检索结果-内蒙古大学图书馆

arXiv 2025年

作者： Wang, Haodong Zhou, Qihua Hong, Zicong Guo, Song Hong Kong University of Science and Technology Hong Kong Hong Kong College of Computer Science and Software Engineering Shenzhen University Shenzhen China

The mixture of experts (MoE) model is a sparse variant of large language models (LLMs), designed to hold a better balance between intelligent capability and computational overhead. Despite its benefits, MoE is still too expensive to deploy on resource-constrained edge devices, especially with the demands of on-device inference services. Recent research efforts often apply model compression techniques, such as quantization, pruning and merging, to restrict MoE complexity. Unfortunately, due to their predefined static model optimization strategies, they cannot always achieve the desired quality-overhead trade-off when handling multiple requests, finally degrading the on-device quality of service. These limitations motivate us to propose the D2MoE, an algorithm-system co-design framework that matches diverse task requirements by dynamically allocating the most proper bit-width to each expert. Specifically, inspired by the nested structure of matryoshka dolls, we propose the matryoshka weight quantization (MWQ) to progressively compress expert weights in a bit-nested manner and reduce the required © 2025, CC BY.

关键词： Quality of service

来源：评论

学校读者我要写书评

暂无评论

Video Surveillance-Based Intrusion Detection System in Edge Cloud Environment 3rd

Video Surveillance-Based Intrusion Detection System in Edge ...

引用

3rd International Conference on Emergent Converging Technologies and Biomedical Systems, ETBS 2023

作者： Sharma, Annu Devasenapathy, Deepa Raja, M. Shadrach, Finney Daniel Shirgire, Anil Arun, R. Yau, Thomas Moh Shan Department of Computer Science and Applications Rajarajeshwari College of Engineering Karnataka Bengaluru560074 India Instructor-II/Computing and Software Engineering U.A. Whitaker College of Engineering Florida Gulf Coast University Fort Myers United States Department of Computer Science and Engineering CMR Institute of Technology Karnataka Bengaluru560037 India Department of Electronics and Communication Engineering KPR Institute of Engineering and Technology Tamil Nadu Coimbatore641407 India Department of Civil Engineering Dr. D Y Patil Institute of Technology Maharashtra Pimpri Pune411018 India Department of Computer Science and Engineering Hindusthan Institute of Technology Coimbatore India Director of Centre of University Courses and Innovative Learning University of Technology Sarawak Jalan Universiti Sibu Sarawak96000 Malaysia

ISBN: (纸本)9789819986453

Video surveillance is mostly used for monitoring distant videos and the cloud memory in the system provides more secure to the data. But, in the existing methodologies poor intrusion detection accuracy as well as poor performance rate is considered as the main demerits. So, to overcome the above-mentioned demerits, a novel video surveillance-based Intrusion Detection System (IDS) (VS-IDS) in the edge cloud environment via deep learning method has been developed in this research. Different algorithms are used for object detection and intrusion detection. Subsequently, the proposed model can be implemented and verified in NVIDIA Jetson TX2 platform. The proposed model attains better rate of performance than the existing models. The Performance of the proposed model can be validated with accuracy, precision, recall, MSE and F-measure. Higher performance rate indicates that the system can detect intrusion more accurately than the existing mechanisms. 99% of detection accuracy can be attained through the proposed model. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024.

关键词： Intrusion detection

来源：评论

学校读者我要写书评

暂无评论

Spark Spectrum Allocation for D2D Communication in Cellular Networks

引用

computers, Materials & Continua 2022年第3期70卷 6381-6394页

作者： Tanveer Ahmad Imran Khan Azeem Irshad Shafiq Ahmad Ahmed TSoliman Akber Abid Gardezi Muhammad Shafiq Jin-Ghoo Choi Department of Electrical and Electronic Engineering Auckland University of TechnologyAuckland1010New Zealand Department of Electrical Engineering University of Engineering and TechnologyPeshawar25120Pakistan Department of Computer Science and Software Engineering International Islamic University IslamabadIslamabad44000Pakistan Industrial Engineering Department College of EngineeringKing Saud UniversityP.O.Box 800Riyadh11421Saudi Arabia Department of Computer Science COMSATS University IslamabadIslamabadPakistan Department of Information and Communication Engineering Yeungnam UniversityGyeongsan38541Korea

The device-to-device(D2D)technology performs explicit communication between the terminal and the base station(BS)terminal,so there is no need to transmit data through the BS *** establishment of a short-distance D2D communication link can greatly reduce the burden on the BS *** present,D2D is one of the key technologies in 5G technology and has been studied in depth.D2D communication reuses the resources of cellular users to improve system key parameters like utilization and ***,repeated use of the spectrum and coexistence of cellular users can cause co-channel *** at the interference problem under the constraint of fair resource allocation and improving the system throughput,this paper proposes an effective resource optimization scheme based on the firework *** main idea is to expand the weighted sum rate and convert the allocated resource expression into fireworks to determine the correlation *** simulation results show that,compared with the existing scheme,this scheme improves system performance by reducing interference.

