检索结果-内蒙古大学图书馆

32nd ACM International Conference on Multimedia, MM 2024

作者： Liu, Xinyue Wang, Jianyuan Leng, Biao Zhang, Shuo School of Computer Science and Engineering Beihang University Beijing China Key Laboratory of Intelligent Bionic Unmanned Systems Ministry of Education School of Intelligence Science and Technology University of Science and Technology Beijing Beijing China Beijing Key Lab of Traffic Data Analysis and Mining School of Computer & Technology Beijing Jiaotong University Beijing China

ISBN: (纸本)9798400706868

knowledge distillation based on student-teacher network is one of the mainstream solution paradigms for the challenging unsupervised Anomaly Detection task, utilizing the difference in representation capabilities of the teacher and student networks to implement anomaly localization. However, over-generalization of the student network to the teacher network may lead to negligible differences in representation capabilities of anomaly, thus affecting the detection effectiveness. Existing methods address the possible over-generalization by using differentiated students and teachers from the structural perspective or explicitly expanding distilled information from the content perspective, which inevitably results in an increased likelihood of underfitting of the student network and poor anomaly detection capabilities in anomaly center or edge. In this paper, we propose Dual-Modeling Decouple Distillation (DMDD) for the unsupervised Anomaly Detection. In DMDD, a Decouple Student-Teacher Network is proposed to decouple the initial student features into normality and abnormality features. We further introduce Dual-Modeling Distillation based on normal-anomalous image pairs, fitting normality features of anomalous image and the teacher features of the corresponding normal image, widening the distance between abnormality features and the teacher features in anomalous regions. Synthesizing these two distillation ideas, we achieve anomaly detection which focuses on both edge and center of anomaly. Finally, a Multi-perception Segmentation Network is proposed to achieve focused anomaly map fusion based on multiple attention. Experimental results on MVTec AD show that DMDD surpasses SOTA localization performance of previous knowledge distillation-based methods, reaching 98.85% on pixel-level AUC and 96.13% on PRO. © 2024 ACM.

关键词： Students

来源：评论

学校读者我要写书评

暂无评论

A Multi-Objective Evolutionary Machine Learning Method for the Yield Strength Prediction of Strip Steel in Continuous Annealing

A Multi-Objective Evolutionary Machine Learning Method for t...

引用

IISE Annual Conference and Expo 2024

作者： Liu, Chang Tang, Lixin Zhang, Kainan Guo, Jiajing National Frontiers Science Center for Industrial Intelligence and Systems Optimization North-eastern University Shenyang110819 China Ministry of Education Shenyang110819 China Liaoning Engineering Laboratory of Data Analytics and Optimization for Smart Industry Shenyang110819 China Liaoning Key Laboratory of Manufacturing System and Logistics Optimization Shenyang110819 China

ISBN: (纸本)9781713877851

In the continuous annealing process, the yield strength is the key property index to evaluate the quality of strip steel and guide the production. It is difficult to perform online quality measurement for numerous products in the steel industry, this brings significant challenges for process monitoring and operation optimization. In this paper, a multiobjective evolutionary machine learning method is proposed for the yield strength prediction of strip steel in continuous annealing. First, the factors affecting the yield strength of strip steel are analyzed and the prediction model based on a hybrid kernel support vector regression (HKSVR) is established. To improve the prediction accuracy, a multi-objective evolutionary optimization algorithm with a knee solution strategy is used to optimize the key parameters of the HKSVR. It is verified by the actual production data that the proposed method has high prediction accuracy and can meet the actual production demand. Moreover, the proposed method can help practitioners to construct quality analytics models for many other similar production lines, providing assurance for quality control and optimization with a good generalization. © IISE Annual Conference and Expo *** rights reserved.

