检索结果-内蒙古大学图书馆

arXiv 2022年

作者： Sun, Li Ye, Junda Peng, Hao Wang, Feiyang Yu, Philip S. School of Control and Computer Engineering North China Electric Power University Beijing102206 China School of Computer Science Beijing University of Posts and Telecommunications Beijing100876 China Beijing Advanced Innovation Center for Big Data and Brain Computing Beihang University Beijing100191 China Department of Computer Science University of Illinois ChicagoIL United States

Continual graph learning routinely finds its role in a variety of real-world applications where the graph data with different tasks come sequentially. Despite the success of prior works, it still faces great challenges. On the one hand, existing methods work with the zero-curvature Euclidean space, and largely ignore the fact that curvature varies over the coming graph sequence. On the other hand, continual learners in the literature rely on abundant labels, but labeling graph in practice is particularly hard especially for the continuously emerging graphs on-the-fly. To address the aforementioned challenges, we propose to explore a challenging yet practical problem, the self-supervised continual graph learning in adaptive Riemannian spaces. In this paper, we propose a novel self-supervised Riemannian Graph Continual Learner (RieGrace). In RieGrace, we first design an Adaptive Riemannian GCN (AdaRGCN), a unified GCN coupled with a neural curvature adapter, so that Riemannian space is shaped by the learnt curvature adaptive to each graph. Then, we present a Label-free Lorentz Distillation approach, in which we create teacher-student AdaRGCN for the graph sequence. The student successively performs intra-distillation from itself and inter-distillation from the teacher so as to consolidate knowledge without catastrophic forgetting. In particular, we propose a theoretically grounded Generalized Lorentz Projection for the contrastive distillation in Riemannian space. Extensive experiments on the benchmark datasets show the superiority of RieGrace, and additionally, we investigate on how curvature changes over the graph sequence. © 2022, CC BY-NC-SA.

关键词： Distillation

来源：评论

学校读者我要写书评

暂无评论

TransKD: Transformer Knowledge Distillation for Efficient Semantic Segmentation

arXiv

引用

arXiv 2022年

作者： Liu, Ruiping Yang, Kailun Roitberg, Alina Zhang, Jiaming Peng, Kunyu Liu, Huayao Wang, Yaonan Stiefelhagen, Rainer The Institute for Anthropomatics and Robotics Karlsruhe Institute of Technology Karlsruhe76131 Germany The School of Robotics The National Engineering Laboratory of Robot Visual Perception and Control Technology Hunan University Changsha410082 China The Institute for Artificial Intelligence The University of Stuttgart Stuttgart70569 Germany The Institute for Visual Computing ETH Zurich Zurich8092 Switzerland NIO Shanghai201804 China

Semantic segmentation benchmarks in the realm of autonomous driving are dominated by large pre-trained transformers, yet their widespread adoption is impeded by substantial computational costs and prolonged training durations. To lift this constraint, we look at efficient semantic segmentation from a perspective of comprehensive knowledge distillation and aim to bridge the gap between multi-source knowledge extractions and transformer-specific patch embeddings. We put forward the Transformer-based Knowledge Distillation (TransKD) framework which learns compact student transformers by distilling both feature maps and patch embeddings of large teacher transformers, bypassing the long pre-training process and reducing the FLOPs by >85.0%. Specifically, we propose two fundamental modules to realize feature map distillation and patch embedding distillation, respectively: (1) Cross Selective Fusion (CSF) enables knowledge transfer between cross-stage features via channel attention and feature map distillation within hierarchical transformers;(2) Patch Embedding Alignment (PEA) performs dimensional transformation within the patchifying process to facilitate the patch embedding distillation. Furthermore, we introduce two optimization modules to enhance the patch embedding distillation from different perspectives: (1) Global-Local Context Mixer (GL-Mixer) extracts both global and local information of a representative embedding;(2) Embedding Assistant (EA) acts as an embedding method to seamlessly bridge teacher and student models with the teacher’s number of channels. Experiments on Cityscapes, ACDC, NYUv2, and Pascal VOC2012 datasets show that TransKD outperforms state-of-the-art distillation frameworks and rivals the time-consuming pre-training method. The source code is publicly available at https://***/RuipingL/TransKD. Copyright © 2022, The Authors. All rights reserved.

