating the impact of relative corruption on tourist demand, the findings suggest that relative corruption has a negative impact on tourist demand. Moreover, the effect of relative corruption is greater as the difference of the corruption perception index of the tourist source countries and Thailand increases, indicating that tourists from countries with low levels of corruption tend to visit Thailand less frequently.
The findings of this study support the Sanding the wheels hypothesis, indicating that corruption adversely affects tourist demand. Consequently, highlighting the significance of combating corruption to foster sustainable tourism growth in Thailand
Development of a Remotely Operated System for a Waterjet Propulsion Unmanned Surface Vehicle
No full textUnmanned Surface Vehicles (USVs) showcase capabilities such as remote operation, autonomous navigation, and obstacle avoidance, making them applicable in military, commercial, and academic fields. The current development trends in USVs highlight progress in communication, positioning, object identification, and path following with obstacle avoidance. To address the challenges of extremely shallow water environments, this study introduces the development of a remotely operated system for a water-jet propulsion USV, utilizing the MOKAI ES-Kape watercraft. The original joystick control unit on the MOKAI ES-Kape watercraft is replaced with a single-board computer for remote wireless control. Engine start-stop, nozzle direction adjustment, and throttle control of the USV are executed through long-distance wireless communication. The resulting remotely operated system is divided into the land-based station and the USV onboard system. The land-based component includes a wireless communication system, a central computer, and a human-machine interface. Simultaneously, the USV onboard system encompasses a central controller, propulsion system, communication system, and power system. The central controller on the USV onboard system utilizes a Raspberry Pi single-board computer. Commands from the land-based station are wirelessly transmitted to the USV onboard system through a wireless communication module. The USV\ue2s central controller initiates engine start, stops the engine, manipulates the propulsion system based on the received commands, and relays sensor data back to the land-based station. To remotely verify the successful activation of the USV engine, this study uses an infrared sensor to detect engine rotational speed. Furthermore, the USV\ue2s central controller is equipped with a mechanism to handle wireless communication interruptions, ensuring that the operational USV can be stopped to prevent accidents in case of disruptions in wireless communication
A Study on the Public Dimension of Performing Arts under the Arts Subsidy Mechanism
No full textThe purpose of this study is to reexamine the public nature of performing arts under the award and subsidy mechanism. The research employs a qualitative strategy, using the regular cultural and arts subsidies in Kaohsiung City as a sample, and conducts a two-stage progressive exploration and examination model. In the first stage, a literature re-view method is utilized to review and analyze relevant literature on the public nature of public administration and non-profit organizations. The study explores the potential public nature of performing arts under the cultural subsidy mechanism. In the second stage, an expert in-depth interview method is employed. Through consultations with government subsidy review committee members, the various aspects identified in the first stage are further investigated. The analysis and inference are conducted sequen-tially from four main perspectives. Firstly, the study investigates the reasons for public resource subsidies for performing arts to clarify the government's role in the cultural domain. To ensure fair resource allocation, the research also analyzes the existing sub-sidy evaluation mechanisms, including the qualifications of reviewers, methods of ef-fectiveness assessment, and the selection of evaluation indicators. Under such mecha-nisms, cultural and arts organizations are naturally assigned public responsibilities, in-cluding continuous innovation, cultural heritage, expanding audiences, and enhancing quality of life from diverse perspectives, establishing the public nature of performing arts under the subsidy mechanism.
In addition to aiming to fill the theoretical gap in the public nature of performing arts, the study, through in-depth interviews and expert consultations, also seeks to provide practical recommendations and references for current cultural policy makers. Cultural and arts organizations can gain insights into their public responsibilities under the sub-sidy mechanism to offer more public value in their performance services. For the gen-eral public, ongoing attention to the public nature of performing arts subsidies has the potential to increase awareness and support for performing arts, further promoting so-cial participation in the performing arts. Future research is suggested to expand the di-versity of interview subjects, conduct long-term tracking studies, and engage in cross-regional comparative research to gain a deeper understanding of the development and changes in performing arts subsidy systems. Practical recommendations include providing guidelines for cultural and arts organizations to fulfill public responsibilities and delving into the internal organizational structure and management of these organi-zations to better comprehend the substantive impact of subsidies on the public nature of performing arts
An Adaptive Protection Method for Approximate DRAM and Its Application to LeNet Neural Network as A Case Study
No full textApproximate memory is a solution for power consumption bottlenecks in edge devices, often used in applications that inherently have error tolerance. It maintains acceptable output quality even when partial errors are present in memory. However, the energy efficiency of refresh power saving in Dynamic Random Access Memory (DRAM) tends to saturate as the error rate increases. To further reduce memory power consumption, many theses have proposed protection methods to increase the error tolerance of memory data. When setting the approximation level of approximate DRAM at the design stage, operating the refresh interval at the longest interval tolerable by the target application can achieve the highest error tolerance. If the operating temperature of the DRAM increases or it is affected by aging, the number of retention time errors that occur under a specific refresh interval may exceed expectations, posing a threat to reliability. However, simply enhancing the effectiveness of protection methods can lead to unnecessary power consumption costs under different levels of reliability threats.
