检索结果-内蒙古大学图书馆

5th International Conference on Artificial Intelligence and Computer Engineering, ICAICE 2024

作者： Zheng, Fengkai The Key Laboratory of Embedded System and Service Computing Ministry of Education Tongji University Shanghai China

ISBN: (纸本)9798400718007

Accurate long-term traffic forecasting is crucial for urban traffic management and planning. Existing methods focus mainly on short-term predictions and struggle with long-term spatial-temporal dependencies due to computational complexities. We propose a Spatial-Temporal Multi-View Graph Transformer (STMGT) based on spatial-temporal sequence decomposition. STMGT enhances long-term forecasting by decomposing traffic sequences into temporal position embeddings, periodicity-trend features, and residual components. This approach captures long-term correlations while managing computational cost by dividing data into multiple time blocks. A multi-head graph attention mechanism is used to explore dynamic spatial correlations. Experimental results on real-world datasets show that STMGT significantly outperforms baseline models, proving it effective for long-term traffic prediction. © 2024 Copyright held by the owner/author(s). Publication rights licensed to ACM.

关键词： graph attention mechanism

来源：评论

学校读者我要写书评

暂无评论

Edge-weight-Based link prediction in heterogeneous graph 19

Edge-weight-Based link prediction in heterogeneous graph

引用

19th IEEE International Conference on Networking, Sensing and Control, ICNSC 2022

作者： Zong, Jie Ding, Zhijun The Key Laboratory of Embedded System and Service Computing Ministry of Education Tongji University Shanghai China

ISBN: (纸本)9781665472432

Link prediction is to predict whether there is a link between two nodes in the graph, it is a very important application and plays a great role in various industries. In recent years, with the development of graph neural network technology, many algorithms make effort to study the expression of each node from the original graph data and use them to infer new links. However, most of the existing algorithms have a common problem when facing heterogeneous graphs, which is, they do not consider the weight of edges in graphs. Instead, they put all their energies into computing node features. Although a few algorithms such as RGCN are trying to take the influence of different link types into account while extracting node features, these implicit feature extractions do not start from the global information, but just calculate independently for each node. In other words, in these algorithms, even the same type of links will be abstracted into different features on different nodes. This is obviously inconsistent with reality. On the same map, the feature of the same link should be relatively fixed and should not be changed just because of different positions. In addition, when the current graph neural network algorithm is applied to link prediction, the link type to be predicted must be specified in advance, which makes the algorithm extremely inflexible. In order to solve these problems, we propose an edge weight calculation algorithm that extracts the edge feature from the whole graph. We also propose the edge-weight-based link prediction algorithm. By introducing edge weight into the MLP, there is no need to specify the target link type at the beginning of model training. It improves both the performance and efficiency of the link prediction model. Experiments on two datasets show that this edge-weight-based link prediction algorithm performs better than current algorithms and reaches SOTA. © 2022 IEEE.

关键词： Graph neural networks

来源：评论

学校读者我要写书评

暂无评论

Adaptive Random Forest with Dynamic Detectors for Evolving Data Stream Classification 23

Adaptive Random Forest with Dynamic Detectors for Evolving D...

引用

9th International Conference on computing and Artificial Intelligence, ICCAI 2023

作者： Jin, Xuancheng Zhang, Yaying Key Laboratory of Embedded System and Service Computing Ministry of Education Tongji University Shanghai China

ISBN: (纸本)9781450399029

Evolving data streams containing concept drift breaks the assumption that data are independent and identically distributed (IID) in traditional machine learning models. A series of ensemble-based models have been adapted to data stream learning tasks due to the abilities of dynamic partially update and cooperate with drift detection methods. Adaptive Random Forests (ARF), one of the state-of-the-art in the stream setting, encounters high resource requirements in both efficiency and memory like all ensemble methods. We are committed to reducing running cost without compromising ARF’s performance. In this paper, we propose the Adaptive Random Forest with Dynamic Detectors (ARFDD) which utilizes diversity measurement to combine homogeneous detectors and split heterogeneous ones accordingly, leads to lower computing costs and thus speed up in applications. Experiments on all 8 synthetic and 4 real datasets show a great CPU-time thrift while avoiding performance degradation. © 2023 Copyright held by the owner/author(s).