关键词： Semantic Segmentation

来源：评论

学校读者我要写书评

暂无评论

Attention-Driven Graph Convolutional Networks for Deadline-Constrained Virtual Machine Task Allocation in Edge computing

引用

IEEE Transactions on Consumer Electronics 2025年

作者： Ali, Ahmad Ullah, Inam Singh, Sushil Kumar Jiang, Weiwei Alturise, Fahad Bai, Xiaoshan Shenzhen University College of Mechatronics and Control Engineering Shenzhen518060 China Shenzhen University National Engineering Laboratory for Big Data System Computing Technology Shenzhen518060 China Beijing University of Posts and Telecommunications Beijing China Qassim University College of Computer Department of Cybersecurity Buraydah Saudi Arabia

Effective load balancing is critical in contemporary mobile computing environments to ensure optimal resource utilization and adherence to strict deadlines. This study presents a novel framework for deadline-constrained load balancing, termed DCLD-net, which leverages the integration of Graph Convolutional Networks (GCN) and an attention-based Bidirectional Gated Recurrent Unit (BiGRU). The proposed framework utilizes the spatial modeling capabilities of GCNs to capture intricate network topologies, while the temporal dynamics and attention mechanisms of BiGRU enable the identification of critical patterns in traffic flow data. By incorporating key factors such as sticky sessions, load distribution, content-based factors such as videos or file downloads, instance health checks, and other relevant parameters, the DCLD-net model achieves high accuracy in predicting task schedules. These schedules allocate tasks to virtual machines (VMs), enabling dynamic traffic routing through a load balancer that efficiently assigns tasks. Extensive experimental evaluations demonstrate that the proposed attention-based BiGRU-GCN model significantly outperforms traditional methods in both deadline compliance and load balancing efficiency, offering a robust solution to the challenges of modern cloud computing. Experimental results demonstrate that DCLD-net outperforms traditional methods by 27% in deadline compliance and 29.3% in load balancing efficiency. © 1975-2011 IEEE.

关键词： Mobile edge computing

来源：评论

学校读者我要写书评

暂无评论

Novel Scheme Towards Interfacial Charge Transfer between Znin2s4 and Biobr for Efficient Photocatalytic Removal of Organics and Chromium (Vi) from Water

SSRN

引用

SSRN 2022年

作者： He, Zuming Yang, Hanpei Sunarso, Jaka Wong, Ngie Hing Huang, Zhengyi Xia, Yongmei Wang, Yong Su, Jiangbin Wang, Lina Kang, Li Key Laboratory of Integrated Regulation and Resource Development on Shallow Lakes Ministry of Education College of Environment Hohai University Nanjing210098 China Huaide School Changzhou University Jingjiang214500 China School of Microelectronics and Control Engineering Changzhou University Changzhou213164 China Research Centre for Sustainable Technologies Faculty of Engineering Computing and Science Swinburne University of Technology Jalan Simpang Tiga Sarawak Kuching93350 Malaysia School of Materials and Engineering Jiangsu University of Technology Changzhou213001 China School of Pharmaceutical and Materials Engineering Taizhou University Jiaojiang318000 China

Construction of Z-scheme heterostructure is an effective strategy to enhance the charge carriers' separation. However, successfully achieving this on the defect heterojunction to improve the photocatalytic activity remains challenging. This work successfully obtained sulfur vacancy in the ZnIn2S4/BiOBr (SZIS/BOB) heterojunction composites with S-O covalent bonding using a hydrothermal method. As a result, they exhibited superior photocatalytic and stability performance. The optimized SZIS/BOB-10 exhibited excellent rhodamine B degradation (95.2%) and chromium (VI) reduction (97.8%) within 100 minutes under visible light. The enhanced composites with S-vacancies, S-O bond, and internal electric field induced the Z-scheme charge transfer mechanism. We had verified this mechanism based on the surface photovoltage spectra, electron spin response spectra, and density functional theory calculations. This work not only provides valuable insights into designing photocatalysts with a direct Z scheme heterostructure but also delineates a promising strategy for developing efficient photocatalysts to degrade organic pollutants. © 2022, The Authors. All rights reserved.

关键词： Density functional theory

来源：评论

学校读者我要写书评

暂无评论

Optimal Minimum Energy Path Planning of Mobile Robot Based on Improved WOA-GA with Hybrid Strategy for Community-Centric Application

Optimal Minimum Energy Path Planning of Mobile Robot Based o...

引用

International Conference on Machine Learning and Cybernetics (ICMLC)

作者： Jiang Du Zhou Da-Lin Li Gong-Fa Lu Bao-Li Yun Jun-Tong Liu Ying Key Laboratory of Metallurgical Equipment and Control Technology of Ministry of Education Wuhan University of Science and Technology Wuhan China School of Computing University of Portsmouth Portsmouth UK Hubei key laboratory of hydroelectric machinery design & maintenance China Three Gorges University Yichang China Key Laboratory of Mechanical Transmission and Manufacturing Engineering of Hubei Province Wuhan University of Science and Technology Wuhan China Precision Manufacturing Research Institute Wuhan University of Science and Technology Wuhan China Institute of Semiconductors Chinese Academy of Sciences Beijing China Research Center for Biomimetic Robot and Intelligent Measurement and Control Wuhan University of Science and Technology Wuhan China

