检索结果-内蒙古大学图书馆

MAML^(2):meta reinforcement learning via meta-learning for task categories

Frontiers of Computer science 2023年第4期17卷 57-67页

作者： Qiming FU Zhechao WANG Nengwei FANG Bin XING Xiao ZHANG Jianping CHEN School of Electronics and Information Engineering Suzhou University of Science and TechnologySuzhou 215009China Chongqing Industrial Big Data Innovation Center Co.Ltd. Chongqing 404100China School of Medical Informatics Xuzhou Medical UniversityXuzhou 221004China School of Architecture and Urban Planning Suzhou University of Science and TechnologySuzhou 215009China

Meta-learning has been widely applied to solving few-shot reinforcement learning problems,where we hope to obtain an agent that can learn quickly in a new ***,these algorithms often ignore some isolated tasks in pursuit of the average performance,which may result in negative adaptation in these isolated tasks,and they usually need sufficient learning in a stationary task *** this paper,our algorithm presents a hierarchical framework of double meta-learning,and the whole framework includes classification,meta-learning,and ***,in the classification process,we classify tasks into several task subsets,considered as some categories of tasks,by learned parameters of each task,which can separate out some isolated tasks ***,in the meta-learning process,we learn category parameters in all subsets via ***,based on the gradient of each category parameter in each subset,we use meta-learning again to learn a new metaparameter related to the whole task set,which can be used as an initial parameter for the new ***,in the re-adaption process,we adapt the parameter of the new task with two steps,by the meta-parameter and the appropriate category parameter ***,we demonstrate our algorithm prevents the agent from negative adaptation without losing the average performance for the whole task ***,our algorithm presents a more rapid adaptation process within ***,we show the good performance of our algorithm with fewer samples as the agent is exposed to an online meta-learning setting.

关键词： meta-learning reinforcement learning few-shot learning negative adaptation

来源：评论

学校读者我要写书评

暂无评论

Robust twin-field quantum key distribution through sending or not sending

引用

National science Review 2023年第4期10卷 76-85页

作者： Cong Jiang Zong-Wen Yu Xiao-Long Hu Xiang-Bin Wang Jinan Institute of Quantum Technology State Key Laboratory of Low Dimensional Quantum Physics Department of Physics Tsinghua University Data Communication Science and Technology Research Institute School of Physics State Key Laboratory of Optoelectronic Materials and Technologies Sun Yat-sen University Shanghai Branch CAS Center for Excellence and Synergetic Innovation Center in Quantum Information and Quantum Physics University of Science and Technology of China Shenzhen Institute for Quantum Science and Engineering and Physics Department Southern University of Science and Technology Frontier Science Center for Quantum Information

The sending-or-not-sending(SNS) protocol is one of the most major variants of the twin-field(TF)quantum key distribution(QKD) protocol and has been realized in a 511-km field fiber, the farthest field experiment to date. In practice, however, all decoy-state methods have unavoidable source errors, and the source errors may be non-random, which compromises the security condition of the existing TF-QKD protocols. In this study, we present a general approach for efficiently calculating the SNS protocol's secure key rate with source errors, by establishing the equivalent protocols through virtual atenuation and the tagged model. This makes the first result for TF QKD in practice where source intensity cannot be controlled exactly. Our method can be combined with the two-way classical communication method such as active odd-parity pairing to further improve the key rate. The numerical results show that if the intensity error is within a few percent, the key rate and secure distance only decrease marginally. The key rate of the recent SNS experiment in the 511-km field fiber is still positive using our method presented here, even if there is a ±9.5% intensity fluctuation. This shows that the SNS protocol is robust against source errors.

关键词： quantum information quantum key distribution sending or not sending twin field source error

来源：评论

学校读者我要写书评

暂无评论

Set pair three-way overlapping community discovery algorithm for weighted social internet of things

引用

Digital Communications and Networks 2023年第1期9卷 3-13页

作者： Chunying Zhang Jing Ren Lu Liu Shouyue Liu Xiaoqi Li Liya Wang College of Science North China University of Science and TechnologyTangshanHebeiChina Hebei Engineering Research Center for the Intelligentization of Iron Ore Optimization and Ironmaking Raw Materials Preparation Processes North China University of Science and TechnologyTangshanHebeiChina Hebei Key Laboratory of Data Science and Application North China University of Science and TechnologyTangshanHebeiChina The Key Laboratory of Engineering Computing in Tangshan City North China University of Science and TechnologyTangshanHebeiChina Tangshan Intelligent Industry and Image Processing Technology Innovation Center North China University of Science and TechnologyTangshanHebeiChina

