检索结果-内蒙古大学图书馆

data optimization in Deep Learning: A Survey

IEEE TRANSACTIONS ON KNOWLEDGE AND data ENGINEERING 2025年第5期37卷 2356-2375页

作者： Wu, Ou Yao, Rujing Tianjin Univ Ctr Appl Math Tianjin 300072 Peoples R China Univ Chinese Acad Sci HIAS Hangzhou 310000 Peoples R China Nankai Univ Business Sch Dept Informat Resources Management Tianjin 300071 Peoples R China

Large-scale, high-quality data are considered an essential factor for the successful application of many deep learning techniques. Meanwhile, numerous real-world deep learning tasks still have to contend with the lack of sufficient amounts of high-quality data. Additionally, issues such as model robustness, fairness, and trustworthiness are also closely related to training data. Consequently, a huge number of studies in the existing literature have focused on the data aspect in deep learning tasks. Some typical data optimization techniques include data augmentation, logit perturbation, sample weighting, and data condensation. These techniques usually come from different deep learning divisions and their theoretical inspirations or heuristic motivations may seem unrelated to each other. This study aims to organize a wide range of existing data optimization methodologies for deep learning from the previous literature, and makes the effort to construct a comprehensive taxonomy for them. The constructed taxonomy considers the diversity of split dimensions, and deep sub-taxonomies are constructed for each dimension. On the basis of the taxonomy, connections among the extensive data optimization methods for deep learning are built in terms of five aspects. We probe into rendering several promising and interesting future directions. The constructed taxonomy and the revealed connections will enlighten the better understanding of existing methods and the design of novel data optimization techniques. Furthermore, our aspiration for this survey is to promote data optimization as an independent subdivision of deep learning.

关键词： Training data Training optimization Deep learning Taxonomy Surveys Noise measurement data models Noise Artificial intelligence data optimization data augmentation sample weighting data perturbation

来源：评论

学校读者我要写书评

暂无评论

Efficient Parallel data optimization for Homogeneous Diffusion Inpainting of 4K Images\ast

引用

SIAM JOURNAL ON IMAGING SCIENCES 2025年第1期18卷 701-734页

作者： Kaemper, Niklas Chizhov, Vassillen Weickert, Joachim Saarland Univ Fac Math & Comp Sci Math Image Anal Grp D-66041 Saarbrucken Germany

Homogeneous diffusion inpainting can reconstruct missing image areas with high quality from a sparse subset of known pixels, provided that their location as well as their gray or color values are well optimized. This property is exploited in inpainting-based image compression, which is a promising alternative to classical transform-based codecs such as JPEG and JPEG2000. However, optimizing the inpainting data is a challenging task. Current approaches are either fairly slow or do not produce high quality results. As a remedy we propose fast spatial and tonal optimization algorithms for homogeneous diffusion inpainting that efficiently utilize GPU parallelism, with a careful adaptation of some of the most successful numerical concepts. We propose a densification strategy using ideas from error-map dithering combined with a Delaunay triangulation for the spatial optimization. For the tonal optimization we design a domain decomposition solver that solves the corresponding normal equations in a matrix-free fashion and supplement it with a Voronoi-based initialization strategy. With our proposed methods we are able to generate high quality inpainting masks for homogeneous diffusion and optimized tonal values in a runtime that outperforms prior state-of-the-art by a wide margin.

关键词： inpainting homogeneous diffusion data optimization domain decomposition GPU

来源：评论

学校读者我要写书评

暂无评论

Research on sensor data optimization technology for thermal hydraulic experiment of nuclear reactor

引用

NUCLEAR ENGINEERING AND DESIGN 2024年 423卷

作者： Liu, Yongchao Li, Tong Xiao, Kai Chen, Jie Tan, Xin Cheng, Jiahao Tan, Sichao Wang, Bo He, Zhengxi Shen, Jihong Gao, Puzhen Tian, Ruifeng Harbin Engn Univ Coll Nucl Sci & Technol Harbin 150001 Peoples R China Harbin Engn Univ Heilongjiang Prov Key Lab Nucl Power Syst & Equipm Harbin 150001 Peoples R China Nucl Power Inst China Sci & Technol Reactor Syst Design Technol Lab Chengdu 610213 Peoples R China Harbin Engn Univ Coll Math Sci Harbin 150001 Peoples R China

