检索结果-内蒙古大学图书馆

Video-based body geometric aware network for 3D human pose estimation

Optoelectronics Letters 2022年第5期18卷 313-320页

作者： LI Chaonan LIU Sheng YAo Lu ZOU Siyu College of Computer Science and Technology Zhejiang University of TechnologyHangzhou 310023Chiana

Three-dimensional human pose estimation(3 D HPE) has broad application prospects in the fields of trajectory prediction, posture tracking and action analysis. However, the frequent self-occlusions and the substantial depth ambiguity in two-dimensional(2 D) representations hinder the further improvement of accuracy. In this paper, we propose a novel video-based human body geometric aware network to mitigate the above problems. Our network can implicitly be aware of the geometric constraints of the human body by capturing spatial and temporal context information from 2 D skeleton data. Specifically, a novel skeleton attention(SA) mechanism is proposed to model geometric context dependencies among different body joints, thereby improving the spatial feature representation ability of the network. To enhance the temporal consistency, a novel multilayer perceptron(MLP)-Mixer based structure is exploited to comprehensively learn temporal context information from input sequences. We conduct experiments on publicly available challenging datasets to evaluate the proposed approach. The results outperform the previous best approach by 0.5 mm in the Human3.6 m dataset. It also demonstrates significant improvements in Human Eva-I dataset.

关键词： network geometric estimation

来源：评论

学校读者我要写书评

暂无评论

Classification of Pulmonary Diseases Using a Deep Learning Stacking Ensemble Model

引用

Informatica (Slovenia) 2024年第14期48卷 43-64页

作者： Sadoon, Ruaa N. Chaid, Adala M. College of Computer Science & Information Technology University of Basrah Basrah Iraq

This paper presents our research in the area of medical imaging diagnostics, focusing specifically on countering the devastating impact of the COVID-19 pandemic and numerous pulmonary pathologies. Using new deep-learning approaches and techniques, we aim to create an advanced classification tool that will be able to capture complex patterns and features in chest image data. This paper introduces the use of state-of-the-art strategies, such as stacked ensemble models, transfer learning, and artificial neural networks, to build a model with unprecedented precision, recall, F1-score, and accuracy. The core idea of our research is to combine different convolutional neural network architectures to bring together their best extraction and classification qualities. The combination of DenseNet, Xception and Inception achieves the best performance and provides the most reliable classification tool. We also use transfer learning to quickly train our model and optimize generalization, making it suitable for the detection of multiple pulmonary pathologies, including COVID-19. Our model also includes an artificial neural network, trained as a meta-learner, which processes the outputs of the CNNs to make classification decisions. We have thoroughly validated and optimized the meng-learner to improve the model’s accuracy on diagnostic images. The provided paper proposed the successful merge of cutting-edge deep-learning methodologies and image-processing algorithms with the medical imaging industry’s specifics. We aim to disrupt the pulmonary disease diagnosis field with our model, offering medical institutions a reliable tool to fight the current and future threats and challenges posed by COVID-19. © 2024 Slovene Society Informatika. All rights reserved.

关键词： COVID-19

来源：评论

学校读者我要写书评

暂无评论

Unlocking the Non-deterministic Computing Power with Memory-Elastic Multi-Exit Neural Networks 24

Unlocking the Non-deterministic Computing Power with Memory-...

引用

33rd ACM Web Conference, WWW 2024

作者： Huang, Jiaming Gao, Yi Dong, Wei College of Computer Science Zhejiang University China

ISBN: (纸本)9798400701719

With the increasing demand for Web of Things (WoT) and edge computing, the efficient utilization of limited computing power on edge devices is becoming a crucial challenge. Traditional neural networks (NNs) as web services rely on deterministic computational resources. However, they may fail to output the results on non-deterministic computing power which could be preempted at any time, degrading the task performance significantly. Multi-exit NNs with multiple branches have been proposed as a solution, but the accuracy of intermediate results may be unsatisfactory. In this paper, we propose MEEdge, a system that automatically transforms classic single-exit models into heterogeneous and dynamic multi-exit models which enables Memory-Elastic inference at the Edge with non-deterministic computing power. To build heterogeneous multi-exit models, MEEdge uses efficient convolutions to form a branch zoo and High Priority First (HPF)-based branch placement method for branch growth. To adapt models to dynamically varying computational resources, we employ a novel on-device scheduler for collaboration. Further, to reduce the memory overhead caused by dynamic branches, we propose neuron-level weight sharing and few-shot knowledge distillation(KD) retraining. Our experimental results show that models generated by MEEdge can achieve up to 27.31% better performance than existing multi-exit NNs. © 2024 ACM.

