检索结果-内蒙古大学图书馆

A Novel CAPTCHA Recognition System Based on Refined Visual Attention

Computers, Materials & Continua 2025年第4期83卷 115-136页

作者： Zaid Derea Beiji Zou Xiaoyan Kui Monir Abdullah Alaa Thobhani Amr Abdussalam School of Computer Science and Engineering Central South UniversityChangsha410083China College of Computer Science and Information Technology Wasit UniversityWasit52001Iraq Department of Computer Science and Artificial Intelligence College of Computing and Information TechnologyUniversity of BishaBisha67714Saudi Arabia Electronic Engineering and Information Science Department University of Science and Technology of ChinaHefei230026China

Improving website security to prevent malicious online activities is crucial,and CAPTCHA(Completely Automated Public Turing test to tell Computers and Humans Apart)has emerged as a key strategy for distinguishing human users from automated ***-based CAPTCHAs,designed to be easily decipherable by humans yet challenging for machines,are a common form of this ***,advancements in deep learning have facilitated the creation of models adept at recognizing these text-based CAPTCHAs with surprising *** our comprehensive investigation into CAPTCHA recognition,we have tailored the renowned UpDown image captioning model specifically for this *** approach innovatively combines an encoder to extract both global and local features,significantly boosting the model’s capability to identify complex details within CAPTCHA *** the decoding phase,we have adopted a refined attention mechanism,integrating enhanced visual attention with dual layers of Long Short-Term Memory(LSTM)networks to elevate CAPTCHA recognition *** rigorous testing across four varied datasets,including those from Weibo,BoC,Gregwar,and Captcha 0.3,demonstrates the versatility and effectiveness of our *** results not only highlight the efficiency of our approach but also offer profound insights into its applicability across different CAPTCHA types,contributing to a deeper understanding of CAPTCHA recognition technology.

关键词： Text-based CAPTCHA recognition refined visual attention web security computer vision

来源：评论

学校读者我要写书评

暂无评论

Optimally arranged TiO_(2)@MoS_(2) heterostructures with effectively induced built-in electric field for high-performance lithium-sulfur batteries

引用

Journal of Energy Chemistry 2023年第8期83卷 496-508,I0012页

作者： Jeongyoub Lee Changhoon Choi Jung Been Park Seungho Yu Jinho Ha Hyungsoo Lee Gyumin Jang Young Sun Park Juwon Yun Hayoung Im Subin Moon Soobin Lee Jung-Il Choi Dong-Wan Kim Jooho Moon Department of Materials Science and Engineering Yonsei UniversitySeoul 03722Republic of Korea School of Civil Environmental and Architectural EngineeringKorea UniversitySeoul 02841Republic of Korea Energy Storage Research Center Korea Institute of Science and Technology(KIST)Seoul 02792Republic of Korea School of Mathematics and Computing(Computational Science and Engineering) Yonsei UniversitySeoul 03722Republic of Korea

To overcome the serious technological issues affecting lithium-sulfur(Li-S) batteries,such as sluggish sulfur redox kinetics and the detrimental shuttle effect,heterostructure engineering has been investigated as a strategy to effectively capture soluble lithium polysulfide intermediates and promote their conversion reaction by integrating highly polar metal oxides with catalytically active metals ***,to fully exploit the outstanding properties of heterostructure-based composites,their detailed structure and interfacial contacts should be designed ***,optimally arranged TiO_(2)and MoS_(2)-based heterostructures(TiO_(2)@MoS_(2)) are fabricated on carbon cloth as a multifunctional interlayer to efficiently trap polysulfide intermediates and accelerate their redox *** to the synergistic effects between TiO_(2)and MoS_(2)and the uniform heterointerface distribution that induces the ideally oriented built-in electric field,Li-S batteries with TiO_(2)@MoS_(2)interlayers exhibit high rate capability(601 mA h g^(-1)at 5 C),good cycling stability(capacity-fade rate of 0.067% per cycle over 500 cycles at2 C),and satisfactory areal capacity(5.2 mA h cm^(-2)) under an increased sulfur loading of 5.2 mg cm^(-2).Moreover,by comparing with a MoS_(2)@TiO_(2)interlayer composed of reversely arranged heterostructures,the effect of the built-in electric field’s direction on the electrocatalytic reactions of polysulfide intermediates is thoroughly investigated for the first *** superior electrocatalytic activities of the rationally arranged TiO_(2)@MoS_(2)interlayer demonstrate the importance of optimizing the built-in electric field of heterostructures for producing high-performance Li-S batteries.