The sensor signals collected by the nuclear reactor thermal hydraulic experimental system are mixed with complex noise information. The uncertainty of sensor data directly affects the analysis and evaluation effect of machine learning algorithms on the operating status of the experimental system. Traditional denoising methods have poor adaptability to different types of noisy data, are highly dependent on researchers and have high design cost. This paper proposes a data optimization technology based on a hybrid adaptive real-time denoising (HART) model, which can realize data distribution self-adaptation and algorithm hyperparameter self-optimization according to the characteristics of noisy source data. Through the joint denoising algorithm with Variational Mode Decomposition (VMD) algorithm as the core, the partition denoising of source data is realized. In addition, the practical application of the source data of thermal hydraulic experiments in the nuclear field and the verification of simulation and noise addition experiments have been completed. The results show that the denoising optimization model proposed in this paper has the characteristics of good adaptability, self-optimization and realtime processing for different sensor signals. It can effectively remove noise information, restrain data uncertainty, and provide high-quality data source for experimental data analysis based on machine learning. At the same time, the optimization technology can be further applied to the optimization of sensor data of nuclear power plants (NPP).

关键词： Artificial intelligence Thermal hydraulic experiment Sensor data Self-adaptation data optimization VMD

来源：评论

学校读者我要写书评

暂无评论

NIDS-VSB: Network Intrusion Detection System for VANET Using Spark-Based Big data optimization and Transfer Learning

引用

IEEE TRANSACTIONS ON CONSUMER ELECTRONICS 2024年第1期70卷 1798-1809页

作者： Ullah, Farhan Srivastava, Gautam Ullah, Shamsher Yoshigoe, Kenji Zhao, Yue Northwestern Polytech Univ Sch Software Xian 710072 Shaanxi Peoples R China Brandon Univ Dept Math & Comp Sci Brandon MB R7A 6A9 Canada China Med Univ Res Ctr Interneural Comp Taichung 40402 Taiwan Lebanese Amer Univ Dept Comp Sci & Math Beirut 1102 Lebanon Shenzhen Univ Natl Engn Lab Big Data Syst Comp Technol Shenzhen 518060 Peoples R China Embry Riddle Aeronaut Univ Dept Elect Engn & Comp Sci Daytona Beach FL 32114 USA

Vehicle Ad Hoc Networks (VANET) are a technology that allows existing transportation systems to deliver critical information safely. Due to real-time communication, VANET generate massive amounts of data, making them vulnerable to eavesdropping and interference. Intrusion Detection Systems (IDS) are crucial devices that detect and respond to malicious activity. This paper develops a novel network intrusion detection system for VANET that leverages Spark-based big data optimization and transfer learning (NIDS-VSB). First, a packet parser is used to crawl network traffic and filter required flow events. Next, a Spark-based optimization method is implemented to process large amounts of data efficiently. Furthermore, a transfer learning strategy is designed to learn comprehensive feature representations using their semantic anchors. A Convolutional Neural Network (CNN) model is used to extract deep features from semantic anchors. Finally, a stacking generalization ensemble model uses deep features to detect various attacks. The proposed approach is tested on three standard and publicly available datasets, namely the CIC-IoT 2022, CIC-IDS2017, and NSL_KDD datasets. Our findings show that the proposed approach delivers the best detection results for a variety of attacks including audio, HTTP, BruteForce, Web, and others.