关键词： Edge computing

来源：评论

学校读者我要写书评

暂无评论

Memory-efficient detection of large-scale obfuscated malware

引用

International Journal of Wireless and Mobile Computing 2024年第1期26卷 48-60页

作者： Wang, Yueming Zhang, Meng College of Computer Science and Technology Jilin University Jilin Changchun China

Obfuscation techniques are frequently used in malicious programs to evade detection. However, current effective methods often require much memory space during training. This paper proposes a machine-learning-based solution to the malware detection problem that consumes fewer memory resources. We use hash and sparse matrix to build a text bag of words to reduce memory usage during training. Experiments show that our approach reduces the memory footprint by 95% when using 110,000 text data for confusion recognition training compared to the existing model. In the de-obfuscation step, our method improves the recognition accuracy of the import table function by 40%. Our model achieves shallow memory usage during confusion recognition training and enhances the accuracy of imported table recognition. Additionally, the confusion recognition accuracy is only about 10% lower than the confusion recognition model before the improvement. Copyright © 2024 Inderscience Enterprises Ltd.

关键词： Malware

来源：评论

学校读者我要写书评

暂无评论

ERNN: Error-Resilient RNN for Encrypted Traffic Detection towards Network-Induced Phenomena

引用

IEEE Transactions on Dependable and Secure Computing 2023年 1-18页

作者： Zhao, Ziming Li, Zhaoxuan Jiang, Jialun Yu, Fengyuan Zhang, Fan Xu, Congyuan Zhao, Xinjie Zhang, Rui Guo, Shize College of Computer Science and Technology Zhejiang University Hangzhou China State Key Laboratory of Information Security CAS Beijing China Jiaxing University College of Information Science and Engineering Jiaxing Zhejiang China

Traffic detection systems based on machine learning have been proposed to defend against cybersecurity threats, such as intrusion attacks and malware. However, they did not take the impact of network-induced phenomena into consideration, such as packet loss, retransmission, and out-of-order. These phenomena will introduce additional misclassifications in the real world. In this paper, we present ${\sf ERNN}$, a robust and end-to-end RNN model that is specially designed against network-induced phenomena. As its core, ${\sf ERNN}$ is designed with a novel gating unit named as session gate that includes: (i) four types of actions to simulate common network-induced phenomena during model training;and (ii) the Mealy machine to update states of session gate that adjusts the probability distribution of network-induced phenomena. Taken together, ${\sf ERNN}$ advances state-of-the-art by realizing the model robustness for network-induced phenomena in an error-resilient manner. We implement ${\sf ERNN}$ and evaluate it extensively on both intrusion detection and malware detection systems. By practical evaluation with dynamic bandwidth utilization and different network topologies, we demonstrate that ${\sf ERNN}$ can still identify 98.63% of encrypted intrusion traffic when facing about 16% abnormal packet sequences on a 10 Gbps dataplane. Similarly, ${\sf ERNN}$ can still robustly identify more than 97% of the encrypted malware traffic in multi-user concurrency scenarios. ${\sf ERNN}$ can realize $\sim$4% accuracy more than SOTA methods. Based on the Integrated Gradients method, we interpret the gating mechanism can reduce the dependencies on local packets (termed dependency dispersion). Moreover, we demonstrate that ${\sf ERNN}$ possesses superior stability and scalability in terms of parameter settings and feature selection. IEEE

关键词： Malware

来源：评论

学校读者我要写书评

暂无评论

Biometric-Based Secure Encryption Key Generation Using Convolutional Neural Networks and Particle Swarm Optimization

Informatica (Slovenia)

