检索结果-内蒙古大学图书馆

Enhancing medical vision-language contrastive learning via inter-matching relation modelling

IEEE Transactions on Medical Imaging 2025年第6期44卷 2463-2476页

作者： Li, Mingjian Meng, Mingyuan Fulham, Michael Feng, David Dagan Bi, Lei Kim, Jinman University of Sydney Biomedical data analysis and visualization lab School of Computer Science NSW Australia Royal Prince Alfred Hospital Department of Molecular Imaging NSW Australia Shanghai Jiao Tong University Institute of Translational Medicine Shanghai China

Medical image representations can be learned through medical vision-language contrastive learning (mVLCL) where medical imaging reports are used as weak supervision through image-text alignment. These learned image representations can be transferred to and benefit various downstream medical vision tasks such as disease classification and segmentation. Recent mVLCL methods attempt to align image sub-regions and the report keywords as local-matchings. However, these methods aggregate all local-matchings via simple pooling operations while ignoring the inherent relations between them. These methods therefore fail to reason between local-matchings that are semantically related, e.g., local-matchings that correspond to the disease word and the location word (semantic-relations), and also fail to differentiate such clinically important local-matchings from others that correspond to less meaningful words, e.g., conjunction words (importance-relations). Hence, we propose a mVLCL method that models the inter-matching relations between local-matchings via a relation-enhanced contrastive learning framework (RECLF). In RECLF, we introduce a semantic-relation reasoning module (SRM) and an importance-relation reasoning module (IRM) to enable more fine-grained report supervision for image representation learning. We evaluated our method using six public benchmark datasets on four downstream tasks, including segmentation, zero-shot classification, linear classification, and cross-modal retrieval. Our results demonstrated the superiority of our RECLF over the state-of-the-art mVLCL methods with consistent improvements across single-modal and cross-modal tasks. These results suggest that our RECLF, by modelling the inter-matching relations, can learn improved medical image representations with better generalization capabilities. © 2025 IEEE.

关键词： Contrastive Learning

来源：评论

学校读者我要写书评

暂无评论

PraK Tool V3: Enhancing Video Item Search Using Localized Text and Texture Queries 31st

PraK Tool V3: Enhancing Video Item Search Using Localized Te...

引用

31st International Conference on Multimedia Modeling, MMM 2025

作者： Stroh, Michael Kloda, Vojtěch Verner, Benjamin Vopálková, Zuzana Buchmüller, Raphael Jäckl, Bastian Hajko, Jakub Lokoč, Jakub Visual Computing Department of Computer Science Faculty of Natural Sciences Konstanz University Konstanz Germany SIRET Research Group Department of Software Engineering Faculty of Mathematics and Physics Charles University Prague Czech Republic Data Analysis and Visualization Department of Computer Science Faculty of Natural Sciences Konstanz University Konstanz Germany

ISBN: (纸本)9789819620739

We present a third version of the PraK system designed around an effective text-image and image-image search model. The system integrates sub-image search options for localized context search for CLIP and image color/texture and interfaces tailored to specific task types. A user study shows the benefits of sub-image search in homogeneous video collections. We further enhance the user experience of the PraK framework by re-designing the frontend and introducing task-specific views. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.

关键词： Video analysis

来源：评论

学校读者我要写书评

暂无评论

ControlVideo: conditional control for one-shot text-driven video editing and beyond

引用

science China(Information sciences) 2025年第3期68卷 150-162页

作者： Min ZHAO Rongzhen WANG Fan BAO Chongxuan LI Jun ZHU Department of Computer Science and Technology Institute for AI Tsinghua-Bosch Joint ML CenterTsinghua Laboratory of Brain and Intelligence Lab Tsinghua University ShengShu Technology Gaoling School of Artificial Intelligence Renmin University of China Beijing Key Laboratory of Big Data Management and Analysis Methods Pazhou Laboratory (Huangpu)

This paper presents ControlVideo for text-driven video editing — generating a video that aligns with a given text while preserving the structure of the source video. Building on a pre-trained text-to-image diffusion model, ControlVideo enhances the fidelity and temporal consistency by incorporating additional conditions(such as edge maps), and fine-tuning the key-frame and temporal attention on the source video-text pair via an in-depth exploration of the design space. Extensive experimental results demonstrate that ControlVideo outperforms various competitive baselines by delivering videos that exhibit high fidelity w.r.t. the source content, and temporal consistency, all while aligning with the text. By incorporating low-rank adaptation layers into the model before training, ControlVideo is further empowered to generate videos that align seamlessly with reference images. More importantly, ControlVideo can be readily extended to the more challenging task of long video editing(e.g., with hundreds of frames), where maintaining long-range temporal consistency is crucial. To achieve this, we propose to construct a fused ControlVideo by applying basic ControlVideo to overlapping short video segments and key frame videos and then merging them by pre-defined weight functions. Empirical results validate its capability to create videos across 140 frames, which is approximately 5.83 to 17.5 times more than what previous studies achieved. The code is available at https://***/thu-ml/controlvideo.