The path planning of mobile robots is investigated to minimize energy loss in intelligent manufacturing. To achieve optimal energy consumption planning, this paper proposes a hybrid strategy that enhances the whale optimization algorithm (WOA) and the genetic algorithm (GA). The WOA-GA algorithm combines their respective advantages and effectively mitigates premature convergence exhibited by the standard traditional algorithm. To enhance the global and local search capabilities of the standard WOA, we employ adaptive weighting and a Cauchy mutation strategy. Additionally, we introduce an adaptive crossover mutation probability and an elite strategy to mitigate the detrimental effects of the genetic algorithm's evolutionary operation in the late iterations, thereby improving its performance. This paper compares the performance of the standard GA, the standard WOA-GA algorithm, and the enhanced WOA-GA algorithm. The experimental results demonstrate that the enhanced WOA-GA algorithm can achieve optimal energy-efficient path planning using the hybrid strategy. The average energy consumption of the enhanced WOA-GA algorithm can be reduced by 30%, while the maximum energy consumption can be reduced by 47%. The results confirm the effectiveness and feasibility of the proposed method in enhancing the level of green manufacturing.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Short-term Wind Power Predication Based on Feature engineering and SOA-SVM 5

Short-term Wind Power Predication Based on Feature Engineeri...

引用

5th IEEE Conference on Energy Internet and Energy System Integration, EI2 2021

作者： Zhang, Wenfeng Yan, Renwu Yu, Ning Liu, Yunzhang Zhang, Yang College of Electronic Electrical and Physics FuJian University of Technology Fuzhou350118 China College of Electronic Electrical and Physics Fujian Provincial University Fuzhou350118 China Engineering Research Center for Smart Grid Simulation Analysis and Integrated Control FuJian University of Technology Fuzhou350118 China Department of Computing Sciences SUNY Brockport BrockportNY14420 United States College of Electronic EFuJian University of Tlectrical and Physics Echnology Fuzhou350118 China

ISBN: (纸本)9781665434256

Contraposing to the issues of strong randomness, large volatility, and high prediction error of wind power, this paper proposes a short-term wind power prediction method based on feature engineering and support vector machine optimized by the seagull optimization algorithm (ISOA-SVM). First, in feature engineering, the abnormal values are preprocessed and utilizing the Pearson correlation coefficient (PCC) method to analyze the correlation between wind power and every feature. On this basis, a set of new features are constructed. Then, in the construction of the combined model, the seagull optimization algorithm is utilized to optimize the parameters of the support vector machine, so as to realize the complementary advantages of the optimization characteristics of the seagull algorithm and the support vector machine in simplifying the high-dimensional space complexity problems. Finally, the short-term wind power prediction model is constructed based on the ISOA-SVM model. The emulation experiment is carried out with the relevant measurement data of a wind power field. The results show that, in relation to other models, the proposed ISOA-SVM model has more excellent precision and dependability of short-term wind power prediction, which provides a certain reference for power dispatching. © 2021 IEEE

关键词： Support vector machines

来源：评论

学校读者我要写书评

暂无评论

Spectrometric Characterization of Suspension Liquid and Light Extinction Model Update

SSRN

引用

SSRN 2022年

作者： Liu, Sheng Shen, Yang Gao, Zihui Gao, Hongwen School of Computer Science and Technology Huaibei Normal University Huaibei235000 China State Key Laboratory of Enviromental Pollution Control and Resource Reuse Tongji University Shanghai200092 China Shanghai GreenEmpire Environmental Science and Technology Corporation Shanghai200092 China Anhui Engineering Research Center for Intelligent Computing and Application on Cognitive Behavior Huaibei235000 China

On the basis of the classical light scattering models, the light extinction model is the first to establish as [[EQUATION]] (, N and γ – average diameter in μm, number and relative refractive index of the suspending particles, λ , A and δ – incident light wavelength in μm, absorbance and optical path in cm of the suspension liquid) by spectrometric characterization of ten standard suspension liquids. It has been used to determine the suspending particles in the calcium oxalate, Formazine, soil, milk and sewage suspension water samples. As the result, the light extinction model method brought out less than 12% error of and 18% error of the suspending particles’ quality by comparing with the conventional methods. It provides a simple and reliable spectroptometric determination of a suspension liquid. Also, it is very potential to in-situ monitor the growth and working state of the suspending particles using in synthesis of materials, culture of cells, treatment of wastewater and safety of drinking water and foods. © 2022, The Authors. All rights reserved.

关键词： Suspensions (fluids)

来源：评论

学校读者我要写书评

暂无评论

DeepContact: High throughput quantification of membrane contact site based on electron microscopy imaging

DeepContact: High throughput quantification of membrane cont...