引用

Informatica (Slovenia) 2025年第16期49卷 213-234页

作者： Almola, Sahera A.S. Khudeyer, Raidah S. Younis, Hameed Abdulkareem Department of Computer Information Systems College of Computer Science and Information Technology University of Basrah Basrah Iraq

With the rapid expansion of computer networks and information technology, ensuring secure data transmission is increasingly vital—especially for image data, which often contains sensitive information. This research presents a biometric-based encryption system that uses fingerprint recognition and deep learning to generate strong, random encryption keys. Two convolutional neural networks (CNNs) are employed: one to verify identity based on a user’s ID and another to extract fingerprint features for key generation. These keys are optimized using Particle Swarm Optimization (PSO), enhancing their randomness and resistance to brute-force attacks. The system generates keys in real-time, eliminating the need for storage and minimizing the risk of theft or leakage. To further improve security, encryption keys are automatically updated after every ten messages, with different keys generated from multiple fingerprints of the same individual. Testing with the SOCOFing dataset (6,000 original and 49,270 synthetic images) achieved 99.75% identity verification and 99.83% classification accuracy. Performance metrics—entropy of 7.89, correlation factor of 0.00628, and zero repetition—demonstrate high robustness. This approach offers a secure, adaptive, and personalized encryption method ideal for sensitive domains like finance and healthcare. © 2025 Slovene Society Informatika. All rights reserved.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

A self-learning human-machine cooperative control method based on driver intention recognition

引用

CAAI Transactions on Intelligence technology 2024年第5期9卷 1101-1115页

作者： Yan Jiang Yuyan Ding Xinglong Zhang Xin Xu Junwen Huang College of Intelligence Science and Technology National University of Defense TechnologyChangshaChina Department of Computer Science Technical University of MunichGarchingGermany

Human-machine cooperative control has become an important area of intelligent driving,where driver intention recognition and dynamic control authority allocation are key factors for improving the performance of cooperative decision-making and *** this paper,an online learning method is proposed for human-machine cooperative control,which introduces a priority control parameter in the reward function to achieve optimal allocation of control authority under different driver intentions and driving safety ***,a two-layer LSTM-based sequence prediction algorithm is proposed to recognise the driver's lane change(LC)intention for human-machine cooperative steering ***,an online reinforcement learning method is developed for optimising the steering authority to reduce driver workload and improve driving *** driver-in-the-loop simulation results show that our method can accurately predict the driver's LC intention in cooperative driving and effectively compensate for the driver's non-optimal driving *** experimental results on a real intelligent vehicle further demonstrate the online optimisation capability of the proposed RL-based control authority allocation algorithm and its effectiveness in improving driving safety.

关键词： Self-leaming Control Hmman-machine Cooperation Intelligent Vehicles Reinfocement Learning

来源：评论

学校读者我要写书评

暂无评论

Plasma-assisted fabrication of multiscale materials for electrochemical energy conversion and storage

引用

Carbon Energy 2025年第2期7卷 144-171页

作者： Chen Li Tengfei Zhang Zhong Qiu Beirong Ye Xinqi Liang Xin Liu Minghua Chen Xinhui Xia Chen Wang Wangjun Wan Yongqi Zhang Institute of Fundamental and Frontier Sciences University of Electronic Science and Technology of ChinaChengduChina Key Laboratory of Engineering Dielectric and Applications(Ministry of Education) School of Electrical and Electronic EngineeringHarbin University of Science and TechnologyHarbinChina State Key Laboratory of New Textile Materials and Advanced Processing Technology Wuhan Textile UniversityWuhanChina College of Materials Science and Engineering Zhejiang University of TechnologyHangzhouChina School of Materials Science and Engineering Zhejiang UniversityHangzhouChina Zhejiang Academy of Science and Technology for Inspection and Quarantine HangzhouChina

Plasma,the fourth state of matter,is characterized by the presence of charged particles,including ions and *** has been shown to induce unique physical and chemical ***,there have been increased applications of plasma technology in the field of multiscale functional materials'preparation,with a number of interesting *** review will begin by introducing the basic knowledge of plasma,including the definition,typical parameters,and classification of plasma *** this,we will provide a comprehensive review and summary of the applications(phase conversion,doping,deposition,etching,exfoliation,and surface treatment)of plasma in common energy conversion and storage systems,such as electrocatalytic conversion of small molecules,batteries,fuel cells,and *** article summarizes the structure-performance relationships of electrochemical energy conversion and storage materials(ECSMs)that have been prepared or modified by *** also provides an overview of the challenges and perspectives of plasma technology,which could lead to a new approach for designing and modifying electrode materials in ECSMs.