关键词： D2D communication 5G technology spectrum allocation

来源：评论

学校读者我要写书评

暂无评论

A review of neuro-symbolic AI integrating reasoning and learning for advanced cognitive systems

Intelligent Systems with Applications

引用

Intelligent Systems with Applications 2025年 26卷

作者： Nawaz, Uzma Anees-ur-Rahaman, Mufti Saeed, Zubair Knowledge and Data Science Research Centre Department of Computer and Software Engineering College of Electrical and Mechanical Engineering National University of Science and Technology Islamabad Pakistan Military College of Signals National University of Science and Technology Islamabad Pakistan Department of Electrical & Computer Engineering Texas A&M University College StationTX United States Department of Electrical and Computer Engineering Texas A&M University at Qatar Doha Qatar

Neuro-symbolic AI represents the convergence of two principal paradigms in artificial intelligence: neural networks, which are efficient in data-driven learning, and symbolic reasoning, which offers explainability and logical inference. This hybrid methodology combines the adaptability of neural networks with symbolic AI's interpretability and formal reasoning abilities, which provide a practical framework for advanced cognitive systems. This paper analyzes the present condition of neuro-symbolic AI, emphasizing essential techniques that combine reasoning and learning. We explore models such as Logic Tensor Networks, Differentiable Logic Programs, and Neural Theorem Provers. The study analyzes their impact on the advancement of cognitive systems in natural language processing, robotics, and decision-making. The paper examines the challenges faced by neuro-symbolic AI, such as scalability, integration with multimodal data, and maintaining interpretability without compromising efficiency. By evaluating the strengths and weaknesses of many methodologies, we comprehensively understand the field's development and its potential to revolutionize intelligent systems. In addition, we identify emerging research areas, including the incorporation of ethical frameworks and the development of adaptive dynamic neuro-symbolic systems that respond in real-time. This review aims to guide future research by providing insights into the potential of neuro-symbolic AI to influence the development of the next generation of intelligent, explainable, and adaptive systems. © 2025

关键词： Natural language processing systems

来源：评论

学校读者我要写书评

暂无评论

Blockchain-Enhanced AI Security and Trust Through Auditable Decision-Making and Data Integrity in Modern Industry 2

Blockchain-Enhanced AI Security and Trust Through Auditable ...

引用

2nd IEEE International Conference on Innovations in High-Speed Communication and Signal Processing, IHCSP 2024

作者： Parmar, Digantkumar N. Putha, Sudharshan Maaliw, Renato R. Kasaraneni, Bhavani Prasad Reegu, Faheem Ahmad Byeon, Haewon Silver Oak University Department of Computer Engineering India Salesforce United States Luzon State University Lucban College of Engineering Southern Quezon Philippines United States Jazan University College of Engineering and Computer Science Department of Electrical and Electronics Engineering Saudi Arabia Inje University Department of Ai and Software Gimhae 50834 South Korea

ISBN: (纸本)9798350368949

AI and blockchain can make companies safer and more trustworthy. AI can handle massive quantities of data and make complicated judgments, but its lack of clarity and reliance on centralized data present safety, accountability, and openness concerns. Blockchain ensures data is accurate, clear, and simple to track with its decentralized and unchangeable record. Blockchain can improve AI systems by making decision-making more open and safer for AI data. Blockchain protects AI model data from unauthorized changes. This prevents illicit alterations and improves AI accuracy. Blockchain's autonomy lets parties independently verify AI decision-making, which fosters confidence. This study examines blockchain's primary AI protections. These include open data storage, validated AI models, and trustworthy smart contracts that make decisions automatically. Real-world examples from healthcare, finance, and supply chain management show how blockchain-enhanced AI could address data breaches, algorithmic bias, and machine accountability. These findings demonstrate that adding blockchain to AI systems increases their trustworthiness and makes AI use in modern organizations safer and more open. © 2024 IEEE.

关键词： AI security Artificial intelligence Auditable decision-making Blockchain Data integrity Decentralized systems Immutable ledger Smart contracts Transparent AI Trust in AI

来源：评论

学校读者我要写书评

暂无评论

Exploring Relationships among Bitcoin's Market Price, Energy Consumption and Carbon Dioxide Emissions: A Machine Learning Approach 1

Exploring Relationships among Bitcoin's Market Price, Energy...

