检索结果-内蒙古大学图书馆

TransEffiVisNet – an image captioning architecture for auditory assistance for the visually impaired

Multimedia Tools and Applications 2024年 1-20页

作者： R, Harshitha B, Lakshmipriya Krishnamurthy, Vallidevi School of Computer Science and Engineering Vellore Institute of Technology Chennai Campus Chennai India School of Engineering Jawaharlal Nehru University New Delhi India

Insights derived out of image captioning systems have potential applications in real life, including providing auditory assistance for the visually impaired. This paper proposes TransEffiVisNet, a novel image captioning system embedded with text-to-speech generation, which can be deployed to aid individuals with visual impairments. The proposed model combines a convolutional neural network (CNN) and a transformer-based encoder and decoder to generate descriptive captions for the visual content provided as an image input. A comparative study was conducted to assess the performance of the proposed model formulated using five prominent CNN models—VGG16, ResNet50, EfficientNetB0, EfficientNetB1, and Inception V3, while maintaining the same transformer architecture. The image captioning model with EfficientNetB1 showcased phenomenal results compared to four other architectures and hence preferred to be used in TransEffiVisNet. Integration of Google Text-to-Speech technology to convert captions into speech further enhances the functionality of the model which in future can be deployed for visually impaired assistance. © The Author(s), under exclusive licence to Springer science+Business Media, LLC, part of Springer Nature 2024.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Explaining the better generalization of label distribution learning for classification

引用

science China(Information sciences) 2025年第5期68卷 348-361页

作者： Jing WANG Xin GENG School of Computer Science and Engineering Southeast University Key Laboratory of New Generation Artificial Intelligence Technology and Its Interdisciplinary Applications(Southeast University) Ministry of Education

Label distribution learning(LDL) has shown advantages over traditional single-label learning(SLL) in many realworld applications, but its superiority has not been theoretically understood. In this paper, we attempt to explain why LDL generalizes better than SLL. Label distribution has rich supervision information such that an LDL method can still choose the sub-optimal label from label distribution even if it neglects the optimal one. In comparison, an SLL method has no information to choose from when it fails to predict the optimal label. The better generalization of LDL can be credited to the rich information of label distribution. We further establish the label distribution margin theory to prove this explanation; inspired by the theory,we put forward a novel LDL approach called LDL-LDML. In the experiments, the LDL baselines outperform the SLL ones, and LDL-LDML achieves competitive performance against existing LDL methods, which support our explanation and theories in this paper.

关键词： label distribution learning classification generalization label distribution margin label ambiguity neural network

来源：评论

学校读者我要写书评

暂无评论

Reinforcement learning based energy efficient robot relay for unmanned aerial vehicles against smart jamming

引用

science China(Information sciences) 2022年第1期65卷 234-246页

作者： Xiaozhen LU Jingfang JIE Zihan LIN Liang XIAO Jin LI Yanyong ZHANG Department of Information and Communication Engineering Xiamen University School of Computer Science and Educational Software Guangzhou University School of Computer Science and Technology University of Science and Technology of China

Unmanned aerial vehicles(UAVs) with limited energy resources, severe path loss, and shadowing to the ground base stations are vulnerable to smart jammers that aim to degrade the UAV communication performance and exhaust the UAV energy. The UAV anti-jamming communication performance, such as the outage probability, degrades if the robot relay is not aware of the jamming policies and the UAV network topology. In this paper, we propose a robot relay scheme for UAVs against smart jamming, which combines reinforcement learning with a function approximation approach named tile coding, to jointly optimize the robot moving distance and relay power with the unknown jamming channel states and locations. The robot mobility and relay policy are chosen based on the received jamming power, the robot received signal quality,location and energy consumption, and the bit error rate of the UAV messages. We also present a deep reinforcement learning version for the robot with sufficient computing resources. It uses three deep neural networks to choose the robot mobility and relay policy with reduced sample complexity, so as to avoid exploring dangerous policies that lead to the high outage probability of the UAV messages. The network architecture of the three networks is designed with fully connected layers instead of convolutional layers to reduce the computational complexity, which is analyzed by theoretical analyses. We provide the performance bound of the proposed schemes in terms of the bit error rate, robot energy consumption and utility based on a game-theoretic study. Simulation results show that the performance of our proposed relay schemes,including the bit error rate, the outage probability, and the robot energy consumption outperforms the existing schemes.