关键词： Lithium-sulfur batteries Shuttle effect TiO_(2)-MoS_(2)heterostructure engineering Built-in electric field Multifunctional interlayers

来源：评论

学校读者我要写书评

暂无评论

Pre-Coded Inter-Satellite Routing Algorithm with Load Balancing for Mega-Constellation Networks

引用

Space(science & Technology) 2024年第1期4卷 465-478页

作者： Chi Zhang Quan Chen Zuping Tang Jiaolong Wei Gaoyang Liu School of Electronic Information and Communication Huazhong University of Science and TechnologyWuhanChina College of Aerospace Science and Engineering National University of Defense TechnologyChangshaChina School of Computing Science Simon Fraser UniversityBritish ColumbiaCanada

In order to achieve global multiple seamless coverage,space-based internet usually adopts low Earth orbit(LEO)mega-constellation networks structure,which has the characteristics of high network topology dynamics,limited on-board computing and storage capacity,and uneven distribution of ground *** features may cause problems such as high transmission delay,network congestion and link *** a stable,efficient and balanced satellite communication link can effectively alleviate the performance of the transmission delay,load balancing,and network *** advantage of the regularity of network topology,a pre-coded inter-satellite routing algorithm with load balancing is proposed,which includes 3 parts:(a)the routing sequence coding method and the concept of gateway satellite Service Region(GSSR)are proposed;(b)the initial routing sequence of GSSR is generated based on the maximum network flow method under the ideal situation of uniform satellite traffic distribution;(c)aiming at the uneven distribution of traffic,the Sinkhorn algorithm is used to improve the load balancing performance of inter-satellite *** results show that,for the Starlink Group-4 constellation,the proposed method can maintain a low transmission delay and improve the load balancing together with the network throughout performance with minimal hops and low time complexity.

关键词： mega constellation networks satellite communication link load balancing space based internet global coverage inter satellite communication pre coded routing low Earth orbit

来源：评论

学校读者我要写书评

暂无评论

Machine Learning Approach for Handling Imbalanced Students' Performance Data 15

Machine Learning Approach for Handling Imbalanced Students' ...

引用

15th International Conference on Advances in computing, Control, and Telecommunication Technologies, ACT 2024

作者： Sujatha, E. Divya, S. Sakthivelan, R.G. Gopirajan, P.V. Thandalam Chennai India Chennai India Department of Computer Science and Engineering Gopalan College of Engineering and Management Bengaluru India Department of Computational Intelligence School of Computing SRM Institute of Science and Technology Kattankulathur campus Chennai India

ISBN: (纸本)9798331300579

Imbalanced student performance data in educational institutions is crucial for any machine learning prediction model. It affects the efficiency of classifiers and challenges the sampling methods for having a more significant number of features. The proposed model was designed to predict the student's performance by comparing the results with popular similar models. The Kaggle dataset having 386 rows and 33 columns of student data, was used in this study. In addition, the Portugal database was considered for experimental analysis, containing 394 students and 19 attributes. It proves that the Random Forest classifier yields the highest accuracy at 84.15% in handling imbalanced datasets. The results show that the SVM-SMOTE is higher accuracy as, 94.76%, than the other sampling methods in predicting the student's performance with various features. © Grenze Scientific Society, 2024.

关键词： Adversarial machine learning

来源：评论

学校读者我要写书评

暂无评论

Fragmentation dynamics of electron-impact double ionization of helium

引用

Chinese Physics B 2023年第6期32卷 9-15页

作者：刘士炜叶地发刘杰 Beijing Computational Science Research Center Beijing 100193China Laboratory of Computational Physics Institute of Applied Physics and Computational MathematicsBeijing 100088China Graduate School China Academy of Engineering PhysicsBeijing 100193China HEDPS Center for Applied Physics and Technologyand College of EngineeringPeking UniversityBeijing 100871China

We study the double ionization dynamics of a helium atom impacted by electrons with full-dimensional classical trajectory Monte Carlo simulation. The excess energy is chosen to cover a wide range of values from 5 e V to 1 ke V for comparative study. At the lowest excess energy, i.e., close to the double-ionization threshold, it is found that the projectile momentum is totally transferred to the recoil-ion while the residual energy is randomly partitioned among the three outgoing electrons, which are then most probably emitted with an equilateral triangle configuration. Our results agree well with experiments as compared with early quantum-mechanical calculation as well as classical simulation based on a two-dimensional Bohr's model. Furthermore, by mapping the final momentum vectors event by event into a Dalitz plot,we unambiguously demonstrate that the ergodicity has been reached and thus confirm a long-term scenario conceived by Wannier. The time scale for such few-body thermalization, from the initial nonequilibrium state to the final microcanonical distribution, is only about 100 attoseconds. Finally, we predict that, with the increase of the excess energy, the dominant emission configuration undergoes a transition from equilateral triangle to T-shape and finally to a co-linear mode. The associated signatures of such configuration transition in the electron–ion joint momentum spectrum and triple-electron angular distribution are also demonstrated.

