检索结果-内蒙古大学图书馆

IEEE Transactions on Green Communications and Networking 2024年第1期8卷 265-278页

作者： Lotfolahi, Amin Ferng, Huei-Wen National Taiwan University of Science and Technology Department of Computer Science and Information Engineering Taipei106319 Taiwan

A novel cluster-based traffic offloading and user association (UA) algorithm alongside a multi-agent deep reinforcement learning (DRL) based base station (BS) activation mechanism is proposed in this paper. Our design aims to maximize the energy efficiency (EE) of the heterogeneous network (HetNet) while maintaining high quality of service (QoS). By taking advantage of the dense deployment of BSs in a HetNet, a clustering algorithm is first proposed to facilitate traffic offloading among BSs. Then, a multi-agent proximal policy optimization (MAPPO) based DRL algorithm is employed to trigger the BS activation decision based on the current environmental condition. Finally, a UA algorithm is deployed to improve further the (normalized) data rate of all users, known as the (normalized) sum rate. Via simulation, we show that our proposed mechanism can remarkably enhance the EE and excel over the closely related mechanisms. It satisfies the required data rate, improves the sum rate, and exhibits excellent scalability when many BSs are deployed. © 2017 IEEE.

关键词： Quality of service

来源：评论

学校读者我要写书评

暂无评论

IensNet: A novel and efficient approach for iris spoof detection via ensemble of deep models

引用

Multimedia Tools and Applications 2025年 1-30页

作者： Sharma, Deepika Selwal, Arvind School of Computer Science Engineering and Technology Bennett University Greater Noida201310 India Department of Computer Science and Information Technology Central University of Jammu Samba India

Iris biometrics allow contactless authentication, which makes it widely deployed human recognition mechanisms since the couple of years. Susceptibility of iris identification systems remains a challenging task due to diversity in spoof or presentation attacks (PAs) that fails to assure consistency while adopting them in real life scenarios. Hence, iris PAs are the growing concerns that gained significant attention in recent past decade. To alleviate these attacks or recognize presentation attack instruments (PAIs), iris presentation attacks detection (IPAD) algorithms are designed to distinguish a real and fabricated iris trait. Aiming at the efficient iris spoof detection mechanism, in this research work we expound a novel ensemble learning-enabled model (IensNet) that learns three pre-trained and fined-tuned deep models (i.e. DenseNet161, ResNet and VGGNet) for better accuracy and generalized performance. The novel IensNet approach offers several merits (i.e. consolidated strengths of multiple models, improved generalization ability, etc.) as compared to a simple transfer learning strategy where the knowledge is drawn from single pre-trained model. Finally, our approach learns a novel fully-connected dual layer classifier via outcome of three fine-tuned models to yield a final classification result as bonafide or spoof iris trait. Our approach is evaluated on Notre Dame LivDet iris 2017 and Notre Dame contact lenses 2015 anti-spoofing datasets. The experimental analysis of IensNet offers outstanding performance with a lower ACER of 0.2% and 1.4% for Iris-LivDet-2017 and Notre Dame contact lenses 2015 dataset respectively. Besides, IensNet exhibit promising results in cross-dataset environment with an ACA of 91.46%. © The Author(s), under exclusive licence to Springer science+Business Media, LLC, part of Springer Nature 2025.

关键词： Contrastive Learning

来源：评论

学校读者我要写书评

暂无评论

Advancements in Data Augmentation and Transfer Learning: A Comprehensive Survey to Address Data Scarcity Challenges

引用

Recent Advances in computer science and Communications 2024年第8期17卷 14-35页

作者： Fayaz, Salma Ahmad Shah, Syed Zubair Ud Din, Nusrat Mohi Gul, Naillah Assad, Assif Department of Computer Science and Engineering Islamic University of Science and Technology Awantipora Jammu and Kashmir Pulwama India