关键词： unmanned aerial vehicles relay jamming game theory reinforcement learning

来源：评论

学校读者我要写书评

暂无评论

A Scalable and Secure Facial Recognition Attendance System Leveraging CNN Algorithms for Real-Time Accuracy and Efficiency

A Scalable and Secure Facial Recognition Attendance System L...

引用

2025 IEEE International Students' Conference on Electrical, Electronics and computer science, SCEECS 2025

作者： Kumar, Aakash Sharma, Jiya Renjith, P.N. Vellore Institute of Technology School of Computer Science and Engineering India

ISBN: (纸本)9798331529833

The facial recognition attendance system represents a modern approach to attendance tracking, designed to improve efficiency, accuracy, and security in various organizational settings. Advanced biometric technologies are leveraged to capture and encode facial features, creating unique digital identifiers for each individual. The methodology employs cutting-edge facial recognition algorithms powered by convolutional neural networks (CNNs) with a 99.5% recognition accuracy rate. These algorithms extract and encode facial vectors with sub-millisecond latency, which are then compared in real-time against a secure, encrypted database of preregistered individuals. Upon a successful match, the system automatically records the attendance, ensuring an accurate and efficient process. With an intuitive web interface built on ReactJS and a backend supported by *** and MongoDB, the system provides seamless access to attendance records and administrative functionalities. The entire attendance recording process is optimized for speed, typically completing within two seconds per individual. Additionally, advanced anti-spoofing measures and integration support with existing HR management systems redefine traditional attendance management practices with a scalable, robust, and secure solution. (Abstract) © 2025 IEEE.

关键词： Efficiency

来源：评论

学校读者我要写书评

暂无评论

A Decoupling-Type Strategy for the Allen-Cahn Equation on Curved Surfaces

引用

Journal of Mathematical Research with Applications 2025年第1期45卷 73-83页

作者： Xiaoman XIE Qing PAN School of Computer and Communication Engineering Changsha University of Science & Technology

In this paper, we construct an efficient decoupling-type strategy for solving the Allen-Cahn equation on curved surfaces. It is based on an FEM-EIEQ(Finite Element Method and explicit-Invariant Energy Quadratization) fully discrete scheme with unconditional energy stability. Spatially the FEM is adopted, using a triangular mesh discretization strategy that can be adapted to complex regions. Temporally, the EIEQ approach is considered, which not only linearizes the nonlinear potential but also gives a new variable that we combine with the nonlocal splitting method to achieve the fully decoupled computation. The strategy can successfully transform the Allen-Cahn system into some completely independent algebraic equations and linear elliptic equations with constant coefficients, we only need to solve these simple equations at each time step. Moreover, we conducted some numerical experiments to demonstrate the effectiveness of the strategy.

关键词： FEM-EIEQ Allen-Cahn equation surface

来源：评论

学校读者我要写书评

暂无评论

Dynamic Event-Triggered Consensus Control for Input Constrained Multi-Agent Systems With a Designable Minimum Inter-Event Time

引用

IEEE/CAA Journal of Automatica Sinica 2024年第3期11卷 649-660页

作者： Meilin Li Yue Long Tieshan Li Hongjing Liang C.L.Philip Chen the School of Automation Engineering University of Electronic Science and Technology of ChinaChengdu 611731China IEEE the Yangtze Delta Region Institute University of Electronic Science and Technology of ChinaHuzhou 313000 the Laboratory of Electromagnetic Space Cognition and Intelligent Control Beijing 100089China the School of Computer Science and Engineering South China University of TechnologyGuangzhou 510006