There are many problems in Social Internet of Things(IoTs),such as complex topology information,different degree of association between nodes and overlapping *** idea of set pair information grain computing and clustering is introduced to solve the above problems so as to accurately describe the similarity between nodes and fully explore the multi-community structure.A Set Pair Three-Way Overlapping Community Discovery Algorithm for Weighted Social Internet of Things(WSIoT-SPTOCD)is *** the local network structure,which fully considers the topological information between nodes,the set pair connection degree is used to analyze the identity,difference and reverse of neighbor *** similarity degree of different neighbor nodes is defined from network edge weight and node degree,and the similarity measurement method of set pair between nodes based on the local information structure is *** to the number of nodes'neighbors and the connection degree of adjacent edges,the clustering intensity of nodes is defined,and an improved algorithm for initial value selection of k-means is *** nodes are allocated according to the set pair similarity between nodes and different ***-way community structures composed of a positive domain,boundary domain and negative domain are generated ***,the overlapping node set is generated according to the calculation results of community node ***,experiments are carried out on artificial networks and real *** results show that WSIoT-SPTOCD performs well in terms of standardized mutual information,overlapping community modularity and F1.

关键词： Social internet of things Set pair analysis K-means Local information structure Overlapping community

来源：评论

学校读者我要写书评

暂无评论

Transformers with Joint Tokens and Local-Global Attention for Efficient Human Pose Estimation

arXiv

引用

arXiv 2025年

作者： Kinfu, Kaleab A. Vidal, René Center for Innovation in Data Engineering and Science University of Pennsylvania PhiladelphiaPA United States

Convolutional Neural Networks (CNNs) and Vision Transformers (ViTs) have led to significant progress in 2D body pose estimation. However, achieving a good balance between accuracy, efficiency, and robustness remains a challenge. For instance, CNNs are computationally efficient but struggle with long-range dependencies, while ViTs excel in capturing such dependencies but suffer from quadratic computational complexity. This paper proposes two ViT-based models for accurate, efficient, and robust 2D pose estimation. The first one, EViTPose, operates in a computationally efficient manner without sacrificing accuracy by utilizing learnable joint tokens to select and process a subset of the most important body patches, enabling us to control the trade-off between accuracy and efficiency by changing the number of patches to be processed. The second one, UniTransPose, while not allowing for the same level of direct control over the trade-off, efficiently handles multiple scales by combining (1) an efficient multi-scale transformer encoder that uses both local and global attention with (2) an efficient sub-pixel CNN decoder for better speed and accuracy. Moreover, by incorporating all joints from different benchmarks into a unified skeletal representation, we train robust methods that learn from multiple datasets simultaneously and perform well across a range of scenarios – including pose variations, lighting conditions, and occlusions. Experiments on six benchmarks demonstrate that the proposed methods significantly outperform state-of-the-art methods while improving computational efficiency. EViTPose exhibits a significant decrease in computational complexity (30% to 44% less in GFLOPs) with a minimal drop of accuracy (0% to 3.5% less), and UniTransPose achieves accuracy improvements ranging from 0.9% to 43.8% across these benchmarks. Copyright © 2025, The Authors. All rights reserved.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Analysis on n-level quantum systems by means of a coordinate transformation 63

Analysis on n-level quantum systems by means of a coordinate...

引用

63rd IEEE Conference on Decision and Control, CDC 2024

作者： Xu, Huilong Miao, Zibo Gao, Qing Harbin Institute of Technology School of Mechanical Engineering and Automation Shenzhen518055 China Beijing Advanced Innovation Center for Big Data and Brain Computing Beihang University School of Automation Science and Electrical Engineering Beijing100191 China

ISBN: (纸本)9798350316339

The structural decomposition of two-level quantum systems has recently been established, related to various applications in quantum information science. As an extension to previous work on two-level quantum systems, this paper aims to investigate the structural properties of n-level quantum systems. We develop algebraic conditions under which the attainment of quantum back-action evading (BAE) measurements, the generation of decoherence-free (DF) subspaces and quantum non-demolition (QND) variables, as well as the existence of steadystate solutions for the Lindblad equations, can be realized. In particular, a specific coordinate transformation is explicitly constructed to facilitate the achievement of above-mentioned objectives, accompanied by an illustrative numerical example. © 2024 IEEE.