关键词： Learning systems

来源：评论

学校读者我要写书评

暂无评论

Research on Behavior Element Decision of Cloud Edge Collaboration Capability Based on 5g Automatic Driving Scene 5

Research on Behavior Element Decision of Cloud Edge Collabor...

引用

5th IEEE Information Technology, Networking, Electronic and Automation Control Conference, ITNEC 2021

作者： Yan, Chungang Sheng, Shuo Key Laboratory of Embedded System and Service Computing Ministry of Education Tongji University Shanghai China

ISBN: (纸本)9781665415989

In order to meet the cloud side requirements of various traffic business scenarios, the construction of traffic cloud side collaboration platform realizes the collaboration and unification of various business resources and data on both sides of the cloud side, and the cloud side integration realizes the real-time and efficient "vehicle road cloud side"collaboration of traffic. In order to quickly meet the needs of various scenarios and provide customized cloud edge capability services, by drawing on the experience and technology R D capabilities of Companies in the industry, it helps to shorten the overall development period of the platform, reduce costs and technical risks, and quickly form product capabilities. Based on the cloud edge collaboration capability, the super edge node collects structured and unstructured data in real time, uses the edge computing power to label, filter and design thresholds, so as to detect and filter the scene behavior elements in 5g + traffic scenarios, further clarify the actor's data, gather it to the cloud for research and judgment, and improve the global decision-making ability of the cloud, Improve the real-time flow of dynamic traffic element resources. © 2021 IEEE.

关键词： 5G mobile communication systems

来源：评论

学校读者我要写书评

暂无评论

A perspective on Petri Net learning

引用

Frontiers of Computer Science 2023年第6期17卷 1-3页

作者： Hongda QI Changjun JIANG Key Laboratory of Embedded System and Service Computing(Ministry of Education) Tongji UniversityShanghai 201804China National(Province-Ministry Joint)Collaborative Innovation Center for Financial Network Security Tongji UniversityShanghai 201804China

Petri Nets(PNs)are used for modeling and analyzing discreteevent systems,such as communication protocols,trafficsystems,human-computer interaction,and fault ***’state space explosion problem means that the state spaceof PNs grows exponentially with PNs’*** thefundamental reachability problem is still an NP-Hard problemin *** has been proved that the equivalence problem forthe reachability sets of arbitrary PNs is undecidable except forsome subclass of PNs[1].That is,the reachability problem ofarbitrary PNs cannot be solved ***,there is noefficient and accurate algorithm to solve the problem.10172In recent years,with the emergence of big data and thedevelopment of computing hardware,a series ofbreakthroughs have been achieved in machine learning,suchas AlphaGo,AlphaFold,and ChatGPT[2−4].As a data-drivenapproach,machine learning can learn potential mappingrelationships between inputs and outputs from large-scaledata.

关键词： breakthrough hardware learning

来源：评论

学校读者我要写书评

暂无评论

Genetic Fuzzy + PI Path Tracking Control of a Nonholonomic Mobile Robot

引用

电子学报(英文版) 2011年第1期20卷 31-34页

作者： CAO Zhengcai ZHAO Yingtao WU Qidi College of Information Science and Technology Beijing University of Chemical Technology Beijing China The Key Laboratory of Embedded System and Service Computing Ministry of Education Tongji University Shanghai China

A mobile robot is one of the well-known nonholonomic systems. The genetic fuzzy + PI path track- ing control applied to a mobile robot is presented. The pro- posed controller composes of two fuzzy + PI controllers which are employed to control the angular velocity to- gether, and one fuzzy + PI controller that is utilized to adjust the speed of mobile robot solely. Moreover, the fuzzy controller is designed to track reference path in ap- proaching phase and PI controller is operated to diminish the tracking error in stable phase. In order to obtain bet- ter tracking performance, Genetic algorithm (GA) is used to optimize controller parameters. Finally, computer sim- ulation results are given to illustrate the effectiveness of the proposed control scheme.