关键词： diffusion models controllable generation text-driven editing video editing long video editing

来源：评论

学校读者我要写书评

暂无评论

PraK Tool: An Interactive Search Tool Based on Video data Services 30th

PraK Tool: An Interactive Search Tool Based on Video Data S...

引用

30th International Conference on MultiMedia Modeling, MMM 2024

作者： Lokoč, Jakub Vopálková, Zuzana Stroh, Michael Buchmueller, Raphael Schlegel, Udo SIRET Research Group Department of Software Engineering Faculty of Mathematics and Physics Charles University Prague Czech Republic Visual Computing Department of Computer Science Faculty of Natural Sciences Konstanz University Konstanz Germany Data Analysis and Visualization Department of Computer Science Faculty of Natural Sciences Konstanz University Konstanz Germany

ISBN: (纸本)9783031533013

This paper presents a tool relying on data service architecture, where technical details of all VBS datasets are completely hidden behind an abstract stateless data layer. The data services allow independent development of interactive search interfaces and refinement techniques, which is demonstrated by a smart front-end component. The component supports common search features and allows users to exploit content-based statistics for effective filtering. We believe that video data services might be a valuable addition to the open-source VBS toolkit, especially when available for the competition on a shared server with all VBS datasets, extracted features, and meta-data behind. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024.

关键词： Video recording

来源：评论

学校读者我要写书评

暂无评论

Artificial intelligence in physiological characteristics recognition for internet of things authentication

引用

Digital Communications and Networks 2024年第3期10卷 740-755页

作者： Zhimin Zhang Huansheng Ning Fadi Farha Jianguo Ding Kim-Kwang Raymond Choo School of Computer and Communication Engineering University of Science and Technology BeijingBeijing100083China Beijing Engineering Research Center for Cyberspace Data Analysis and Applications Beijing100083China Faculty of Informatics Engineering Aleppo UniversityAleppoSyria Department of Computer Science Blekinge Institute of Technology37179KarlskronaSweden Department of Information Systems and Cyber Security University of Texas at San AntonioSan AntonioTX78249-0631USA

Effective user authentication is key to ensuring equipment security,data privacy,and personalized services in Internet of Things(IoT)***,conventional mode-based authentication methods(e.g.,passwords and smart cards)may be vulnerable to a broad range of attacks(e.g.,eavesdropping and side-channel attacks).Hence,there have been attempts to design biometric-based authentication solutions,which rely on physiological and behavioral *** characteristics need continuous monitoring and specific environmental settings,which can be challenging to implement in ***,we can also leverage Artificial Intelligence(AI)in the extraction and classification of physiological characteristics from IoT devices processing to facilitate ***,we review the literature on the use of AI in physiological characteristics recognition pub-lished after *** use the three-layer architecture of the IoT(i.e.,sensing layer,feature layer,and algorithm layer)to guide the discussion of existing approaches and their *** also identify a number of future research opportunities,which will hopefully guide the design of next generation solutions.

关键词： Physiological characteristics recognition Artificial intelligence Internet of things Biological-driven authentication

来源：评论

学校读者我要写书评

暂无评论

Blockchain-Based Cognitive Computing Model for data Security on a Cloud Platform

引用

computers, Materials & Continua 2023年第12期77卷 3305-3323页

作者： Xiangmin Guo Guangjun Liang Jiayin Liu Xianyi Chen Department of Computer Information and Cyber Security Jiangsu Police InstituteNanjing210031China Jiangsu Electronic Data Forensics and Analysis Engineering Research Center Jiangsu Police InstituteNanjing210031China School of Computer and Software Nanjing University of Information Science and TechnologyNanjing210044China

Cloud storage is widely used by large companies to store vast amounts of data and files,offering flexibility,financial savings,and ***,information shoplifting poses significant threats,potentially leading to poor performance and privacy ***-based cognitive computing can help protect and maintain information security and privacy in cloud platforms,ensuring businesses can focus on business *** ensure data security in cloud platforms,this research proposed a blockchain-based Hybridized data Driven Cognitive Computing(HD2C)***,the proposed HD2C framework addresses breaches of the privacy information of mixed participants of the Internet of Things(IoT)in the ***2C is developed by combining Federated Learning(FL)with a Blockchain consensus algorithm to connect smart contracts with Proof of ***“data Island”problem can be solved by FL’s emphasis on privacy and lightning-fast processing,while Blockchain provides a decentralized incentive structure that is impervious to *** with Blockchain allows quick consensus through smart member selection and *** HD2C paradigm significantly improves the computational processing efficiency of intelligent *** analysis results derived from IIoT datasets confirm HD2C *** compared to other consensus algorithms,the Blockchain PoA’s foundational cost is *** accuracy and memory utilization evaluation results predict the total benefits of the *** comparison to the values 0.004 and 0.04,the value of 0.4 achieves good *** to the experiment results,the number of transactions per second has minimal impact on memory *** findings of this study resulted in the development of a brand-new IIoT framework based on blockchain technology.