关键词： double ionization classical trajectory Monte Carlo simulation Wannier threshold law ergodicity

来源：评论

学校读者我要写书评

暂无评论

Uncertainty-Calibrated Test-Time Model Adaptation without Forgetting

引用

IEEE Transactions on Pattern Analysis and Machine Intelligence 2025年 PP卷 1-14页

作者： Tan, Mingkui Chen, Guohao Wu, Jiaxiang Zhang, Yifan Chen, Yaofo Zhao, Peilin Niu, Shuaicheng South China University of Technology School of Software Engineering China Pazhou Laboratory Guangzhou China XVERSE China National University of Singapore School of Computing Singapore Tencent AI Lab China Nanyang Technological University College of Computing and Data Science Singapore

Test-time adaptation (TTA) seeks to tackle potential distribution shifts between training and testing data by adapting a given model w.r.t. any testing sample. This task is particularly important when the test environment changes frequently. Although some recent attempts have been made to handle this task, we still face two key challenges: 1) prior methods have to perform backpropagation for each test sample, resulting in unbearable optimization costs to many applications;2) while existing TTA solutions can significantly improve the test performance on out-of-distribution data, they often suffer from severe performance degradation on in-distribution data after TTA (known as catastrophic forgetting). To this end, we have proposed an Efficient Anti-Forgetting Test-Time Adaptation (EATA) method which develops an active sample selection criterion to identify reliable and non-redundant samples for test-time entropy minimization. To alleviate forgetting, EATA introduces a Fisher regularizer estimated from test samples to constrain important model parameters from drastic changes. However, in EATA, the adopted entropy loss consistently assigns higher confidence to predictions even when the samples are underlying uncertain, leading to overconfident predictions that underestimate the data uncertainty. To tackle this, we further propose EATA with Calibration (EATA-C) to separately exploit the reducible model uncertainty and the inherent data uncertainty for calibrated TTA. Specifically, we compare the divergence between predictions from the full network and its sub-networks to measure the reducible model uncertainty, on which we propose a test-time uncertainty reduction strategy with divergence minimization loss to encourage consistent predictions instead of overconfident ones. To further re-calibrate predicting confidence on different samples, we utilize the disagreement among predicted labels as an indicator of the data uncertainty. Based on this, we devise a min-max entropy

关键词： Semantic Segmentation

来源：评论

学校读者我要写书评

暂无评论

Comparative Analysis of Transfer Learning Based Deep Learning Models for Brain Tumor Detection

Comparative Analysis of Transfer Learning Based Deep Learnin...

引用

2024 International Conference on Artificial Intelligence and Emerging Technology, Global AI Summit 2024

作者： Singh, Amish Priyadarshi, Prakhar Panwar, Arvind Dubey, Arun Kumar Jain, Achin Bharati Vidyapeeth College of Engineering New Delhi India School of computing science engineering Galgotias University India

ISBN: (纸本)9798350379716

Detecting and classifying brain tumors is essential in medical imaging. Accurate and early diagnosis significantly impacts treatment outcomes. Deep learning (DL) models provide automated and highly accurate diagnostic tools. In this paper, we have used publicly available MRI brain tumor dataset from Kaggle. It evaluates the performance of various advanced DL architectures to classify brain tumors. We used several pre-trained deep learning models, including Inception ResNet V2, Inception V3, ResNet 50, VGG 16, Xception, and VGG 19. Transfer learning was employed to fine-tune these models on the Kaggle MRI brain tumor dataset. Each model was trained and validated to assess its classification accuracy. Transfer learning allowed the models to utilize pre-existing knowledge, enhancing their performance in this medical imaging task. The validation accuracies achieved were as follows: Inception ResNet V2 at 62.39%, Inception V3 at 74.56%, ResNet 50 at 81.34%, VGG 16 at 89.83%, Xception at 70.67%, and VGG 19 at 92.37%. Among these, VGG 19 showed the highest accuracy. It demonstrated superior capability in distinguishing between different types of brain tumors, with an improvement of 6.93% over other models. © 2024 IEEE.

关键词： Contrastive Learning

来源：评论

学校读者我要写书评

暂无评论

Bridging the Semantic Gap: A Deep Learning Framework for Semantic Video Search Integrating Vision and Language for Enhanced Video Retrieval 10

Bridging the Semantic Gap: A Deep Learning Framework for Sem...

