检索结果-内蒙古大学图书馆

Recognition of American sign language using modified deep residual CNN with modified canny edge segmentation

Multimedia Tools and Applications 2025年 1-28页

作者： Hariharan, U. Devarajan, Devikanniga Kumar, P. Santhosh Rajkumar, K. Meena, M. Akilan, T. Department of Computer Science and Engineering Apex Institute of Technology Chandigarh University Punjab Mohali140413 India Department of Computer Science and Engineering GITAM University Karnataka Bengaluru561203 India Department of Information Technology Faculty of Engineering and technology SRM Institute of Science and Technology Ramapuram Tamil Nadu Chennai600089 India Department of Data Science and Business Systems School of Computing Faculty of Engineering and Technology SRM Institute of Science and Technology Kattankulathur Tamil Nadu Chennai603203 India Department of Computer Science and Engineering Sri Sairam Engineering College Tamil Nadu Tambaram602109 India Research Coordinator Dept. of IQAC Amity University Kolkata West Bengal 700135 India

American Sign Language (ASL) recognition aims to recognize hand gestures, and it is a crucial solution to communicating between the deaf community and hearing people. However, existing sign language recognition algorithms still have some drawbacks, such as difficulty recognizing hand movements and low recognition accuracy for most sign language recognition. A Modified Convolutional Neural Network (MCNN) deep residual 101 classifier-based American Sign Language identification system has been developed to address this problem. There are three main parts presents in the method. The first part is pre-processing the images to remove the noise, enhance the contrast of the picture, adjust the contrast level and smoothen the picture using various filters. The second part is the segmentation, and it’s used to partition the image using a modified canny edge detection method by removing all weak edges present in the image. Finally, classification will be done using the Modified CNN deep residual 101 classifiers. By classifying the image, the American Sign Language is accurately identified. This process is conducted through images. The outcome shows that the suggested approach has a 0.97% accuracy and a 0.05% False Positive Rate. Other CNN architectures such as resNet 50 and resNet 18 reached 0.90% and 0.80% accuracy, respectively, which is lower than our proposed method. In addition, the resNet 101 classifier effectively recognizes the difficult hand gestures through the image data. It obtains high recognition accuracy for 36 signs is 0 to 9 numbers and a to z alphabets from American Sign Language. © The Author(s), under exclusive licence to Springer science+Business Media, LLC, part of Springer Nature 2025.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Transparency vs Explanation of Machine Learning Algorithms: Perspectives from Recent Legal Proceedings 23rd

Transparency vs Explanation of Machine Learning Algorithms: ...

引用

23rd EPIA Conference on Artificial Intelligence, EPIA 2024

作者： Nnawuchi, Uchenna George, Carlisle Kammueller, Florian ALERT Research Group Department of Computer Science Faculty of Science and Technology Middlesex University London United Kingdom

ISBN: (纸本)9783031734960

The extensive use of Machine Learning (ML) algorithms has significantly altered decision-making processes, societal structures, and power dynamics. These algorithms, known for their efficiency, are increasingly making critical decisions. However, concerns about their opacity and biases persist. Some legal scholars and data scientists argue that the GDPR includes a right to explanation to address ML opacity, while others advocate for transparency as a better solution. This study explores the debate between transparency and explanation within the GDPR’s regulatory framework, focusing on Articles 15, 22 and recent EU court rulings. By examining judicial developments and scholarly perspectives, the study highlights the GDPR’s efforts to promote transparency and accountability in automated decision-making. Despite the GDPR’s provisions, the lack of an explicit right to explanation has led to ongoing legal and ethical discussions. This research aims to provide insights into how transparency, explanation, and individual rights intersect in the realm of algorithmic decision-making under the GDPR. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

关键词： Adversarial machine learning

来源：评论

学校读者我要写书评

暂无评论

LoRA-drop: Efficient LoRA Parameter Pruning based on Output Evaluation 31

LoRA-drop: Efficient LoRA Parameter Pruning based on Output ...