引用

1st IEEE Karachi Section Humanitarian Technology Conference, Khi-HTC 2024

作者： Jamali, Aadil Ali, Najma Imtiaz Brohi, Imtiaz Ali Kanasro, Nadeem Ahmed Murad, Muhammad Umar Jamali, Asad Ali University of Sindh IMCS Jamshoro Pakistan Government College University Dept. of Computer Science Hyderabad Pakistan Istanbul Aydin University Dept. of Software Engineering Istanbul Turkey University of Essex Dept. of Economics Colchester United Kingdom

ISBN: (纸本)9798350373042

The cryptocurrency boom has resulted in a substantial rise in energy usage and carbon emissions. The mining process of Bitcoin is notorious for its energy intensive process because it requires huge computational power. The carbon footprint of Bitcoin mining is a growing concern that's why we have to analyze the number of CO2 emissions resulting from the mining procedure. In this study, we have utilized machine learning techniques to examine the number of CO2 emissions generated by Bitcoin mining, analyzed the correlations between Bitcoin price, electricity consumption, and CO2 emissions, and pinpoint the factors during the process of Bitcoin mining that leads to CO2 emissions. We gathered global data set on Bitcoin electricity consumption, prices, and CO2 emissions. We trained supervised machine learning model (Regression) in Python to analyze and investigate the data. Our research findings have great implications for policymakers and researchers worried about the impact of the latest technologies on the environment, and in future Bitcoin policies and practices;they should prioritize sustainability and reduce the carbon footprint of mining operations. In the end, our research will provide valuable insights into the environmental impact of Bitcoin mining, highlight the factors that cause CO2 emissions and underscored the need for further research and action to reduce the carbon footprint of cryptocurrencies. © 2024 IEEE.

关键词： Carbon footprint

来源：评论

学校读者我要写书评

暂无评论

MLRP-KG: Mine Landslide Risk Prediction Based on Knowledge Graph

IEEE Transactions on Artificial Intelligence

引用

IEEE Transactions on Artificial Intelligence 2022年第1期3卷 78-87页

作者： Ma, Lianbo Wang, Jingwei Cheng, Jian Wang, Xingwei Zhu, Wancheng Northeastern University Shenyang China College of Software Northeastern University Shenyang China Research Institute of Mine Big Data China Coal Research Institute Beijing China College of Computer Science and Engineering Northeastern University Shenyang China School of Resource and Civil Engineering Northeastern University Shenyang China

The mine landslide risk prediction is a fundamental task for the safety management of the digital mining system, which is dependent on the analysis of the open pit mine exploitation slope stability. Such stability analysis involves a lot of local natural and human factors. Training a learning model using these factors, with the aim of discovering their relationship with the slope stability, is an intuitive way for the mine landslide risk prediction. The key issue is how to explore the complex nonlinear relations among these factors by using a small amount of high-dimensional historical slope data. Traditional factor-led methods for the issue only focus on the impacts of landslide factors, but ignore the correlations between historical slope data, which in fact contain more useful information. In this article, we propose a new mine landslide risk prediction model using knowledge graph. The gradient boosting decision tree is applied to further exploit the crossed features within the historical data, and then a landslide semantic network is constructed using knowledge graph with the consideration of correlation information between historical slope data. In this way, the model is suitable to deal with the small set of high-dimensional data, and it makes a joint use of the features within the landslide factor values and correlations between historical slope data. We conduct a set of experiments on real historical slope data from the real-world open pit mining scene. The experiment results validate the effectiveness and efficiency of the proposed model in predicting landslide risks with a small set of high-dimensional historical data. © 2020 IEEE.

关键词： Landslides

来源：评论

学校读者我要写书评

暂无评论

LCM: Locally Constrained Compact Point Cloud Model for Masked Point Modeling 38

LCM: Locally Constrained Compact Point Cloud Model for Maske...