关键词： Prediction models

来源：评论

学校读者我要写书评

暂无评论

Generalized-Extended-State-Observer and Equivalent-Input-Disturbance Methods for Active Disturbance Rejection: Deep Observation and Comparison

引用

IEEE/CAA Journal of Automatica Sinica 2023年第4期10卷 957-968页

作者： Jinhua She Kou Miyamoto Qing-Long Han Min Wu Hiroshi Hashimoto Qing-Guo Wang School of Engineering Tokyo University of TechnologyHachiojiTokyo 192-0982Japan K.Miyamoto is with the Institute of Technology Shimizu CorporationKotoTokyo 135-0044Japan School of Science Computing and Engineering TechnologiesSwinburne University of TechnologyMelbourneVIC 3122Australia School of Automation China University of GeosciencesWuhan 430074 Hubei Key Laboratory of Advanced Control and Intelligent Automation for Complex Systems Engineering Research Center of Intelligent Technology for Geo-Exploration Ministry of EducationWuhan 430074China School of Industrial Technology Advanced Institute of Industrial TechnologyTokyo 140-0011Japan Institute of Artificial Intelligence and Future Networks Beijing Normal UniversityZhuhai 519087 Guangdong Key Lab of AI and Multi-Modal Data Processing Guangdong Provincial Key Laboratory of Interdisciplinary Research and Application for Data Science BNUHKBU United International College Zhuhai 519087China

Active disturbance-rejection methods are effective in estimating and rejecting disturbances in both transient and steady-state *** paper presents a deep observation on and a comparison between two of those methods:the generalized extended-state observer(GESO)and the equivalent input disturbance(EID)from assumptions,system configurations,stability conditions,system design,disturbance-rejection performance,and extensibility.A time-domain index is introduced to assess the disturbance-rejection performance.A detailed observation of disturbance-suppression mechanisms reveals the superiority of the EID approach over the GESO method.A comparison between these two methods shows that assumptions on disturbances are more practical and the adjustment of disturbance-rejection performance is easier for the EID approach than for the GESO method.

关键词： Active disturbance-rejection control(ADRC) disturbance observer(DOB) equivalent input disturbance(EID) extendedstate observer(ESO) generalized extended-state observer(GESO)

来源：评论

学校读者我要写书评

暂无评论

Observation-based sources evolution of non-methane hydrocarbons (NMHCs) in a megacity of China

引用

Journal of Environmental Sciences 2023年第2期124卷 794-805页

作者： Yarong Peng Hongli Wang Qian Wang Shengao Jing Jingyu An Yaqin Gao Cheng Huang Rusha Yan Haixia Dai Tiantao Cheng Qiang Zhang Meng Li Jianlin Hu Zhihao Shi Li Li Shengrong Lou Shikang Tao Qinyao Hu Jun Lu Changhong Chen Department of Environmental Science and Engineering Shanghai Key Laboratory of Atmospheric Particle Pollution and PreventionInstitute of Atmospheric SciencesFudan UniversityShanghai 200438China State Environmental Protection Key Laboratory of Formation and Prevention of Urban Air Pollution Complex Shanghai Academy of Environmental SciencesShanghai 200233China Department of Atmospheric and Oceanic Sciences Institute of Atmospheric SciencesFudan UniversityShanghai 200438China Department of Earth System Science Ministry of Education Key Laboratory for Earth System ModelingTsinghua UniversityBeijing 100084China Big Data Institute for Carbon Emission and Environmental Pollution Fudan UniversityShanghai 200438China School of Environmental Science and Engineering Nanjing University of Information Science&TechnologyNanjing 210044China

Both concentrations and emissions of many air pollutants have been decreasing due to implement of control measures in China,in contrast to the fact that an increase in emissions of non-methane hydrocarbons(NMHCs)has been *** study employed seven years continuous NMHCsmeasurements and the related activities data of Shanghai,a megacity in China,to explore evolution of emissions and effectiveness of air pollution control *** mixing ratio of NMHCs showed no statistical interannual changes,of which their compositions exhibited marked *** resulted in a decreasing trend of ozone formation potential by 3.8%/year(p<0.05,the same below),which should be beneficial to ozone pollution mitigation as its production in Shanghai is in the NMHCs-limited *** alkanes,aromatics and acetylene changed by+3.7%/year,-5.9%/year and-7.4%/year,respectively,and alkenes showed no apparent *** sources were apportioned by a positive matrix factorization ***,vehicular emissions(-5.9%/year)and petrochemical industry emissions(-7.1%/year)decreased significantly,but the decrease slowed down;significant reduction in solvent usage(-9.0%/year)appeared after 2010;however,emissions of natural gas(+12.6%/year)and fuel evaporation(with an increasing fraction)became more *** inconsistency between observations and inventories was found in interannual trend and speciation as well as source contributions,emphasizing the need for further validation in NMHCs emission *** study confirms the effectiveness of measures targeting mobile and centralized emissions from industrial sources and reveals a need focusing on fugitive emissions,which provided new insights into future air policies in polluted region.