引用

31st International Conference on Computational Linguistics, COLING 2025

作者： Zhou, Hongyun Lu, Xiangyu Xu, Wang Zhu, Conghui Zhao, Tiejun Yang, Muyun Faculty of Computing Harbin Institute of Technology China Department of Computer Science & Technology Tsinghua University Beijing China

ISBN: (纸本)9798891761964

Low-Rank Adaptation (LoRA) is currently the most commonly used Parameter-efficient fine-tuning (PEFT) method. However, it still faces high computational and storage costs to models with billions of parameters. Most previous studies have tackled this issue by using pruning techniques. Nonetheless, these efforts only analyze LoRA parameter features to evaluate their importance, such as parameter count, size, and gradient. In fact, the output of LoRA directly impacts the fine-tuned model. Preliminary experiments indicate that a fraction of LoRA possesses significantly high output values, substantially influencing the layer output. Motivated by the observation, we propose LoRA-drop. Concretely, LoRA-drop evaluates the importance of LoRA based on the LoRA output. Then we retain LoRA for important layers and the other layers share the same LoRA. We conduct abundant experiments with models of different scales on NLU and NLG tasks. Results demonstrate that LoRA-drop can achieve performance comparable to full fine-tuning and LoRA while retaining 50% of the LoRA parameters on average. © 2025 Association for Computational Linguistics.

关键词： Computational linguistics

来源：评论

学校读者我要写书评

暂无评论

Network Bandwidth Optimization Using Software Defined Networking (SDN) 5

Network Bandwidth Optimization Using Software Defined Networ...

引用

5th IEEE Middle East and North Africa COMMunications Conference, MENACOMM 2025

作者： Haidar, Gaby Abou Zangana, Ali Noaman Khasro Owayjan, Michel Daou, Roy Abi Zeid Achkar, Roger American University of Science and Technology Faculty of Engineering and Computer Science Beriut Lebanon France La Sagesse University Faculty of Engineering Beirut Lebanon

ISBN: (纸本)9798331519957

This paper provides a detailed comparison of traditional networking architectures and Software Defined Networking (SDN) approaches, with a focus on bandwidth optimization and traffic management. Simulations were conducted under various networking scenarios, utilizing Cisco Packet Tracer for traditional networks and Mininet with the OpenDaylight (ODL) controller for SDN. The evaluated scenarios included single, dual, and triple PC setups, as well as high-Traffic conditions. Traditional networks demonstrated packet loss rates of up to 5%, limited bandwidth utilization (~2.443%), and increasing latency as more devices were added. In contrast, SDN-based setups showed no packet loss and significantly higher bandwidth utilization, reaching up to 94.478%. This study highlights the potential of SDN to enhance network efficiency and scalability, addressing a gap in the literature through scenario-based comparative analysis. Future research directions include exploring large-scale topologies, security enhancements, and advanced optimization techniques. © 2025 IEEE.

关键词： Packet loss

来源：评论

学校读者我要写书评

暂无评论

Machine Translation Performance for Low-Resource Languages: A Systematic Literature Review

引用

IEEE Access 2025年 13卷 72486-72505页

作者： Tafa, Taofik O. Hashim, Siti Zaiton Mohd Othman, Mohd Shahizan Alhussian, Hitham Nasser, Maged Abdulkadir, Said Jadid Huspi, Sharin Hazlin Adeyemo, Sarafa O. Bena, Yunusa Adamu Universiti Teknologi Malaysia Faculty of Computing Johor Bahru81310 Malaysia Department of Computer Science at Gusau Gusau632101 Nigeria Universiti Teknologi PETRONAS Department of Computer and Information Sciences Perak Seri Iskandar32610 Malaysia Kebbi State University of Science and Technology Faculty of Engineering Aliero Nigeria

Machine translation (MT) for low-resource languages continues to face significant challenges because of limited digital resources and parallel corpora, despite remarkable developments in neural machine translation (NMT). Addressing these challenges requires a thorough review of existing research to identify effective strategies and methods. To achieve this, a systematic literature review (SLR) is conducted following PRISMA guidelines and systematically analysing studies published in various academic databases in the last five years (between 2020 and 2024). A total of 69 relevant articles were examined to evaluate the performance of MT, explore persistent challenges and assess the effectiveness of proposed or used solutions. The analysis shows that while NMT has emerged as the predominant approach, its effectiveness is often reduced by the scarcity of training data and the structural complexity of low-resource languages. Strategies such as active learning, data augmentation, multilingual models and transfer learning are identified as critical for improving translation performance. Additionally, emerging research trends, including data pre-processing, optimization of decoder and rule-based approach demonstrate promising directions for addressing existing limitations. In terms of evaluation, most of the studies used Character n-gram F-score (ChrF), Translation Edit Rate (TER), Metric for Evaluation of Translation with Explicit Ordering (METEOR), Word Error Rate (WER) and Bilingual Evaluation Underscore (BLEU) as techniques’ validation metrics. This review provides a detailed evaluation of the current state of MT for low-resource languages and emphasizes the need for further research into underrepresented languages and the development of comprehensive datasets. © 2013 IEEE.