Deep Learning (DL) models have demonstrated remarkable proficiency in image classification and recognition tasks, surpassing human capabilities. The observed enhancement in performance can be attributed to the utilization of extensive datasets. Nevertheless, DL models have huge data requirements. Widening the learning capability of such models from limited samples even today remains a challenge, given the intrinsic constraints of small datasets. The trifecta of challenges, encompassing limited labeled datasets, privacy, poor generalization performance, and the costliness of annotations, further compounds the difficulty in achieving robust model performance. Overcoming the challenge of expanding the learning capabilities of Deep Learning models with limited sample sizes remains a pressing concern even today. To address this critical issue, our study conducts a meticulous examination of established methodologies, such as Data Augmentation and Transfer Learning, which offer promising solutions to data scarcity dilemmas. Data Augmentation, a powerful technique, amplifies the size of small datasets through a diverse array of strategies. These encompass geometric transformations, kernel filter manipulations, neural style transfer amalgamation, random erasing, Generative Adversarial Networks, augmentations in feature space, and adversarial and meta-learning training paradigms. Furthermore, Transfer Learning emerges as a crucial tool, leveraging pre-trained models to facilitate knowledge transfer between models or enabling the retraining of models on analogous datasets. Through our comprehensive investigation, we provide profound insights into how the synergistic application of these two techniques can significantly enhance the performance of classification tasks, effectively magnifying scarce datasets. This augmentation in data availability not only addresses the immediate challenges posed by limited datasets but also unlocks the full potential of working with Big Data in

关键词： Transfer learning

来源：评论

学校读者我要写书评

暂无评论

Centralized RANSAC-Based Point Cloud Registration With Fast Convergence and High Accuracy

引用

IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing 2024年 17卷 5431-5442页

作者： Chung, Kuo-Liang Chang, Wei-Tai National Taiwan University of Science and Technology Department of Computer Science and Information Engineering Taipei106335 Taiwan

For point cloud registration, the purpose of this article is to propose a novel centralized random sample consensus (RANSAC) (C-RANSAC) registration with fast convergence and high accuracy. In our algorithm, the novel contributions are, first, the proposal of a scale histogram-based outlier removal to delete outliers from the initial line vector set L for constructing a reduced line vector set Lred;second, the handshake cooperation between the host RANSAC (H-RANSAC) only working on L and the local RANSAC (LCL-RANSAC) only working on Lred;third, in each handshake process, after receiving the global registration solution and the global iteration number xH from H-RANSAC, LCLRANSAC uses the received global solution as the initial solution of the modified TEASER++ (M-TEASER++) method to calculate its first local registration solution. If the first local registration solution satisfies the global iteration number inheritance condition, LCL-RANSAC directly sends the accumulated iteration number, xH + 1, and the first local solution back to H-RANSAC;otherwise, LCL-RANSAC iteratively refines its local solution using the M-TEASER++ method, and then sends the resultant local solution and the required local iteration number xLCL to H-RANSAC for updating the global solution, the global iteration number to xH := xH + xLCL, and the global confidence level. Due to |Lred| |L| and employing the global iteration number inheritance condition test into our algorithm, we have conducted extensive experiments on testing point cloud pairs to show the registration accuracy and execution time merits of our algorithm relative to the state-of-theart methods. © 2008-2012 IEEE.

关键词： Iterative methods

来源：评论

学校读者我要写书评

暂无评论

CDS: Coupled Data Storage to Enhance Read Performance of 3D TLC NAND Flash Memory

引用

IEEE Transactions on computers 2024年第3期73卷 694-707页

作者： Wu, Wan-Ling Hsieh, Jen-Wei Ku, Hao-Yu National Taiwan University of Science and Technology Department of Computer Science and Information Engineering Taipei106 Taiwan

Due to the strong demand of massive storage capacity, the density of flash memory has been improved in terms of technology node scaling, multi-bit per cell technique, and 3D stacking. However, these techniques also degrade read performance and reliability. The long read latency comes from increased data sensing time and time-consuming ECC decoding time. Storing multiple bits per cell results in more read reference voltages and increased latency of identifying appropriate threshold voltages. To deal with error correction, LDPC is widely used in flash memory to provide stronger ECC capability. However, LDPC incurs a long decoding latency when bit errors are numerous. In this work, we propose coupled data storage (CDS) to improve the read performance of 3D NAND flash-memory storage devices. CDS supports two modes to improve read latency: The high read-speed mode is designed to improve data sensing time with reduced voltage states, while the data correction mode is designed to mitigate bit errors and LDPC overhead. Experiment results showed that CDS could reduce 50∼66.6% read latency and 25.7∼27.5% write latency under the high read-speed mode. For the data correction mode, RBER could be decreased by 37∼52% and the lifetime could be prolonged to 1.6 to 3 times. © 1968-2012 IEEE.