关键词： Characteristics Source apportionment Observation-based Interannual trend Shanghai

来源：评论

学校读者我要写书评

暂无评论

Distilling Dynamic Spatial Relation Network for Human Pose Estimation 32

Distilling Dynamic Spatial Relation Network for Human Pose E...

引用

32nd British Machine Vision Conference, BMVC 2021

作者： Wu, Kewei Wang, Tao Xie, Zhao Guo, Dan Key Laboratory of Knowledge Engineering with Big Data Hefei University of Technology Ministry of Education Hefei China School of Computer Science and Information Engineering Hefei University of Technology Hefei China

Human pose estimation is a challenging task that requires the comprehension of the pose structure. This work can refer to spatial relation inference in a pose structure model;how to model the dynamic spatial relation against various unreliable joints is critical. To this end, we propose a Distilling Dynamic Spatial Relation network (DDSR), which builds pose-based graph representation by exploiting the feature of spatial relation from the location distribution of joints. We use a dynamic message propagation mechanism to update the spatial relation on edges. Specifically, to filter out the noisy predictions, we select the joints with high confidence;to enhance the spatial relation in a large receptive field, we propagate multi-stage messages among joints. Besides, to reduce the computation cost of the multi-stage message propagation, we design a cross-resolution distillation framework. We use a new spatial distillation loss to verify the spatial relation between the teacher model and the student model. Experimental results on COCO and MPII datasets show that our method is superior to the state-of-the-art methods. The visualization results further verify the interpretability of our spatial relation. © 2021. The copyright of this document resides with its authors.

关键词： Distillation

来源：评论

学校读者我要写书评

暂无评论

Electronic Medical Record Sharing System Based on Hyperledger Fabric and InterPlanetary File System 2021

Electronic Medical Record Sharing System Based on Hyperledge...

引用

5th International Conference on Compute and data Analysis, ICCDA 2021

作者： Li, Lei Yue, Zhengxiang Wu, Gongqing Key Laboratory of Knowledge Engineering with Big Data Hefei University of Technology Ministry of Education Hefei China School of Computer Science and Information Engineering Hefei University of Technology Hefei China

ISBN: (纸本)9781450389112

Electronic Medical Records (EMR) and other medical data contain important and sensitive privacy information of patients, which provide important basis and reference for their doctors to diagnose and treat them. With the continuous development of data sharing, big data applications and other technologies, new requirements are proposed for the storage of medical privacy data. At present, medical data have the characteristics of obstructed sharing and exchange, difficult data tracing and data sensitivity, which increases the cost and difficulty of each diagnosis. Moreover, patients lack the right to know and control their own medical data, which is easy to cause data fraud or data leakage. As a decentralized, tamper proof and traceable distributed ledger system, blockchain brings new solutions to these problems. This paper designs a medical data fusion and distributed privacy management system based on the Hyperledger Fabric blockchain development platform and InterPlanetary File System. It can not only solve the problem of mutual distrust in the traditional system, but also realize the safe data storage and the safe sharing of medical record data between medical organizations which is a beneficial attempt to overcome the disadvantages of traditional hospital data separation. At the same time, according to the requirements of different levels and types of medical users or groups for data security and personal privacy protection, the access control technology of users is studied to ensure personal privacy data. © 2021 ACM.