关键词： electrocatalysts energy conversion and storage nanomaterials plasma

来源：评论

学校读者我要写书评

暂无评论

Modeling Price-Aware Session-Based Recommendation Based on Graph Neural Network

引用

computers, Materials & Continua 2023年第7期76卷 397-413页

作者： Jian Feng Yuwen Wang Shaojian Chen College of Computer Science&Technology Xi’an University of Science and TechnologyXi’an710600China

Session-based Recommendation(SBR)aims to accurately recom-mend a list of items to users based on anonymous historical session *** methods for SBR suffer from several limitations:SBR based on Graph Neural Network often has information loss when constructing session graphs;Inadequate consideration is given to influencing factors,such as item price,and users’dynamic interest evolution is not taken into account.A new session recommendation model called Price-aware Session-based Recommendation(PASBR)is proposed to address these *** constructs session graphs by information lossless approaches to fully encode the original session information,then introduces item price as a new factor and models users’price tolerance for various items to influence users’*** addition,PASBR proposes a new method to encode user intent at the item category level and tries to capture the dynamic interest of users over ***,PASBR fuses the multi-perspective features to generate the global representation of users and make a ***,the intent,the short-term and long-term interests,and the dynamic interests of a user are *** on two real-world datasets show that PASBR can outperform representative baselines for SBR.

关键词： Session-based recommendation graph neural network price-aware intention dynamic interest

来源：评论

学校读者我要写书评

暂无评论

An Innovative Approach Using TKN-Cryptology for Identifying the Replay Assault

引用

computers, Materials & Continua 2024年第1期78卷 589-616页

作者： Syeda Wajiha Zahra Muhammad Nadeem Ali Arshad Saman Riaz Muhammad Abu Bakr Ashit Kumar Dutta Zaid Alzaid Badr Almutairi Sultan Almotairi Department of Computer Science Alhamd Islamic UniversityIslamabad44000Pakistan Department of Computer Science and Technology University of Science and Technology BeijingBeijing100083China Department of Computer Science National University of TechnologyIslamabad44000Pakistan Department of Electrical Engineering National University of TechnologyIslamabad44000Pakistan Department of Computer Science and Information Systems College of Applied SciencesAlMaarefa UniversityRiyadh13713Saudi Arabia Department of Computer Science Faculty of Computer and Information SystemsIslamic University of MadinahMedinah42351Saudi Arabia Department of Information Technology College of Computer Sciences and Information Technology CollegeMajmaah UniversityAl-Majmaah11952Saudi Arabia Department of Natural and Applied Sciences Faculty of Community CollegeMajmaah UniversityMajmaah11952Saudi Arabia

Various organizations store data online rather than on physical *** the number of user’s data stored in cloud servers increases,the attack rate to access data from cloud servers also *** researchers worked on different algorithms to protect cloud data from replay *** of the papers used a technique that simultaneously detects a full-message and partial-message replay *** study presents the development of a TKN(Text,Key and Name)cryptographic algorithm aimed at protecting data from replay *** program employs distinct ways to encrypt plain text[P],a user-defined Key[K],and a Secret Code[N].The novelty of the TKN cryptographic algorithm is that the bit value of each text is linked to another value with the help of the proposed algorithm,and the length of the cipher text obtained is twice the length of the original *** the scenario that an attacker executes a replay attack on the cloud server,engages in cryptanalysis,or manipulates any data,it will result in automated modification of all associated values inside the *** mechanism has the benefit of enhancing the detectability of replay ***,the attacker cannot access data not included in any of the papers,regardless of how effective the attack strategy *** the end of paper,the proposed algorithm’s novelty will be compared with different algorithms,and it will be discussed how far the proposed algorithm is better than all other algorithms.

关键词： Replay attack malware message attack file encryption cryptology data security

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：