关键词： computer aided language translation

来源：评论

学校读者我要写书评

暂无评论

Deep Learning for Cross-Domain Few-Shot Visual Recognition: A Survey

引用

ACM Computing Surveys 2025年第8期57卷 1-37页

作者： Xu, Huali Zhi, Shuaifeng Sun, Shuzhou Patel, Vishal Liu, Li College of Computer Science Nankai University Tianjin China College of Electronic Science and Technology National University of Defense Technology Changsha China Faculty of Information Technology and Electrical Engineering University of Oulu Oulu Finland Johns Hopkins University Baltimore United States

While deep learning excels in computer vision tasks with abundant labeled data, its performance diminishes significantly in scenarios with limited labeled samples. To address this, Few-shot learning (FSL) enables models to perform the target tasks with very few labeled examples by leveraging prior knowledge from related tasks. However, traditional FSL assumes that both the related and target tasks come from the same domain, which is a restrictive assumption in many real-world scenarios where domain differences are common. To overcome this limitation, Cross-domain few-shot learning (CDFSL) has gained attention, as it allows source and target data to come from different domains and label spaces. This article presents the first comprehensive review of Cross-domain Few-shot Learning (CDFSL), a field that has received less attention compared to traditional FSL due to its unique challenges. We aim at providing both a position article and a tutorial for researchers, covering key problems, existing methods, and future research directions. The review begins with a formal definition of CDFSL, outlining its core challenges, followed by a systematic analysis of current approaches, organized under a clear taxonomy. Finally, we discuss promising future directions in terms of problem setups, applications, and theoretical advancements. © 2025 Copyright held by the owner/author(s).

关键词： Zero-shot learning

来源：评论

学校读者我要写书评

暂无评论

New logarithmic step size for stochastic gradient descent

引用

Frontiers of computer science 2025年第1期19卷 109-118页

作者： Mahsa Soheil SHAMAEE Sajad Fathi HAFSHEJANI Zeinab SAEIDIAN Department of Computer Science Faculty of Mathematical ScienceUniversity of KashanKashan 87317-53153Iran Department of Applied Mathematics Shiraz University of TechnologyShiraz 13876-71557Iran Department of Mathematical Sciences University of KashanKashan 87317-53153Iran

In this paper, we propose a novel warm restart technique using a new logarithmic step size for the stochastic gradient descent (SGD) approach. For smooth and non-convex functions, we establish an O(1/√T) convergence rate for the SGD. We conduct a comprehensive implementation to demonstrate the efficiency of the newly proposed step size on the FashionMinst, CIFAR10, and CIFAR100 datasets. Moreover, we compare our results with nine other existing approaches and demonstrate that the new logarithmic step size improves test accuracy by 0.9% for the CIFAR100 dataset when we utilize a convolutional neural network (CNN) model.

关键词： stochastic gradient descent logarithmic step size warm restart technique

来源：评论

学校读者我要写书评

暂无评论

Adaptive hybrid enhanced flower pollination-based simulated annealing scheme for task scheduling in edge-cloud continuum

引用

Neural Computing and Applications 2025年 1-26页

作者： Dankolo, Nasiru Muhammad Radzi, Nor Haizan Mohamed Mustaffa, Noorfa Haszlinna Gabi, Danlami Faculty of Computing University Teknologi Malaysia Johor Bahru81310 Malaysia Department of Computer Science Kebbi State University of Science and Technology P.M.B Aliero1144 Nigeria

In the rapidly advancing edge-cloud continuum computing, efficient task scheduling is crucial for optimizing the performance of large-scale and latency-sensitive applications. However, existing metaheuristic techniques often struggle with slow convergence, high computational complexity, and an imbalance between local and global searches, which limit their scalability and real-time adaptability. To address these challenges, this study aims to develop an efficient task scheduling algorithm, the Adaptive Hybrid Enhanced Flower Pollination-Based Simulated Annealing (AHEFPA-SA) algorithm. The AHEFPA-SA algorithm integrates simulated annealing with adaptive mechanisms, employing a chaotic cycle map to improve initial solutions and adaptive inertia weight to control levy flight distance, thereby reducing computational complexity and accelerating convergence. Implemented within the EdgeCloudSim simulator, the algorithm is evaluated against benchmark algorithms across key performance metrics including makespan, execution cost, and resource utilization, the proposed algorithm demonstrates significant improvements. The results show up to 22% reduction in makespan on edge servers and 20% on cloud servers, along with a 16% reduction in execution cost and a 13% improvement in resource utilization. These findings highlight the adaptability and real-time efficiency of the AHEFPA-SA algorithm, making it well-suited for dynamic, resource-constrained edge-cloud environments. © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2025.