关键词： Quantum optics

来源：评论

学校读者我要写书评

暂无评论

Probing the electric double layer structure at nitrogen-doped graphite electrodes by constant-potential molecular dynamics simulations

引用

Journal of Energy Chemistry 2024年第6期93卷 299-305,I0008页

作者： Legeng Yu Nan Yao Yu-Chen Gao Zhong-Heng Fu Bo Jiang Ruiping Li Cheng Tang Xiang Chen Tsinghua Center for Green Chemical Engineering Electrification(CCEE) Department of Chemical EngineeringTsinghua UniversityBeijing 100084China Beijing Advanced Innovation Center for Materials Genome Engineering Institute for Advanced Materials and TechnologyUniversity of Science and Technology BeijingBeijing 100083China National Materials Corrosion and Protection Data Center University of Science and Technology BeijingBeijing 100083China Ordos Carbon Neutral Research and Application Co. Ltd.Ordos 017010Inner Mongoli1China Institute for Carbon Neutrality Tsinghua UniversityBeijing 100084China Center for Next-Generation Energy Materials and School of Chemical Engineering Sungkyunkwan University2066 Seobu-roJangan-guSuwon-siGyeonggi-do 16419Republic of Korea

Electric double layer(EDL)is a critical topic in electrochemistry and largely determines the working performance of lithium ***,atomic insights into the EDL structures on heteroatom-modified graphite anodes and EDL evolution with electrode potential are very ***,a constant-potential molecular dynamics(CPMD)method is proposed to probe the EDL structure under working conditions,taking N-doped graphite electrodes and carbonate electrolytes as an *** interface model was developed,incorporating the electrode potential and atom *** a result,an insightful atomic scenario for the EDL structure under varied electrode potentials has been established,which unveils the important role of doping sites in regulating both the EDL structures and the following electrochemical reactions at the atomic ***,the negatively charged N atoms repel the anions and adsorb Li~+at high and low potentials,*** preferential adsorption suggests that Ndoped graphite can promote Li~+desolvation and regulate the location of Li~+*** CPMD method not only unveils the mysterious function of N-doping from the viewpoint of EDL at the atomic level but also applies to probe the interfacial structure on other complicated electrodes.

关键词： Lithium batteries Graphite N-doping Electric double layer Molecular dynamics Constant potential method Electrode potential

来源：评论

学校读者我要写书评

暂无评论

Residential Energy Consumption Forecasting Based on Federated Reinforcement Learning with data Privacy Protection

引用

Computer Modeling in engineering & sciences 2023年第10期137卷 717-732页

作者： You Lu Linqian Cui YunzheWang Jiacheng Sun Lanhui Liu School of Electronic and Information Engineering Suzhou University of Science and TechnologySuzhou215009China Jiangsu Province Key Laboratory of Intelligent Building Energy Efficiency Suzhou University of Science and TechnologySuzhou215009China Chongqing Industrial Big Data Innovation Center Co. Ltd.Chongqing400707China

Most studies have conducted experiments on predicting energy consumption by integrating data formodel ***, the process of centralizing data can cause problems of data ***,many laws and regulationson data security and privacy have been enacted, making it difficult to centralize data, which can lead to a datasilo problem. Thus, to train the model while maintaining user privacy, we adopt a federated learning ***, in all classical federated learning frameworks secure aggregation, the Federated Averaging (FedAvg)method is used to directly weight the model parameters on average, which may have an adverse effect on te ***, we propose the Federated Reinforcement Learning (FedRL) model, which consists of multiple userscollaboratively training the model. Each household trains a local model on local data. These local data neverleave the local area, and only the encrypted parameters are uploaded to the central server to participate in thesecure aggregation of the global model. We improve FedAvg by incorporating a Q-learning algorithm to assignweights to each locally uploaded local model. And the model has improved predictive performance. We validatethe performance of the FedRL model by testing it on a real-world dataset and compare the experimental results withother models. The performance of our proposed method in most of the evaluation metrics is improved comparedto both the centralized and distributed models.

关键词： Energy consumption forecasting federated learning data privacy protection Q-learning

来源：评论

学校读者我要写书评

暂无评论

Preliminary Studies for Cracking Sound Identification During Coffee Roasting

Preliminary Studies for Cracking Sound Identification During...