关键词： Blockchain Internet of Things(IoT) blockchain based cognitive computing Hybridized data Driven Cognitive Computing(HD2C) Federated Learning(FL) Proof of Authority(PoA)

来源：评论

学校读者我要写书评

暂无评论

Evolving Safety Protocols: Deep Learning-Enabled Detection of Personal Protective Equipment 1

引用

4th International Conference on computer Vision, High-Performance Computing, Smart Devices, and Networks, CHSN 2023

作者： Alahmid, Mustafa Bhimani, Kishankumar Saradva, Khushbu Ghildiyal, Sushil Saeedi, Sayed Aryan Ali, Saraa Yan, Han Filippov, Evgeny Faculty of Computer Science and Technology Saint Petersburg Electrotechnical University "LETI" Saint Petersburg Russia Laboratory on AI for Computational Biology Faculty of Computer Science HSE University Moscow Russia Department of Computer Science and Engineering Indian Institute of Technology Ropar Punjab India Laboratory of Methods for Big Data Analysis Faculty of Computer Science HSE University Moscow Russia Deggendorf Institute of Technology Bavaria Germany

ISBN: (数字)9789819777945

ISBN: (纸本)9789819777938

To give shift in safety protocols, we have employed advanced deep learning algorithms and frameworks (Shrestha and Mahmood in IEEE Access 7:53,040–53,065, 2019 [25]) to construct an innovative AI model. The designed model detects the usage of personal protective equipment (PPE) (Personal protective equipment. Geneva: World Health Organization, 2020 [18]) by workers in high-risk industries such as construction and manufacturing. We have used Google’s TensorFlow object detection API (Sai and Sasikala in Object detection and count of objects in image using tensor flow object detection API, pp 542–546, 2019 [22]) to modify and train a model for dual purposes: PPE detection and face recognition. The state-of-the-art of this research is to substantially enhance safety compliance by addressing the prevalent issue of PPE non-compliance. To emphasis this, we have developed a pioneering software prototype that synergizes PPE detection with a face recognition-based clock-in system. This prototype demonstrates impressive object detection metrics with a mean average precision (mAP) of 0.9 for vests and 0.85 for helmets. Moreover, it exhibited efficient face recognition with a successful threshold range of 17–20%. The implementation of AI in our system promises significant enhancements to worker safety, while concurrently reducing the financial burden associated with big hazards and accidents. Beyond the development and performance of the system, this paper provides a thorough exploration of the encountered challenges, potential real-world applications (particularly in employee monitoring and clock-in systems), and the future implications of this study on research and practical applications in the field of AI-integrated safety compliance. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024.

关键词： Software prototyping

来源：评论

学校读者我要写书评

暂无评论

HOFD: An Outdated Fact Detector for Knowledge Bases

引用

IEEE Transactions on Knowledge and data Engineering 2023年第10期35卷 10775-10789页

作者： Hao, Shuang Chai, Chengliang Li, Guoliang Tang, Nan Wang, Ning Yu, Xiang Beijing Jiaotong University School of Computer and Information Technology Beijing100044 China Beijing Key Laboratory of Traffic Data Analysis and Mining Beijing100044 China Beijing Institute of Technology School of Computer Science and Technology Beijing100811 China Beijing100190 China Qatar Computing Research Institute Hbku Doha34110 Qatar Huawei Company Gauss Department Beijing100031 China

Knowledge bases (KBs), which store high-quality information, are crucial for many applications, such as enhancing search results and serving as external sources for data cleaning. Not surprisingly, there exist outdated facts in most KBs due to the rapid change of information. Naturally, it is important to keep KBs up-to-date. Traditional wisdom has investigated the problem of using reference data (such as new facts extracted from the news) to detect outdated facts in KBs. However, existing approaches can only cover a small percentage of facts in KBs. In this paper, we propose HOFD, a novel human-in-the-loop approach for outdated fact detection in KBs. HOFD trains a binary classifier using features such as historical update frequency and update time of a fact to compute the likelihood of a fact in a KB to be outdated. Then, HOFD interacts with humans to verify whether a fact with high likelihood is indeed outdated. In addition, HOFD also uses logical rules to detect more outdated facts based on human feedback. The outdated facts detected by the logical rules will also be fed back to train the ML model further for data augmentation. Extensive experiments on real-world KBs, such as Yago and DBpedia, show the effectiveness of our solution. © 1989-2012 IEEE.