关键词： Intrusion detection big data VANET data optimization transfer learning cybersecurity

来源：评论

学校读者我要写书评

暂无评论

Four vector intelligent metaheuristic for data optimization

引用

COMPUTING 2024年第7期106卷 2321-2359页

作者： Fakhouri, Hussam N. Awaysheh, Feras M. Alawadi, Sadi Alkhalaileh, Mohannad Hamad, Faten Univ Petra Fac Informat Technol Data Sci & Artificial Intelligence Dept Amman Jordan Tartu Univ Inst Comp Sci Delta Res Ctr Tartu Estonia Blekinge Inst Technol Dept Comp Sci DIDA Karlskrona Sweden Al Ain Univ Coll Educ Humanities & Social Sci Al Ain U Arab Emirates Sultan Qaboos Univ Informat Studies Dept Muscat Oman Univ Jordan Sch Educ Sci Amman Jordan

Swarm intelligence (SI) algorithms represent a class of Artificial Intelligence (AI) optimization metaheuristics used for solving complex optimization problems. However, a key challenge in solving complex problems is maintaining the balance between exploration and exploitation to find the optimal global solution and avoid local minima. This paper proposes an innovative Swarm Intelligence (SI) algorithm called the Four Vector Intelligent Metaheuristic (FVIM) to address the aforementioned problem. FVIM's search strategy is guided by four top-performing leaders within a swarm, ensuring a balanced exploration-exploitation trade-off in the search space, avoiding local minima, and mitigating low convergence issues. The efficacy of FVIM is evaluated through extensive experiments conducted over two datasets, incorporating both qualitative and quantitative statistical measurements. One dataset contains twenty-three well-known single-objective optimization functions, such as fixed-dimensional and multi-modal functions, while the other dataset comprises the CEC2017 functions. Additionally, the Wilcoxon test was computed to validate the result's significance. The results illustrate FVIM's effectiveness in addressing diverse optimization challenges. Moreover, FVIM has been successfully applied to tackle engineering design problems, such as weld beam and truss engineering design.

关键词： Artificial intelligence Swarm intelligence data optimization Exploitation Exploration Global optima

来源：评论

学校读者我要写书评

暂无评论

Quantum enhanced optical sensors in data optimization for huge communication network

引用

OPTICAL AND QUANTUM ELECTRONICS 2024年第3期56卷 422-422页

作者： Gao, Lin'e Nan, Yahui Yuncheng Adv Normal Coll Dept Math & Comp Yuncheng 044029 Peoples R China Lvliang Univ Dept Comp Sci & Technol Lvliang 033000 Shanxi Peoples R China

The Internet of Things (IoT)-based limited WSN (wireless sensor network) has garnered much interest and advancement over the past several years to enhance resource utilization and service delivery. IoT demands a more robust network for communication and an appropriate location for an energy-efficient WSN for data transport across heterogeneous devices. In addition to the large area that needs to be covered and the restricted communication range of the sensors, WSNs with a single sink may not be appropriate in applications like smart cities. Therefore, Multi-sink WSN solutions appear appropriate for these kinds of applications. Although they boost network speed, the lifespan of the network, and energy usage, multi-sink WSNs are becoming more common. This study leverages deep learning architectures to create a unique resource allocation approach for wireless sensor IoT networks that is both energy-efficient and data-optimized. In this work, the integration of the network's energy efficiency, data optimization and optical wireless communication is processed. In this paper, we present Modified Bat for Node optimization, a novel approach that uses reinforcement-based Q-learning approaches to improve energy efficiency and resource allocation in Multi-sink Wireless Sensor Networks (WSNs). Additionally, routing with multiple hops is necessary for the WSNS to gather information from sensor nodes and transmit it to the sinking node for decision-making. A reliable and energy-efficient way of communicating data between sensor nodes is provided by Optical Wireless Communication, which leverages RF technology. The best power distribution and relay selection are accomplished using the suggested strategy. Lowering total power transmission and achieving Quality of Service requirements improve resource allocation and relay selection. The simulation outcomes show that the proposed model outperforms traditional ones regarding throughput, energy efficiency, quality of service, frequency

关键词： Wireless sensor network Internet of Things Resource allocation Energy efficiency data optimization Deep learning Optimal wireless communication

来源：评论

学校读者我要写书评

暂无评论

Quality of Service-Oriented data optimization in Networks using Artificial Intelligence Techniques