关键词：模糊PI控制器移动机器人机器人控制非完整系统路径跟踪遗传算法模糊控制器控制器参数

来源：评论

学校读者我要写书评

暂无评论

Observable liveness of Petri nets with controllable and observable transitions

引用

Science China(Information Sciences) 2017年第11期60卷 256-264页

作者： Guanjun LIU Changjun JIANG Department of Computer Science Tongji University Key Laboratory of Embedded System and Service Computing of Ministry of Education Tongji University E-commerce Transactions and Information Services Collaborative Innovation Center of Shanghai University Knowledge Service Platform Tongji University

Observability and/or controllability are paid much attention to by researchers not only from the control science field[1–3]but also from the computer science field[4–9].Recently,Desel and Kilinc[4]proposed the notion of observable liveness for Petri nets with(un-)observable and(un-)controllable *** purpose of defining observable liveness is to represent that"a user can always enforce the occurrence of any observable transition only by stimulating the net by choosing appropriate enabled controllable transition"[4].An important assumption of

关键词： net cot PN Observable liveness of Petri nets with controllable and observable transitions

来源：评论

学校读者我要写书评

暂无评论

Local-directed network coding in vehicular ad-hoc networks

Local-directed network coding in vehicular ad-hoc networks

引用

18th International Symposium on Intelligent Signal Processing and Communication systems, ISPACS 2010

作者： Haojie, Luo Zhang, Dongdong Ye, Chen Key Laboratory of Embedded System and Service Computing Ministry of Education Tongji University Shanghai 200092 China

ISBN: (纸本)9781424473717

In this paper, we proposed a novel network coding method called local-directed network coding to improve the throughput of the vehicular ad-hoc network. Different from other vehicular network coding methods, we consider the direction of packet transmission. In detail, the intermediate node selects one main packet and some sub-packets from forwards and backwards sending queues, and then encodes these packets into one packet to broadcast. Simulation results show that our approach can efficiently improve the throughput of application layer in vehicular networks. © 2010 IEEE.

关键词： Network coding

来源：评论

学校读者我要写书评

暂无评论

Multi-View Metrics Enhanced Heterogeneous Graph Neural Network for Extractive Summarization

Multi-View Metrics Enhanced Heterogeneous Graph Neural Netwo...

引用

2021 China Automation Congress, CAC 2021

作者： Zhang, Chaobo Wang, Junli Qi, Hongda Yan, Chungang Jiang, Changjun Key Laboratory of Embedded System and Service Computing Ministry of Education Tongji University Shanghai201804 China

ISBN: (纸本)9781665426473

Currently, many works regarded extractive summarization as a binary classification problem. However, the metric of the summary classification is always too singular and cannot fully utilize sentence features. To address this issue, we propose a multi-view metrics enhanced heterogeneous graph neural network for extractive summarization (MHGS). Firstly, the text is modeled as a heterogeneous graph, which contains two types of nodes, namely, words and sentences. Secondly, in the graph update layer, cross-sentence relationships and long-distance associations are obtained through the indirect connection of words and the direct connection between sentences. Then, we design multi-view metrics including relevance, redundancy, new information, and Rouge to help the model improve the quality of the summary. Finally, to evaluate the effectiveness of the method, some experiments are conducted on the CNN & DailyMail dataset. The experimental results show a generally higher performance of MHGS compared with related methods. © 2021 IEEE

关键词： Graph neural networks

来源：评论

学校读者我要写书评

暂无评论

Traffic flow prediction model based on deep belief network and genetic algorithm

Traffic flow prediction model based on deep belief network a...

引用

作者： Zhang, Yaying Huang, Guan Key Laboratory of Embedded System and Service Computing Ministry of Education Tongji University Shanghai200092 China

Traffic flow prediction plays an indispensable role in the intelligent transportation system. The effectiveness of traffic control and management relies heavily on the prediction accuracy. The authors propose a model based on deep belief networks (DBNs) to predict the traffic flow. Moreover, they use Fletcher-Reeves conjugate gradient algorithm to optimise the fine-tuning of model's parameters. Since the traffic flow has various features at different times such as weekday, weekend, daytime and nighttime, the hyper-parameters of the model should adapt to the time. Therefore, they employ the genetic algorithm to find the optimal hyper-parameters of DBN models for different times. The dataset from Caltrans Performance Measurement system was used to evaluate the performance of their models. The experimental results demonstrate that the proposed model achieved better performance in different times. © The Institution of Engineering and Technology 2018.

关键词： Genetic algorithms

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：