关键词： Threshold voltage

来源：评论

学校读者我要写书评

暂无评论

Evaluation of the Practicability of Internet of Things in Fighting Against COVID-19

引用

IEEE Internet of Things Journal 2024年第5期11卷 8064-8077页

作者： Lin, Yan-Ming Chu, Edward T.-H. National Yunlin University of Science and Technology Department of Computer Science and Information Engineering Douliou640 Taiwan

The COVID-19 pandemic has already ravaged the world for two years and infected more than 600 million people, having an irreparable impact on the health, economic, and political dimensions of human society. There have been many papers proposing various Internet of Things (IoT) solutions to stop the spread of COVID-19, such as virus tracking, infection control, medical treatment, and health management. Due to the high popularity of IoT devices, many studies have systematically investigated related technologies, evaluating the development, and application of IoT solutions from different aspects, such as electronic security, personal privacy, disaster prediction, and artificial intelligence (AI). However, the usefulness of IoT in combating COVID-19 has not been thoroughly evaluated. Therefore, this article systematically analyzes high-quality articles published in well-known international journals and conferences from 2019 to 2022 and divides them into the technical development of prevention and treatment according to the application field of the solution, cross-comparing the research results. It provides a comprehensive assessment of the maturity of the current IoT solutions through the Standard technology Readiness Level (TRL). Finally, through the technical efficacy matrix, the application status and development direction of the international IoT technology are analyzed by patent. Our results reveal the maturity and limitations of existing technologies to fight COVID-19. Possible future research directions are also discussed. © 2014 IEEE.

关键词： COVID-19

来源：评论

学校读者我要写书评

暂无评论

A user-adaptive privacy-preserving authentication of IoMT using zero knowledge proofs with ECC

引用

Multimedia Tools and Applications 2025年 1-32页

作者： Misra, Garima Hazela, Bramah Chaurasia, Brijesh Kumar Department of Computer Science and Engineering Amity School of Engineering and Technology Lucknow Amity University Uttar Pradesh India Department of Computer Science & Engineering Pranveer Singh Institute of Technology Utter Pradesh Kanpur India

The Internet of Things (IoT) has developed into a crucial component for meeting the connection needs of the current smart healthcare systems. The Internet of Medical Things (IoMT) consists of medical devices that are connected to patients in order to sense their medical parameters and exchange that information with healthcare professionals so they can deliver remote healthcare services. The need to protect the patient’s private information from being misused makes IoMT security an urgent issue that merits further study. Elliptic Curve cryptography (ECC), a popular substitute for traditional public key encryption like RSA, has several advantages. ECC is a prime contender for implementation on constrained devices with confined memory and speed, as well as wireless communication protocols that use low power like IoMT environment. The economic complexity of the current security protocols makes them inappropriate, hence new security protocols must be developed for resource-constrained and diverse IoT networks. This work suggests an authenticated key agreement protocol for the IoMT that is based on elliptic curve encryption based zero-knowledge proof (ZKP) protocols. The proposed mechanism completes in order of milliseconds up to 300 ms for 160 bits. The research will include the comparison between computational delays in traditional protocols and ZKP-ECC protocol. This paper provides a protocol to obtain privacy of data in IoMT environment on the basis of zero knowledge phenomenon and anonymous communication by ZKP-ECC. © The Author(s), under exclusive licence to Springer science+Business Media, LLC, part of Springer Nature 2025.

关键词： Authentication Protocol

来源：评论

学校读者我要写书评

暂无评论

GRA: Graph Representation Alignment for Semi-Supervised Action Recognition

引用

IEEE Transactions on Neural Networks and Learning Systems 2024年第9期35卷 11896-11905页

作者： Huang, Kuan-Hung Huang, Yao-Bang Lin, Yong-Xiang Hua, Kai-Lung Tanveer, M. Lu, Xuequan Razzak, Imran National Taiwan University of Science and Technology Department of Computer Science and Information Engineering Taipei10607 Taiwan Indian Institute of Technology Indore Department of Mathematics Indore Simrol453552 India La Trobe University Department of Computer Science and IT MelbourneVIC3086 Australia University of New South Wales School of Computer Science and Engineering SydneyNSW2052 Australia

Graph convolutional networks (GCNs) have emerged as a powerful tool for action recognition, leveraging skeletal graphs to encapsulate human motion. Despite their efficacy, a significant challenge remains the dependency on huge labeled datasets. Acquiring such datasets is often prohibitive, and the frequent occurrence of incomplete skeleton data, typified by absent joints and frames, complicates the testing phase. To tackle these issues, we present graph representation alignment (GRA), a novel approach with two main contributions: 1) a self-training (ST) paradigm that substantially reduces the need for labeled data by generating high-quality pseudo-labels, ensuring model stability even with minimal labeled inputs and 2) a representation alignment (RA) technique that utilizes consistency regularization to effectively reduce the impact of missing data components. Our extensive evaluations on the NTU RGB+D and Northwestern-UCLA (N-UCLA) benchmarks demonstrate that GRA not only improves GCN performance in data-constrained environments but also retains impressive performance in the face of data incompleteness. © 2012 IEEE.