This paper investigates the consensus control of multi-agent systems(MASs) with constrained input using the dynamic event-triggered mechanism(ETM).Consider the MASs with small-scale networks where a centralized dynamic ETM with global information of the MASs is first ***,a distributed dynamic ETM which only uses local information is developed for the MASs with large-scale *** is shown that the semi-global consensus of the MASs can be achieved by the designed bounded control protocol where the Zeno phenomenon is eliminated by a designable minimum inter-event *** addition,it is easier to find a trade-off between the convergence rate and the minimum inter-event time by an adjustable ***,the results are extended to regional consensus of the MASs with the bounded control *** simulations show the effectiveness of the proposed approach.

关键词： Constrained input designable minimum inter-event time directed communication topology dynamic event-triggered mechanism MASs consensus control

来源：评论

学校读者我要写书评

暂无评论

Abnormal Traffic Detection for Internet of Things Based on an Improved Residual Network

引用

computers, Materials & Continua 2024年第6期79卷 4433-4448页

作者： Tingting Su Jia Wang Wei Hu Gaoqiang Dong Jeon Gwanggil School of Computer Science and Technology Xinjiang UniversityUrumqi830046China School of Cyber Science and Engineering Jinling Institute of Technology CityNanjing210000China College of Information Technology Incheon National UniversityIncheon22012Korea

Along with the progression of Internet of Things(IoT)technology,network terminals are becoming continuously more *** has been widely applied in various scenarios,including urban infrastructure,transportation,industry,personal life,and other socio-economic *** introduction of deep learning has brought new security challenges,like an increment in abnormal traffic,which threatens network *** feature extraction leads to less accurate classification *** abnormal traffic detection,the data of network traffic is high-dimensional and *** data not only increases the computational burden of model training but also makes information extraction more *** address these issues,this paper proposes an MD-MRD-ResNeXt model for abnormal network traffic *** fully utilize the multi-scale information in network traffic,a Multi-scale Dilated feature extraction(MD)block is *** module can effectively understand and process information at various scales and uses dilated convolution technology to significantly broaden the model’s receptive *** proposed Max-feature-map Residual with Dual-channel pooling(MRD)block integrates the maximum feature map with the residual *** module ensures the model focuses on key information,thereby optimizing computational efficiency and reducing unnecessary information *** results show that compared to the latest methods,the proposed abnormal traffic detection model improves accuracy by about 2%.

关键词： Abnormal network traffic deep learning residual network multi-scale feature extraction max-feature-map

来源：评论

学校读者我要写书评

暂无评论

Adaptive Marine Predator Optimization Algorithm(AOMA)-Deep Supervised Learning Classification(DSLC)based IDS framework for MANET security

引用

Intelligent and Converged Networks 2024年第1期5卷 1-18页

作者： M.Sahaya Sheela A.Gnana Soundari Aditya Mudigonda C.Kalpana K.Suresh K.Somasundaram Yousef Farhaoui Department of Electronics and Communication Engineering Vel Tech Rangarajan Dr.Sagunthala R&D Institute of Science and TechnologyChennai 600069India Department of Computer Science and Engineering Saveetha School of EngineeringSIMATSChennai 602105India JNIAS School of Planning and Architecture Hyderabad 500034India Department of Computer Science and Engineering NPR College of Engineering and TechnologyNatham Dindigul 624401India Department of Computer Science and Engineering PSNA College of Engineering and TechnologyPoolangulathupatti 620009India Department of Computer Science and Engineering Sri Muthukumaran Institute of TechnologyChennai 600069India Department of Computer Science Moulay Ismail UniversityMeknes 5003Morocco