关键词： Blockchain

来源：评论

学校读者我要写书评

暂无评论

An Encoder-Based Framework for Privacy-Preserving Machine Learning 24th

An Encoder-Based Framework for Privacy-Preserving Machine L...

引用

24th International Conference on Algorithms and Architectures for Parallel Processing, ICA3PP 2024

作者： Wu, Jiayun Ren, Wei Zhang, Xianchao Zheng, Xianghan School of Computer Science China University of Geosciences Wuhan430074 China State Key Laboratory of Geo-Information Engineering and Key Laboratory of Surveying and Mapping Science and Geospatial Information Technology of MNR CASM Beijing China Ministry of Education Chengdu China Key Laboratory of Medical Electronics and Digital Health of Zhejiang Province Jiaxing University Jiaxing China Engineering Research Center of Intelligent Human Health Situation Awareness of Zhejiang Province Jiaxing University Jiaxing China College of Computer and Big Data Fuzhou University Fujian Fuzhou China School of Information and Intelligent Engineering Sanya College Hainan China

ISBN: (纸本)9789819615506

As a data-driven science, machine learning requires vast amounts of training data and computational resources. However, for highly privacy-sensitive data, it is crucial to protect the privacy of the data during both the training and utilization of machine learning models. In this paper, we propose a privacy-preserving machine learning approach using autoencoders and differential privacy mechanisms to safeguard data privacy while minimizing the impact on data availability. Specifically, we augment logistic regression and ResNet18 models with different architectures of autoencoders to perform data encryption? without compromising the machine learning tasks. Additionally, we employ differential privacy mechanisms to introduce gradient perturbations in the encoding part of the autoencoder, enhancing the algorithm’s security and further protecting data privacy. We also design the cosine similarity between the encoded and original data as a metric for evaluating data privacy, considering model performance, privacy budget, and data privacy collectively to balance data availability and privacy. Extensive experiments conducted on MNIST, CIFAR-10, PathMNIST, and BloodMNIST datasets demonstrate that for simple logistic regression models handling easily classifiable datasets, employing simple autoencoder structures can enhance classification accuracy, with significant performance impact after adding differential privacy. For ResNet18, utilizing convolutional autoencoders for data encryption generally has minimal impact on model classification performance and can even improve accuracy in most cases. Adding differential privacy has minor effects on model classification performance. Selecting appropriate model structures and privacy budgets for different usage scenarios can ensure both data availability and privacy. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.

关键词： Adversarial machine learning

来源：评论

学校读者我要写书评

暂无评论

Marginal Value-Based Edge Resource Pricing and Allocation for Deadline-Sensitive Tasks

Marginal Value-Based Edge Resource Pricing and Allocation fo...

引用

IEEE Annual Joint Conference: INFOCOM, IEEE Computer and Communications Societies

作者： Puwei Wang Zhouxing Sun Ying Zhan Haoran Li Xiaoyong Du School of Information Renmin University of China Key Laboratory of Data Engineering and Knowledge Engineering of Ministry of Education Renmin University of China Beijing China Guizhou University of Finance and Economics Guiyang City China

In edge computing (EC), resource allocation is to allocate computing, storage and networking resources on the edge nodes (ENs) efficiently and reasonably to tasks generated by users. Due to the resource-limitation of ENs, the tasks often need to compete for the resources. Pricing mechanisms are widely used to deal with the resource allocation problem, and the valuations of tasks play a critical role in the price mechanisms. However, users naturally are not willing to expose the valuations of their tasks due to conflicts of interests. Current research works usually adopt truthful auctions to motivate the users to report honestly the valuations of their tasks. In this paper, we introduce the marginal value to estimate the valuations of tasks, and propose a marginal value-based pricing mechanism using the incentive theory, which motivates the tasks with higher marginal values to actively request more resources. The EC platform sets the resource prices using the price mechanism, and then the users determine their resource requests relying on the resource prices and the valuations of their tasks. After receiving the deadline-sensitive tasks from the users, the resource allocation can be modeled as a knapsack problem with the deadline constraints. Extensive experimental results demonstrate that our approach is computationally efficient and is promising in enhancing the utility of the EC platform and the tasks.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Enhancing Unsupervised Graph Few-shot Learning via Set Functions and Optimal Transport 25

Enhancing Unsupervised Graph Few-shot Learning via Set Funct...