关键词： Alignment

来源：评论

学校读者我要写书评

暂无评论

Enhancing patient-independent detection of freezing of gait in Parkinson’s disease with deep adversarial network

引用

Neural Computing and Applications 2025年 1-21页

作者： Fahad, Md Shah Ranjan, Ashish Kumar, Gautam Department of Computer Science and Engineering Birla Institute of Technology Mesra Jharkhand Ranchi India Department of Computer Science and Engineering C. V. Raman Global University Odisha Bhubaneswar India Department of Computer Science and Engineering National Institute of Technology Delhi Delhi India

Freezing of gait (FoG) refers to sudden, relatively brief episodes of gait arrest in Parkinson’s disease, known to manifest in the advanced stages of the condition. Events of freezing are associated with tumbles, traumas, and psychological repercussions, significantly impacting the patient’s quality of life. The use of accelerometer data derived from sensors put on a patient’s body to detect FoG has previously been proposed using convolutional neural network (CNN)-based deep learning algorithms. Here, we combine a CNN + Long short-term Memory (LSTM)-attention model to detect FoG episodes—a first for the detection of FoG using the accelerometer data. CNN facilitates automatic feature extraction from the accelerometer data itself. The output from the CNN is fed to the LSTM network, which is known for capturing sequential information. Further, the attention mechanism introduces relative focus on individual sub-sequences during the training of the LSTM network. The proposed model is made patient independent using adversarial training. The proposed model achieved improvements of +7.60% (without adversarial training) and + 8.36% (with adversarial training) in the sensitivity values over state of the art. This is obtained with little or no compromise in specificity values. The corresponding improvements in accuracy values are +3.81 and +2.23%, respectively. These findings imply that the proposed method can detect FoG gaits satisfactorily and can be effective in achieving accurate monitoring and gait assistance for Parkinson’s disease patients during daily life and rehabilitation therapy. © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2025.

关键词： Accelerometers

来源：评论

学校读者我要写书评

暂无评论

A Novel Eccentric Intrusion Detection Model Based on Recurrent Neural Networks with Leveraging LSTM

引用

computers, Materials & Continua 2024年第3期78卷 3089-3127页

作者： Navaneetha Krishnan Muthunambu Senthil Prabakaran Balasubramanian Prabhu Kavin Kishore Senthil Siruvangur Kavitha Chinnadurai Jehad Ali Department of Computer Applications Karpagam College of EngineeringCoimbatore641032India Department of Computer Science and Engineering Karpagam College of EngineeringCoimbatore641032India Department of Data Science and Business Systems College of Engineering and TechnologySRM Institute of Science and TechnologyKattankulathur603203India Department of Computer Science and Engineering Sathyabama Institute of Science and TechnologyJeppiaar NagarRajiv Gandhi SalaiChennai600119India Department of AI Convergence Network Ajou UniversitySuwon16499Korea

The extensive utilization of the Internet in everyday life can be attributed to the substantial accessibility of online services and the growing significance of the data transmitted via the ***,this development has expanded the potential targets that hackers might *** adequate safeguards,data transmitted on the internet is significantly more susceptible to unauthorized access,theft,or *** identification of unauthorised access attempts is a critical component of cybersecurity as it aids in the detection and prevention of malicious *** research paper introduces a novel intrusion detection framework that utilizes Recurrent Neural Networks(RNN)integrated with Long Short-Term Memory(LSTM)*** proposed model can identify various types of cyberattacks,including conventional and distinctive *** networks,a specific kind of feedforward neural networks,possess an intrinsic memory *** Neural Networks(RNNs)incorporating Long Short-Term Memory(LSTM)mechanisms have demonstrated greater capabilities in retaining and utilizing data dependencies over extended *** such as data types,training duration,accuracy,number of false positives,and number of false negatives are among the parameters employed to assess the effectiveness of these models in identifying both common and unusual *** are utilised in conjunction with LSTM to support human analysts in identifying possible intrusion events,hence enhancing their decision-making capabilities.A potential solution to address the limitations of Shallow learning is the introduction of the Eccentric Intrusion Detection *** model utilises Recurrent Neural Networks,specifically exploiting LSTM *** proposed model achieves detection accuracy(99.5%),generalisation(99%),and false-positive rate(0.72%),the parameters findings reveal that it is superior to state-of-the-art techniques.

关键词： Cybersecurity intrusion detection machine learning leveraging long short-term memory(LLSTM) CICIDS2019 dataset innovative cyberattacks

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：