引用

10th International Conference on Communication and Signal Processing, ICCSP 2024

作者： Annamalai, J. Arjun Sneha, S. Sharon Ashokkumar, C. Srm Institute of Science and Technology School of Computing College of Engineering and Technology Department of Computing Technologies Chennai Kattankulathur603203 India

ISBN: (纸本)9798350353068

Traditional video search engines often rely on tags or manual annotations for content retrieval, limiting the accuracy and efficiency of search results. Moreover, keyword-centric searches may not adeptly capture the nuanced and intricate queries users pose when seeking specific video content. The envisioned video search system integrates machine learning and natural language processing components to enable efficient and effective video retrieval based on user queries. The pipeline includes video processing employing Vision Transformer with GPT-2 (ViT-GPT2) architecture to analyze intricate details within video frames, Speech-to-Text (STT) models for transcribing spoken content, and the combination of transcriptions with video captions to form textual descriptions. These descriptions are then embedded using BERT, a transformer-based model, to grasp contextual relationships. Semantic matching is achieved through deep learning models, and a ranking mechanism based on similarity scores facilitates efficient retrieval. Leveraging datasets such as the MSR-VTT dataset, spanning visual recognition domains, contributes to the evaluation and training of the system's components. The system's efficacy is evaluated using metrics such as precision, recall, F1-score, and ranking metrics, with datasets spanning visual recognition, speech transcription, and semantic similarity domains. The comprehensive evaluation approach ensures the robustness and relevance of the video search system, aligning with user expectations and real-world use cases. © 2024 IEEE.

关键词： Semantics

来源：评论

学校读者我要写书评

暂无评论

Toward a Framework to Improve Employees’ Compliance with Cybersecurity Policy in Organizations 8th

Toward a Framework to Improve Employees’ Compliance with Cy...

引用

8th International Congress on Information and Communication Technology, ICICT 2023

作者： Thamae, Reneuoe Abdullah, Hanifa Mujinga, Mathias School of Computing College of Science Engineering and Technology University of South Africa Pretoria South Africa

ISBN: (纸本)9789819930425

Digital transformation has influenced organizations’ operations significantly. However, non-compliance with cybersecurity policy (CSP) is a growing concern for organizations. Technology alone cannot protect organizational cyber assets such as computer systems, networks, and data. Human aspects should be considered when designing and implementing a CSP. The lack of effective CSP training and awareness programs (CSPTAP) is attributable to employees’ non-compliance with the CSP. This paper aims to develop a framework to enhance employees’ compliance with the CSP by implementing effective CSPTAP. Drawing from the present literature and reflecting on the existing behavior change wheel (BCW) framework, and capability, opportunity, motivation, and behavior (COM-B) model, the cybersecurity policy compliance (CSPC) framework is developed. The CSPC framework comprises the following key concepts: learning existing policies, conducting employees’ gap analysis, reviewing existing policies/developing new policies, provision of relevant content and delivery mode, and periodic auditing. These key concepts are essential elements in cybersecurity policy compliance. The model indicates that the implementation of the essential elements will substantially influence employees’ compliance with CSP. Moreover, when organizations consider these key elements, cybersecurity policy training and awareness, can positively enhance employees’ CSP compliance. The proposed development of CSPTAP provides a firm base for future empirical work including action research. © 2024, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

关键词： Personnel training

来源：评论

学校读者我要写书评

暂无评论

Enhancing Cybersecurity Policy Awareness Programs in the Financial Sector: A Comprehensive Framework and Assessment of Effectiveness 1

引用

International Conference on Business and Technology, ICBT 2024

作者： Thamae, Reneuoe Abdullah, Hanifa Mujinga, Mathias School of Computing College of Science Engineering and Technology University of South Africa Pretoria South Africa

ISBN: (数字)9783031674440

ISBN: (纸本)9783031674433

In today's tech-driven landscape and amid rising cyber threats, prioritizing cybersecurity is crucial for financial organizations. While traditional measures like firewalls are insufficient, human vulnerability persists due to inadequate employee training. To solve this, institutions must enact comprehensive awareness programs on cybersecurity policies. Despite their recognized importance, research gaps exist regarding their effectiveness. This study proposes a framework to enhance such programs, drawing from literature reviews and expert insights. Using a quantitative approach, it examines employee perspectives, revealing gender disparities in cybersecurity knowledge. The framework integrates gender mainstreaming to address these gaps and recommends proactive measures such as inclusive training and reporting procedures. By emphasizing inclusivity and diversity, it aims to foster a robust cybersecurity culture and ensure policy adherence across financial organizations. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024.

关键词： Decentralized finance

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：