引用

INTERNATIONAL JOURNAL OF ADVANCED COMPUTER SCIENCE AND APPLICATIONS 2024年第6期15卷 897-908页

作者： Yang, Zhenhua Yang, Qiwen Yang, Minghong Hunan Appl Technol Univ Sch Informat Engn Changde Peoples R China Beihang Univ Sch Comp Sci Beijing Peoples R China Hunan Appl Technol Univ Sch Econ & Management Changde Peoples R China

This paper outlines a comprehensive AI-driven Quality of Service (QoS) optimization method, presenting a rigorous examination of its effectiveness through extensive experimentation and analysis. By applying real-world datasets to simulate network environments, the study systematically evaluates the proposed method's impact across various QoS metrics. Key findings reveal substantial enhancements in reducing average latency, minimizing packet loss, and boosting bandwidth utilization compared to baseline scenarios, with the Deep Deterministic Policy Gradient (DDPG) model showcasing the most notable improvements. The research demonstrates that AI optimization strategies, particularly those leveraging DQN and DDPG algorithms, significantly improve upon conventional methods. Specifically, post-migration optimizations lead to a recovery and even surpassing of pre-migration QoS levels, with delays dropping to levels below initial readings, packet loss nearly eliminated, and bandwidth utilization markedly improved. The study further illustrates that while lower learning rates necessitate longer convergence times, they ultimately facilitate superior model performance and stability. In-depth case studies within a cloud data center setting underscore the system's proficiency in handling large-scale Virtual Machine (VM) migrations with minimal disruption to network performance. The AI-driven optimization successfully mitigates the typical latency spikes, packet loss increases, and resource utilization dips associated with VM migrations, thereby affirming its practical value in maintaining high network efficiency and stability during such operations. Comparative analyses against traditional traffic engineering methods, rule-based controls, and other machine learning approaches consistently place the AI optimization method ahead, achieving up to an 8% increase in throughput alongside a 2 ms decrease in latency. Furthermore, the technique excels in reducing packet loss by 25% and e

关键词： Artificial intelligence networking quality of service-oriented data optimization

来源：评论

学校读者我要写书评

暂无评论

Information-Centric IoT-Based Smart Farming with Dynamic data optimization

引用

Computers, Materials & Continua 2023年第2期74卷 3865-3880页

作者： Souvik Pal Hannah VijayKumar D.Akila N.Z.Jhanjhi Omar A.Darwish Fathi Amsaad Department of Computer Science and Engineering Sister Nivedita UniversityKolkataIndia Post Doctoral Researcher Sambalpur UniversitySambalpurIndia Department of Computer Science Anna Adarsh College for WomenChennaiIndia Department of Information Technology Vels Institute of ScienceTechnology and Advanced StudiesChennaiIndia School of Computer Science and Engineering SCETaylor’s UniversitySubang Jaya47500SelangorMalaysia FITTaylor’s UniversitySubang Jaya47500SelangorMalaysia Department of Information Security and Applied Computing Eastern Michigan UniversityUSA School of Information Security and Applied Computing Eastern Michigan UniversityUSA

Smart farming has become a strategic approach of sustainable agriculture management and monitoring with the infrastructure to exploit modern technologies,including big data,the cloud,and the Internet of Things(IoT).Many researchers try to integrate IoT-based smart farming on cloud platforms *** define various frameworks on smart farming and monitoring system and still lacks to define effective data management *** IoT-cloud systems involve massive structured and unstructured data,data optimization comes into the ***,this research designs an Information-Centric IoT-based Smart Farming with Dynamic data optimization(ICISF-DDO),which enhances the performance of the smart farming infrastructure with minimal energy consumption and improved ***,a conceptual framework of the proposed scheme and statistical design model has beenwell *** information storage and management with DDO has been expanded individually to show the effective use of membership parameters in data *** simulation outcomes state that the proposed ICISF-DDO can surpass existing smart farming systems with a data optimization ratio of 97.71%,reliability ratio of 98.63%,a coverage ratio of 99.67%,least sensor error rate of 8.96%,and efficient energy consumption ratio of 4.84%.