引用

2022生物物理大会

作者： Liqing Liu Shuxin Yang Yang Liu Xixia Li Junjie Hu Li Xiao Tao Xu National Laboratory of Biomacromolecules Institute of Biophysics Chinese Academy of Sciences Center for Biological Imaging Institute of Biophysics Chinese Academy of Sciences Key Laboratory of Intelligent Information Processing Institute of Computing TechnologyChinese Academy of Sciences School of Computer and Control Engineering University of Chinese Academy of Sciences College of Life Science University of Chinese Academy of Sciences Ningbo HuaMei Hospital University of Chinese Academy of Sciences School of Biomedical Engineering Guangzhou Medical University

Membrane contact site (MCS)-mediated organelle interactions plays essential roles in the cell. Quantitative analysis of the MCS reveals vital clues for cellular responses under various physiological and pathological conditions. However, an efficient tool is lacking. Here, we developed "DeepContact", a deep learning protocol for optimizing organelle segmentation and contact analysis based on label-free electron microscopy(EM).

关键词： DeepContact High throughput quantification of membrane contact site based on electron microscopy imaging

来源：评论

学校读者我要写书评

暂无评论

Research on Attention-Based Predictive Model in Multi-Agent Reinforcement Learning

Research on Attention-Based Predictive Model in Multi-Agent ...

引用

Chinese Automation Congress (CAC)

作者： Weixin Hong Huaiyu Wu Zhen Shen Gang Xiong Bin Hu Xisong Dong Fei-Yue Wang The State Key Laboratory for Management and Control of Complex Systems Institute of Automation Chinese Academy of Sciences Beijing China School of Artificial Intelligence University of Chinese Academy of Sciences Beijing China The Beijing Engineering Research Center of Intelligent Systems and Technology Institute of Automation Chinese Academy of Sciences Beijing China The Guangdong Engineering Research Center of 3D Printing and Intelligent Manufacturing Cloud Computing Center Chinese Academy of Sciences Dongguan China

Agent actions planning is a challenging problem in multi-agent reinforcement learning. Recent methods typically build their predictive models by full connection layers, but the shortage of utilizing action and observation information in complex multi-agent environment limits the performance of those methods. To address this issue, we propose a novel predictive model based on self-attention mechanism which can effectively use information in the environment. Specifically, we use the MADDPG algorithm to make a dataset where the input is the observations and actions and the output is the predicted observations and rewards. We also introduce the multi-head attention mechanism into the predictive model so that the predictive model can better combine different information for prediction. We perform extensive experiments to assess the proposed method and compare it with the representative baseline approach. Results demonstrate that our method outperforms competitors with higher accuracy in different tasks. Our implementation will be publicly available.

关键词： Automation Reinforcement learning Predictive models Prediction algorithms Planning Task analysis

来源：评论

学校读者我要写书评

暂无评论

Entity-enhanced Adaptive Reconstruction Network for Weakly Supervised Referring Expression Grounding

arXiv

引用

arXiv 2022年

作者： Liu, Xuejing Li, Liang Wang, Shuhui Zha, Zheng-Jun Li, Zechao Tian, Qi Huang, Qingming Institute of Computing Technology CAS Beijing100190 China University of Chinese Academy of Sciences Beijing100190 China Peng Cheng Laboratory Shenzhen518066 China The School of Information Science and Technology University of Science and Technology of China Hefei230027 China The School of Computer Science Nanjing University of Science and Technology Nanjing210094 China Huawei Cloud & AI Shenzhen518129 China The School of Computer and Control Engineering University of Chinese Academy of Sciences Beijing100190 China The Institute of Computing Technology CAS Beijing100190 China

Weakly supervised Referring Expression Grounding (REG) aims to ground a particular target in an image described by a language expression while lacking the correspondence between target and expression. Two main problems exist in weakly supervised REG. First, the lack of region-level annotations introduces ambiguities between proposals and queries. Second, most previous weakly supervised REG methods ignore the discriminative location and context of the referent, causing difficulties in distinguishing the target from other same-category objects. To address the above challenges, we design an entity-enhanced adaptive reconstruction network (EARN). Specifically, EARN includes three modules: entity enhancement, adaptive grounding, and collaborative reconstruction. In entity enhancement, we calculate semantic similarity as supervision to select the candidate proposals. Adaptive grounding calculates the ranking score of candidate proposals upon subject, location and context with hierarchical attention. Collaborative reconstruction measures the ranking result from three perspectives: adaptive reconstruction, language reconstruction and attribute classification. The adaptive mechanism helps to alleviate the variance of different referring expressions. Experiments on five datasets show EARN outperforms existing state-of-the-art methods. Qualitative results demonstrate that the proposed EARN can better handle the situation where multiple objects of a particular category are situated together. Copyright © 2022, The Authors. All rights reserved.

关键词： Semantics

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：