引用

2023 International Conference on Advanced Mechatronics, Intelligent Manufacture and Industrial Automation, ICAMIMIA 2023

作者： Wibowo, Jony Winaryo Ilham Rizqyawan, Muhammad Fakhrurroja, Hanif Nadiya, Ulfah Munandar, Aris Mahendra, Oka Simbolon, Artha Ivonita Asfy Fakhry Anto, Irfan Rahmawati, Annida National Research and Innovation Agency Research Center for Smart Mechatronic Bandung Indonesia Telkom University School of Industrial Engineering Bandung Indonesia National Research and Innovation Agency Research Center for Smart Mechatronics Bandung Indonesia National Research and Innovation Agency Research Center for Data and Information Science Bandung Indonesia

ISBN: (纸本)9798350309225

A coffee roasting plays an important role in determining the taste of coffee. The process consists of drying, browning, first crack, roast development, and second crack. The first crack is a beginning to form the characteristics of coffee beans. At this stage, a professional coffee roaster must ensure the spesific temperature and time, hence the beans are not getting burn. In manual coffee roasters, this causes inconsistency, because it really depends on the ability and experience of the coffee bean roaster. Therefore, in this study, we conducted a machine learning-based preliminary study to detect the cracking sound of coffee beans. The preliminary study was conducted by collecting videos of the roasting process from the internet. The videos are converted into audio format and separated between audio containing the sound of coffee beans cracking and audio that does not contain the sound of coffee beans cracking. The audios are then converted into an image format using the Mel-frequency cepstral coefficients (MFCC) method and modeled using convolutional neural network (CNN). From the 216 dataset files, we obtained an accuracy value of 89.7% with data lost during the testing process as much as 34.9% with 50 epochs of training and validation processes. © 2023 IEEE.

关键词： Machine learning

来源：评论

学校读者我要写书评

暂无评论

Optimization of Fracture Reduction Robot Controller 8

Optimization of Fracture Reduction Robot Controller

引用

8th International Symposium on Computer science and Intelligent Control, ISCSIC 2024

作者： An, Baichuan Chen, Jianwen Liu, Minghe Yin, Minghuan Song, Zicheng Zhuang, Chao Chang, Cheng Sun, Hao Hebei University of Technology School of Artificial Intelligence and Data Science Tianjin China Southern University of Science and Technology Hospital Limb Deformity Correction and Function Reconstruction Department Shenzhen China Hebei University of Technology School of Artificial Intelligence and Data Science Engineering Research Center of Intelligent Rehabilitation Device and Detection Technology Ministry of Education Control Engineering Technology Innovation Center of Hebei Province Tianjin China

ISBN: (纸本)9798350380286

To improve the accuracy of fracture reduction, combining inverse kinematic analysis with robot dynamics modelling might lead to a more effective controller for the fracture reduction process. The trajectory of the actuator driving rod's motion and any deviations from it are calculated using Simulink simulation. The Levy-SSA-optimized fracture reduction robot's positioning inaccuracy is two orders of magnitude lower than the PID adjustment approach. Using this method also improves the reduction trajectory's accuracy and speeds up the convergence of the fracture-resetting robot control system to the ideal solution. © 2024 IEEE.

关键词： Inverse kinematics

来源：评论

学校读者我要写书评

暂无评论

Security Analysis of Audit Tools 11

Security Analysis of Audit Tools

引用

11th International Conference on Computer, Control, Informatics and its Applications, IC3INA 2024

作者： Widyasuri, Armita Priambodo, Dimas Febriyan Ajhari, Abdul Azzam Sunaringtyas, Septia Ulfa Research Center for Data and Information Science National Research and Innovation Agency Jakarta Indonesia Cybersecurity Engineering National Cyber and Crypto Polytechnic Bogor Indonesia Departement of Informatics Universitas Siber Asia Jakarta Indonesia

ISBN: (纸本)9798331542313

BRIN has built an Electronic-Based Government System (EBS) application and infrastructure audit tool called Application Audit Tools by the mandate of Presidential Regulation No. 95/2018. However, it is necessary to ensure that the web-based audit tool that has been built is guaranteed security. This research was conducted to measure the vulnerability of the audit tool with a blackbox testing method that utilizes the Burpsuite and The Open Worldwide Application Security Project Zed Attack Proxy (OWASP ZAP) applications. The results obtained state that the audit tool has vulnerabilities. Although the results are still at a medium level such as Content Security Policy, Cross Domain Misconfiguration, and Vulnerability JS library by OWASP ZAP and TLS cookies without secure flag set, all of them need attention to improve system security. Recommendations are also given focusing on improving Content Security Policy, Cross Domain Misconfiguration, patches on javascript components such as jQuery, and handling Cookies and Cache. © 2024 IEEE.

关键词： Black-box testing

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：