关键词： Knowledge based systems

来源：评论

学校读者我要写书评

暂无评论

APT Adversarial Defence Mechanism for Industrial IoT Enabled Cyber-Physical System

引用

IEEE Access 2023年 11卷 74000-74020页

作者： Javed, Safdar Hussain Ahmad, Maaz Bin Asif, Muhammad Akram, Waseem Mahmood, Khalid Das, Ashok Kumar Shetty, Sachin College of Computing and Information Sciences Karachi75190 Pakistan Lahore Garrison University Department of Computer Science Lahore54810 Pakistan National Yunlin University of Science and Technology Graduate School of Intelligent Data Science Douliu64002 Taiwan International Institute of Information Technology at Hyderabad Center for Security Theory and Algorithmic Research Hyderabad500032 India Old Dominion University Simulation and Visualization Engineering Virginia Modeling Analysis and Simulation Center Department of Modeling SuffolkVA23435 United States Old Dominion University Center for Cybersecurity Education and Research SuffolkVA23435 United States

The objective of Advanced Persistent Threat (APT) attacks is to exploit Cyber-Physical Systems (CPSs) in combination with the Industrial Internet of Things (I-IoT) by using fast attack methods. Machine learning (ML) techniques have shown potential in identifying APT attacks in autonomous and malware detection systems. However, detecting hidden APT attacks in the I-IoT-enabled CPS domain and achieving real-time accuracy in detection present significant challenges for these techniques. To overcome these issues, a new approach is suggested that is based on the Graph Attention Network (GAN), a multi-dimensional algorithm that captures behavioral features along with the relevant information that other methods do not deliver. This approach utilizes masked self-attentional layers to address the limitations of prior Deep Learning (DL) methods that rely on convolutions. Two datasets, the DAPT2020 malware, and Edge I-IoT datasets are used to evaluate the approach, and it attains the highest detection accuracy of 96.97% and 95.97%, with prediction time of 20.56 seconds and 21.65 seconds, respectively. The GAN approach is compared to conventional ML algorithms, and simulation results demonstrate a significant performance improvement over these algorithms in the I-IoT-enabled CPS realm. © 2013 IEEE.

关键词： Real time systems

来源：评论

学校读者我要写书评

暂无评论

Blockchain and PUF-based secure key establishment protocol for cross-domain digital twins in industrial Internet of Things architecture

引用

Journal of Advanced Research 2024年 62卷 155-163页

作者： Mahmood, Khalid Shamshad, Salman Saleem, Muhammad Asad Kharel, Rupak Das, Ashok Kumar Shetty, Sachin Rodrigues, Joel J.P.C. School of Psychology and Computer Science University of Central Lancashire Preston United Kingdom Graduate School of Intelligent Data Science National Yunlin University of Science and Technology Yunlin 64002 Taiwan Department of Software Engineering The University of Lahore Lahore 54590 Pakistan Department of Computer Science University of Sahiwal Punjab Sahiwal 57000 Pakistan Center for Security Theory and Algorithmic Research International Institute of Information Technology Hyderabad 500 032 India Department of Modeling Simulation and Visualization Engineering Virginia Modeling Analysis and Simulation Center and Center for Cybersecurity Education and Research Old Dominion University Suffolk 23435 VA United States COPELABS Lusófona University Campo Grande 376 Lisbon 1749-024 Portugal

Introduction: The Industrial Internet of Things (IIoT) is a technology that connects devices to collect data and conduct in-depth analysis to provide value-added services to industries. The integration of the physical and digital domains is crucial for unlocking the full potential of the IIoT, and digital twins can facilitate this integration by providing a virtual representation of real-world entities. Objectives: By combining digital twins with the IIoT, industries can simulate, predict, and control physical behaviors, enabling them to achieve broader value and support industry 4.0 and 5.0. Constituents of cooperative IIoT domains tend to interact and collaborate during their complicated operations. Methods: To secure such interaction and collaborations, we introduce a blockchain-based cross-domain authentication protocol for IIoT. The blockchain maintains only each domain's dynamic accumulator, which accumulates crucial materials derived from devices, decreasing the overhead. In addition, we use the on-chain accumulator to effectively validate the unlinkable identities of cross-domain IIoT devices. Results: The implementation of the concept reveals the fact that our protocol is efficient and reliable. This efficiency and reliability of our protocol is also substantiated through comparison with state-of-the-art literature. In contrast to related protocols, our protocol exhibits a minimum 22.67% increase in computation cost efficiency and a 16.35% rise in communication cost efficiency. Conclusion: The developed protocol guarantees data transfer security across the domain and thwarts IoT devices from potential physical attacks. Additionally, in order to protect privacy, anonymity and unlinkability are also guaranteed. © 2024

关键词： Digital twins Industrial Internet of Things (IIoT) Key agreement Mutual authentication Physically Unclonable Functions (PUFs)

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

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

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：