引用

Proceedings of the 31st ACM SIGKDD Conference on knowledge Discovery and data Mining V.1

作者： Yonghao Liu Fausto Giunchiglia Ximing Li Lan Huang Xiaoyue Feng Renchu Guan Key Laboratory of Symbolic Computation and Knowledge Engineering of the Ministry of Education College of Computer Science and Technology Jilin University Changchun China Department of Information Engineering and Computer Science University of Trento Trento Italy

ISBN: (纸本)9798400712456

Graph few-shot learning has garnered significant attention for its ability to rapidly adapt to downstream tasks with limited labeled data, sparking considerable interest among researchers. Recent advancements in graph few-shot learning models have exhibited superior performance across diverse applications. Despite their successes, several limitations still exist. First, existing models in the meta-training phase predominantly focus on instance-level features within tasks, neglecting crucial set-level features essential for distinguishing between different categories. Second, these models often utilize query sets directly on classifiers trained with support sets containing only a few labeled examples, overlooking potential distribution shifts between these sets and leading to suboptimal performance. Finally, previous models typically require necessitate abundant labeled data from base classes to extract transferable knowledge, which is typically infeasible in real-world scenarios. To address these issues, we propose a novel model named STAR, which leverages Set funcTions and optimAl tRansport for enhancing unsupervised graph few-shot learning. Specifically, STAR utilizes expressive set functions to obtain set-level features in an unsupervised manner and employs optimal transport principles to align the distributions of support and query sets, thereby mitigating distribution shift effects. Theoretical analysis demonstrates that STAR can capture more task-relevant information and enhance generalization capabilities. Empirically, extensive experiments across multiple datasets validate the effectiveness of STAR. Our code can be found https://***/KEAML-JLU/STAR here.

关键词： few-shot learning

来源：评论

学校读者我要写书评

暂无评论

Maximum Entropy Inverse Reinforcement Learning based Trajectory Planning for Autonomous Driving 3

Maximum Entropy Inverse Reinforcement Learning based Traject...

引用

3rd International Conference on High Performance Computing and Communication engineering, HPCCE 2023

作者： Zhang, Puhao Xie, Shumin Lu, Xiaoya Zhong, Zuodong Li, Qing School of Automation and Electrical Engineering Beijing University of Science and Technology 30 Xueyuan Road Haidian District Beijing100083 China School of Mechanics and Civil Engineering and Architecture Northwestern Polytechnical University 127 Youyi West Road Xi'an710072 China Key Laboratory of Knowledge Automation for Industrial Processes Ministry of Education 30 Xueyuan Road Haidian District Beijing100083 China

ISBN: (数字)9781510674653

ISBN: (纸本)9781510674646

With the rapid advancement of autonomous driving technology, effective trajectory planning has become crucial for ensuring road safety and driving efficiency. Traditional trajectory planning methods often rely on preset rules and models, making them ill-suited for the complex and dynamic traffic environment. To address this, a Maximum Entropy Inverse Reinforcement Learning (MaxEnt IRL) based trajectory planning method is proposed in this paper, aiming at learning from expert driving behaviors to infer an efficient reward function, which in turn guides decision-making and path planning for autonomous vehicles. This study begins by analyzing expert driving data to extract key state and action features. Then, the MaxEnt IRL algorithm is applied to learn the reward function underlying these features, reflecting the decision-making logic of expert drivers. The learned reward function is subsequently used to guide the trajectory planning of the autonomous driving system, generating safe and efficient driving paths. A series of experiments conducted in a simulated environment demonstrate that the MaxEnt IRL-based method proposed in this paper exhibits higher adaptability and efficiency in handling complex traffic scenarios compared to traditional trajectory planning methods. © 2024 SPIE.

关键词： Decision making

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：