Therefore, this thesis, based on the neural network weight data protection methods proposed in past thesis, explores the protection effectiveness of zero-masking combined with error correction codes of different correction abilities and the problem of DRAM refresh power saving saturation. We also demonstrate with hardware power analysis that simply increasing the correction ability of error correction codes in the single protection method of zero-masking, as mentioned in past thesis, does not further reduce summed power consumption. Considering this issue and reliability, we propose an adaptive protection method. Using the LeNet network as a case study, we compare it with the single-mode protection method of zero-masking combined with error correction codes proposed in past thesis. Under conditions where both consider a maximum of double error correction and the same tolerable error quantity, our developed method, compared to the summed power consumption including refresh power in past thesis, can save 19.7%. Additionally, our developed technique supports user configuration for different networks offline. Apart from the adaptive multi-mode protection method, it also supports a single protection mode, offering high usage flexibility.
To realize the adaptive protection method, this thesis proposes a BCH LUT-based decoder that can adjust error correction capabilities, offering three different error quantity correction modes: No ECC (No Error Correction Code), SEC (Single Error Correction), and DEC (Double Error Correction). Notably, when no additional error correction is needed, we turn off part of the hardware to save significant power. When the encoded data error rate is 8\uc310^(-3), the power consumption of our developed decoder, compared to traditional decoders, saves 82.5% in No ECC mode, 37.43% in SEC mode, and the cost in DEC mode is only 10.98%
UN Members\ue2 Positions on Russia-Ukraine War Resolutions
No full textThe 11th Emergency Special Session of the United Nations General Assembly has recently adopted six resolutions addressing the Russia-Ukraine War. However, the voting behavior among UN members reveals a notable inconsistency. This study seeks to investigate the impact of interstate strategic relationships, national geopolitical security, regimes, and national economic interests on the voting behavior of UN members, employing the perspectives of realism and liberalism. Utilizing a quantitative approach, specifically regression analysis, the study explores the positive or negative correlations among five explanatory variables in relation to the level of voting support for resolutions condemning and sanctioning Russia. These variables include the strategic relationship with Russia, the strategic relationship with the U.S., geographical proximity to Russia, the level of democracy, and the level of economic interdependence with Russia. Moreover, the study delves into a detailed analysis of each resolution\ue2s content to provide a nuanced understanding of voting behavior among UN members. The research findings indicate that all five factors mentioned above play a role in influencing decision to vote for condemning and imposing sanctions against Russia. Notably, democracy emerges as the most significant and consistent factor, followed by the strategic relationships with the U.S. and Russia. Furthermore, the study reveals that the content of resolution itself serves as a potential element influencing the voting behavior of UN members. This comprehensive analysis illuminates the intricate dynamics and multifaceted considerations in the decision-making of UN members within the United Nations General Assembly in response to the Russia-Ukraine War
Synthesis and Application of Conjugated Microporous Polymers as Efficient Electrodes for Energy Storage
No full textAbstract
This thesis is centered around two primary themes. The first theme revolves around the construction of innovative conjugated microporous polymers (CMPs) for use in supercapacitor applications (Chapter 3 and 4). The second theme explores the creation of microporous carbonaceous materials derived from conjugated polymers, with a focus on their testing as effective electrodes for energy storage (Chapter 5 and 6).