Due to the dynamic nature and node mobility,assuring the security of Mobile Ad-hoc Networks(MANET)is one of the difficult and challenging tasks *** MANET,the Intrusion Detection System(IDS)is crucial because it aids in the identification and detection of malicious attacks that impair the network’s regular *** machine learning and deep learning methodologies are used for this purpose in the conventional works to ensure increased security of ***,it still has significant flaws,including increased algorithmic complexity,lower system performance,and a higher rate of ***,the goal of this paper is to create an intelligent IDS framework for significantly enhancing MANET security through the use of deep learning ***,the min-max normalization model is applied to preprocess the given cyber-attack datasets for normalizing the attributes or fields,which increases the overall intrusion detection performance of ***,a novel Adaptive Marine Predator Optimization Algorithm(AOMA)is implemented to choose the optimal features for improving the speed and intrusion detection performance of ***,the Deep Supervise Learning Classification(DSLC)mechanism is utilized to predict and categorize the type of intrusion based on proper learning and training *** evaluation,the performance and results of the proposed AOMA-DSLC based IDS methodology is validated and compared using various performance measures and benchmarking datasets.

关键词： Intrusion Detection System(IDS) Security Mobile Ad-hoc Network(MANET) min-max normalization Adaptive Marine Predator Optimization Algorithm(AOMA) Deep Supervise Learning Classification(DSLC)

来源：评论

学校读者我要写书评

暂无评论

Hybrid Optimization Algorithm for Handwritten Document Enhancement

引用

computers, Materials & Continua 2024年第3期78卷 3763-3786页

作者： Shu-Chuan Chu Xiaomeng Yang Li Zhang Václav Snášel Jeng-Shyang Pan College of Computer Science and Engineering Shandong University of Science and TechnologyQingdao266590China School of Information and Control Engineering Qingdao University of TechnologyQingdao266520China Faculty of Electrical Engineering and Computer Science VŠB-TU OstravaOstrava70080Czech Republic Department of Information Management Chaoyang University of TechnologyTaichung41349Taiwan

The Gannet Optimization Algorithm (GOA) and the Whale Optimization Algorithm (WOA) demonstrate strong performance;however, there remains room for improvement in convergence and practical applications. This study introduces a hybrid optimization algorithm, named the adaptive inertia weight whale optimization algorithm and gannet optimization algorithm (AIWGOA), which addresses challenges in enhancing handwritten documents. The hybrid strategy integrates the strengths of both algorithms, significantly enhancing their capabilities, whereas the adaptive parameter strategy mitigates the need for manual parameter setting. By amalgamating the hybrid strategy and parameter-adaptive approach, the Gannet Optimization Algorithm was refined to yield the AIWGOA. Through a performance analysis of the CEC2013 benchmark, the AIWGOA demonstrates notable advantages across various metrics. Subsequently, an evaluation index was employed to assess the enhanced handwritten documents and images, affirming the superior practical application of the AIWGOA compared with other algorithms.

关键词： Metaheuristic algorithm gannet optimization algorithm hybrid algorithm handwritten document enhancement

来源：评论

学校读者我要写书评

暂无评论

Crop weed separation through image-level segmentation: an ensemble of modified U-Net and encoder–decoder

引用

Neural Computing and Applications 2025年第16期37卷 9887-9910页

作者： Ganapathy, Sannasi Srinivasan, Srinitish Bhopal462 002 India School of Computer Science and Engineering Vellore Institute of Technology Chennai600 127 India

Crop weed segmentation is one of the most challenging tasks in the field of computer vision. This is because, unlike other object detection or segmentation tasks, crop and weed are similar in terms of spectral features, shape, dimensions, etc. For precision agriculture to flourish in terms of smart spraying of crops, efficient systems to distinguish between crop and weed are the need of the hour, which if precise, will take a huge step toward solving the issue of food scarcity. To tackle this issue, we propose new ensemble architecture of two models—a U-Net with a modified backbone and an encoder–decoder. These networks learn to distinguish between soil and crop and soil and weed, respectively, whose ensemble gives state-of-the-art results on pixel-wise annotations of combined crop and weed images. Moreover, it also learns that the model captures un-annotated features since each component of the architecture learnt either crop or weed features to high precision. Finally, the proposed architecture is compared with the U-Net and SegNet, which are popular segmentation networks, and consistently achieved better results. © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2025.

关键词： Semantic Segmentation

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：