引用

38th Conference on Neural Information Processing Systems, NeurIPS 2024

作者： Zha, Yaohua Li, Naiqi Wang, Yanzi Dai, Tao Guo, Hang Chen, Bin Wang, Zhi Ouyang, Zhihao Xia, Shu-Tao Tsinghua Shenzhen International Graduate School Tsinghua University China Institute of Visual Intelligence Pengcheng Laboratory China College of Computer Science and Software Engineering Shenzhen University China Harbin Institute of Technology Shenzhen China Bytedance Inc. China

The pre-trained point cloud model based on Masked Point Modeling (MPM) has exhibited substantial improvements across various tasks. However, these models heavily rely on the Transformer, leading to quadratic complexity and limited decoder, hindering their practice application. To address this limitation, we first conduct a comprehensive analysis of existing Transformer-based MPM, emphasizing the idea that redundancy reduction is crucial for point cloud analysis. To this end, we propose a Locally constrained Compact point cloud Model (LCM) consisting of a locally constrained compact encoder and a locally constrained Mamba-based decoder. Our encoder replaces self-attention with our local aggregation layers to achieve an elegant balance between performance and efficiency. Considering the varying information density between masked and unmasked patches in the decoder inputs of MPM, we introduce a locally constrained Mamba-based decoder. This decoder ensures linear complexity while maximizing the perception of point cloud geometry information from unmasked patches with higher information density. Extensive experimental results show that our compact model significantly surpasses existing Transformer-based models in both performance and efficiency, especially our LCM-based Point-MAE model, compared to the Transformer-based model, achieved an improvement of 1.84%, 0.67%, and 0.60% in average accuracy on the three variants of ScanObjectNN while reducing parameters by 88% and computation by 73%. Code is available at https://***/zyh16143998882/LCM. © 2024 Neural information processing systems foundation. All rights reserved.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Predicting Students' Performance using Machine Learning Classifiers 25

Predicting Students' Performance using Machine Learning Clas...

引用

25th International Arab Conference on Information Technology, ACIT 2024

作者： Alsalem, Gheed M. Sarhan, Noor Hammad, Mustafa Zawaideh, Bayan University of Jordan Counseling and Special Education Department Amman Jordan University of Bahrain Department of Computer Science Sakheer Bahrain Mutah University Department of Software Engineering Alkarak Jordan Luminus Technical College University Foundation and Support Programs Amman Jordan

ISBN: (纸本)9798331540012

This paper presents students' performance prediction model based on Machine Learning (ML) classifiers. The used dataset provides three groups of features that may affect students' performance. Features selection approaches have been applied to the dataset. Several well-known ML classifiers including Neural Network (ANN), Bayes Network (Bayes Net), and Decision Tree (DT). Along with the original approach, hyper approach is applied using boosting, bagging, and voting classifiers. The performance of these algorithms is examined and compared using a confusion matrix. Results show that behavioral features have a major impact on students' performance and ML algorithms can be used effectively to build students' performance prediction model. © 2024 IEEE.

关键词： Prediction models

来源：评论

学校读者我要写书评

暂无评论

Neural Networks and Deep Learning: Innovations and Applications in Soft Computing 7

Neural Networks and Deep Learning: Innovations and Applicati...

引用

7th International Conference on Contemporary Computing and Informatics, IC3I 2024

作者： Abdulkadar, Jahagirdar Imran Mohan, Chinnem Rama Bharati, P. Vijaya Palit, Shamik Rao, Gunda Srinivasa Mudholkar, Pankaj Savitribai Phule Pune University India Narayana Engineering College Department of Computer Science and Engineering Andhra Pradesh Nellore524004 India Vignans Institute of Engineering for Women Department of Cse Andhra Pradesh Visakhapatnam530049 India University of Stirling Department of Computing Science and Software Engineering RAK Campus United Arab Emirates Mallareddy University School of Management and Commerce Kompally500100 India Marwadi University Faculty of Computer Applications Gujarat Rajkot360003 India

ISBN: (纸本)9798350350067

The Paper explores different aspects of deep learning techniques and neural networks in the fields of healthcare, time-series forecasting, agriculture, and other relevant sectors through soft computing. The objective of the report is to examine the significant potential and applications of neural networks. While machine learning techniques called neural networks (NN) are utilized for audio, picture, and the processing of natural languages, it has huge implications in robotics where several machine learning algorithms are significantly required. Improvements in deep machine learning are also investigated in the article specifically in the fields of drug construction, genetics, recognizing faces, farming, biological medicine, biological informatics, medical treatment, natural language, multimedia analysis of data, and mobility prediction. However black box challenges and privacy concerns with data are often ignored in this research. The paper also outlines deep learning-driven data mining techniques for precisely predicting, displaying, evaluating, and classifying data. The report has also emphasized on addressing the potential privacy concerns and challenges associated to information representation. By reviewing recent literature sources about neural network techniques, the study proposes innovation strategies and potential improvement areas for the development of advanced technologies like deep learning and neural networks while addressing the key privacy concerns in data processing. © 2024 IEEE.

关键词： Adversarial machine learning

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：