CMPs are a subset of POPs distinguished by their extended \ucf-conjugation, diverse building units, and substantial incorporation of carbon, nitrogen, and non-metallic elements. As a result, they exhibit notable features such as large specific surface areas, adjustable porosities, and exceptional thermal and chemical stability, making them highly versatile for a wide spectrum of potential applications. In chapter 3, a novel series of carbazole-containing CMPs with redox activity were constructed through Suzuki coupling polymerization. The as-synthesized polymers possessed a great surface area (623 m2 g\ue21) and large pore size (up to 0.49 nm). The electrochemical results were measured using CV and GCD techniques. The data revealed that the Cz-TPA polymer was the excellent electrode which featured a great specific capacitance of 271.82 F g\ue21 at 0.5 A g\ue21 and superior cycling retention (99.87 %) over 2000 cycles at 0.5 A g\ue21. In chapter 4, two dimensional redox-active pyrenetetraone-based CMPs\ue2BC-PT and TPA-PT\ue2have been successfully synthesized via [4+2] Suzuki coupling condensations. Notably, the BC-PT CMP exhibited the best capacitance of 530 F g\ue21 and energy density of 73.61 W h kg\ue21 in 1.0 M KOH in three electrodes system at 1.0 A g\ue21 and 500 W kg\ue21 current density and power density, respectively. Interestingly, a two-electrode coins SC holding BC-PT CMP displayed a superb electrochemical capacitance of 250 F g\ue21 at 1.0 A g\ue21, a considerable retention of 95.12% after 5000 cycles at 10 A g\ue21, and a higher energy density up to 34.72 W h kg\ue21.
Furthermore, numerous porous carbonaceous materials have been engineered due to their expansive surface area, impressive chemical, thermal, and mechanical robustness, as well as their exceptional electrical conductivity. These intriguing attributes have paved the way for their utilization across diverse applications, including but not limited to gas adsorption, catalysis, storing energy, pollutant removal, fuel cells, and electromagnetic shielding. In chapter 5, we have conducted Friedel-Craft polymerization, employing a cyanuric chloride building block and tetraphenylcyclopentadiene-holding as linkage monomers, to create innovative tetraphenylcyclopentadiene-based polymeric materials. Subsequently, these materials were subjected to high-temperature thermal treatment to transform them into porous carbons. According to the applied electrochemical tests, The CP-TCT-700 exhibited a remarkable performance with capacitance of 83 F g\ue21 and energy density of 11.65 W h kg\ue21 at an operating current of 1.0 A g\ue21. In addition, it demonstrated good stability (88 %) over 5000 GCD cycles. In chapter 6, two novel triazatruxene-based CMPs\ue2TAT-Cz and TAT-BCz\ue2have been synthesized via conventional Suzuki coupling reaction. Then, these porous polymers underwent high-temperature carbonization followed by KOH activation to convert them into microporous carbons. Remarkably, the TAT-BCz-800 carbons demonstrated the highest capacitance of 723 F g\ue21 and an energy density of 100.61 W h kg\ue21 in a 1.0 M KOH solution at 1.0 A g\ue21 and a power density of 500 W kg\ue21. Moreover, TAT-BCz-800 exhibited superb stability over a span of 5000 cycles at 10 A g\ue21 reaching 98.89%
Exploring the Visual Development Design Analysis of Network Management System - Taking the Military IP-PBX System as the Case Study
No full textThe common operation screens of network management systems are mostly based on list-type function menus, command lines, and column-type system information. The proportion of chart content is low, and it is difficult for non-professional network management personnel to understand the information and status displayed by the system. The purpose of this study is to help the National Army Network Private Switch System develop and design a visual common operation screen dashboard, and use the expert interview method to gradually analyze and confirm the appropriate network management system functions for this case, and plan and design the system requirements Function list, with the focus of information visualization, develop and design a prototype of common operation screen dashboard; the dashboard presented after the network management system information is visualized can integrate, analyze and process a large amount of system background data, which can be clearly understood through charts It can present important statistical information such as equipment alarms, network topology status, system usage efficiency, and overall equipment availability rate, so that network management personnel and senior managers can quickly understand and grasp the real-time important information of the system and the overall availability rate of the system equipment. , the connection status of each computer room, the usage status of important users, etc.
This study is based on the past literature on information system visualization research, which only focused on effectiveness, configuration design, price, risk and introduction time. It mainly converts the back-end database of the network management system into a front-end user interface dashboard. The development and design steps add different research directions. Instead of focusing on the development of system users' needs and system function orientation in the past, we add the perspective of senior managers and possible needs, aiming at the overall planning of senior management. The management perspective and the corresponding display of system performance are taken into consideration to enhance the trust of senior managers in the effectiveness of the network management system