关键词： Smart farming data optimization information management internet of things cloud

来源：评论

学校读者我要写书评

暂无评论

High Definition Map data optimization for Autonomous Driving in Vehicular Named data Networks

High Definition Map Data Optimization for Autonomous Driving...

引用

IEEE International Conference on Communications (IEEE ICC)

作者： Doe, Daniel Mawunyo Chen, Dawei Han, Kyungtae Wang, Haoxin Xie, Jiang Han, Zhu Univ Houston Elect & Comp Engn Dept Houston TX 77004 USA Toyota Motor North Amer R&D InfoTech Labs Mountain View CA USA Georgia State Univ Dept Comp Sci Atlanta GA 30303 USA Univ N Carolina Dept Elect & Comp Engn Charlotte NC USA

ISBN: (纸本)9781538674628

High-definition (HD) map is an essential building block in the autonomous driving era, which enables fine-grained environmental awareness, exact localization, and route planning. However, because HD maps include rich, multidimensional information, the volume of HD map data is enormous, making it expensive and time-consuming to transmit on vehicular networks. Therefore, in this paper, we propose a data optimization scheme for effective HD map updates in vehicular named data networking (NDN) scenarios. We formulate the HD map data optimization problem as a convex optimization problem and solve it with modified convolutional neural networks (CNNs) from YOLOX's real-time object detection system. Specifically, we modify the YOLOX object detection algorithm to detect and compress redundant pixels in local map data before transmission to the MEC server. To deploy our proposed scheme, we construct a vehicular NDN environment for data collection, processing, and transmission using the CARLA simulator and robot operating system 2 (ROS2). Extensive simulations show that our proposed scheme can significantly reduce the transmission data size and time by 48.25%- 65.78% and 46.85%- 78.84% compared with state-of-the-art HD map update techniques like RLSS, Pro-RTT, and Loss-based systems.

关键词： High definition map data optimization object detection named data networking and vehicular networks

来源：评论

学校读者我要写书评

暂无评论

Dynamic data optimization in IoT-Assisted Sensor Networks on Cloud Platform

引用

Computers, Materials & Continua 2022年第7期72卷 1357-1372页

作者： Nguyen A.Tuan D.Akila Souvik Pal Bikramjit Sarkar Thien Khai Tran G.Mothilal Nehru Dac-Nhuong Le Faculty of Information Technology Ho Chi Minh City University of Foreign Languages-Information TechnologyHo Chi Minh City70000Vietnam Department of Information Technology Vels Institute of ScienceTechnology&Advanced StudiesChennai600117India Department of Computer Science and Engineering Sister Nivedita UniversityKolkata700156India Department of Computer Science and Engineering JIS College of EngineeringNadia741235India Institute of Research and Development Duy Tan UniversityDanang550000Vietnam School of Computer Science Duy Tan UniversityDanang550000Vietnam

This article presents a new scheme for dynamic data optimization in IoT(Internet of Things)-assisted sensor *** various components of IoT assisted cloud platform are *** addition,a new architecture for IoT assisted sensor networks is ***,a model for data optimization in IoT assisted sensor networks is proposed.A novel Membership inducing Dynamic data optimization Membership inducing Dynamic data optimization(MIDDO)algorithm for IoT assisted sensor network is proposed in this *** proposed algorithm considers every node data and utilized membership function for the optimized data *** proposed framework is compared with two stage optimization,dynamic stochastic optimization and sparsity inducing optimization and evaluated in terms of reliability ratio,coverage ratio and sensing *** optimization was performed based on the availability of cloud resource,sensor energy,data flow volume and the centroid of each *** was inferred that the proposed MIDDO algorithm achieves an average performance ratio of 76.55%,reliability ratio of 94.74%,coverage ratio of 85.75%and sensing error of 0.154.

关键词： data optimization IoT sensor network reliability membership sensor energy

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：