关键词： Simulated annealing

来源：评论

学校读者我要写书评

暂无评论

Phasmatodea Population Evolution Algorithm Based on Spiral Mechanism and Its Application to Data Clustering

引用

computers, Materials & Continua 2025年第4期83卷 475-496页

作者： Jeng-Shyang Pan Mengfei Zhang Shu-Chuan Chu Xingsi Xue Václav Snášel College of Computer Science and Engineering Shandong University of Science and TechnologyQingdao266590China School of Artificial Intelligence Nanjing University of Information Science and TechnologyNanjing210044China Department of Information Management Chaoyang University of TechnologyTaichung41349TaiwanChina Fujian Provincial Key Laboratory of Big Data Mining and Applications Fujian University of TechnologyFuzhou350118China Faculty of Electrical Engineering and Computer Science VŠB-Technical University of OstravaOstrava70833Czech Republic

Data clustering is an essential technique for analyzing complex datasets and continues to be a central research topic in data *** clustering algorithms,such as K-means,are widely used due to their simplicity and *** paper proposes a novel Spiral Mechanism-Optimized Phasmatodea Population Evolution Algorithm(SPPE)to improve clustering *** SPPE algorithm introduces several enhancements to the standard Phasmatodea Population Evolution(PPE)***,a Variable Neighborhood Search(VNS)factor is incorporated to strengthen the local search capability and foster population ***,a position update model,incorporating a spiral mechanism,is designed to improve the algorithm’s global exploration and convergence ***,a dynamic balancing factor,guided by fitness values,adjusts the search process to balance exploration and exploitation *** performance of SPPE is first validated on CEC2013 benchmark functions,where it demonstrates excellent convergence speed and superior optimization results compared to several state-of-the-art metaheuristic *** further verify its practical applicability,SPPE is combined with the K-means algorithm for data clustering and tested on seven *** results show that SPPE-K-means improves clustering accuracy,reduces dependency on initialization,and outperforms other clustering *** study highlights SPPE’s robustness and efficiency in solving both optimization and clustering challenges,making it a promising tool for complex data analysis tasks.

关键词： Phasmatodea population evolution algorithm data clustering meta-heuristic algorithm

来源：评论

学校读者我要写书评

暂无评论

Neutrosophic Sets and Systems: A Decade of Scientific Contribution and Growth

引用

Neutrosophic Sets and Systems 2025年 81卷 438-465页

作者： Saqlain, Muhammad Merigó, José M. Kumam, Poom Department of Mathematics King Mongkut’s University of Technology Thonburi Bangkok10140 Thailand School of Computer Science Faculty of Engineering and Information Technology University of Technology Sydney 81 Broadway UltimoNSW2007 Australia

Neutrosophic Sets and Systems (NSS) has become an important Journal for neutrosophic theory and its applications in uncertainty modeling and decision sciences. In 2023, NSS celebrated its 10th anniversary, marking a decade of contributions to scientific literature. Motivated by this event, this paper provides a critical bibliometric analysis of NSS’s evolution as a leading academic journal in neutrosophy. This study explores the publication trends, citation impact, most productive authors, institutions, and global collaboration networks of NSS. The study maps bibliographic coupling, keyword co-occurrence and co-citation analysis using data from the Scopus database, secondary search using NSS webpage, and visualization tools like the VOSviewer. The findings identify Neutrosophic Sets, Decision-Making, and Fuzzy Sets as the most frequently studied keywords that highlight the increasing integration of neutrosophic approaches into multi-criteria decision-making (MCDM) and soft computing. The results indicate the most significant areas of research that are topology, soft set theory, hypersoft sets, and uncertainty modeling. The findings highlight the NSS increasing international collaboration, with contributions from America, Middle East, and Asia. Additionally, the analysis explores NSS’s alignment with sustainable development goals (SDGs), revealing notable contributions to good health and well-being (SDG 3), quality education (SDG 4), and peace, justice, and strong institutions (SDG 16). These contributions highlight how neutrosophic sets and systems is valuable across many researchers, from public health to governance to education. Overall, this paper serves as a fundamental study for researchers looking to explore emerging trends and novel applications of neutrosophic theories across diverse scientific and practical domains. © 2025, University of New Mexico. All rights reserved.

关键词： Bibliographies

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：