检索结果-内蒙古大学图书馆

40th National Conference of Theoretical Computer Science (NCTCS)

作者： Ba, Bing Fu, Qi Hang, Chongjing Jiang, YunXia Hunan Univ Sci & Technol Xiangtan Peoples R China

ISBN: (纸本)9789811981517;9789811981524

Traffic forecasting is fundamental to realizing intelligent transportation systems (ITS) and challenging due to the complicated spatial dependencies of traffic data and nonlinear temporal trends. graph convolutional networks (GCNs) have been employed in the latest studies to capture the intricate spatial relationships of roadways. However, due to the intrinsic restrictions of traditional GCNs, these methods cannot represent high-order or mixed neighborhood information. TheMixHop graphWaveNet (MH-GWN), a novel graph neural network architecture for traffic forecasting, is proposed in this research. In MH-GWN, a spatial-temporal convolutional layer that effectively integrates the MixHop graph convolutional layer and the dilated causal convolutional layer is designed, which can aggregate arbitrary-order neighborhood information and model the complex spatial-temporal dependencies of traffic data. Furthermore, via stacking spatial-temporal convolutional layers, the model's receptive field in the spatial-temporal domain can be exponentially improved. Extensive experiments on two real-world road network traffic datasets show that MH-GWN model is better than other baselines.

关键词： graph convolution neural network Traffic forecasting MixHop graph convolution Spatial-temporal structure

来源：评论

学校读者我要写书评

暂无评论

SVR: A Shard-aware Vertex Reordering Method for Efficient GNN Execution and Memory Access 19

SVR: A Shard-aware Vertex Reordering Method for Efficient GN...

引用

19th International SoC Design Conference (ISOCC) - SoC Technology Towards a New Era of Innovation

作者： Hu, Xingyuan Shao, Zhuang Xie, Chenjia Du, Li Du, Yuan Nanjing Univ Nanjing Peoples R China

ISBN: (纸本)9781665459716

graph neural networks (GNNs) appear to be a powerful approach to analyzing non-euclidean data structures and achieving unprecedented performance on graph processing tasks. However, due to the inherent randomness and sparsity of graph structures, GNNs require irregular execution and memory access patterns. This paper proposed a lightweight shard-aware vertex reordering (SVR) algorithm to exploit data reuse and reduce memory access based on the adjacency of vertices. The vertices are divided into fixed-size groups and the vertex indices are updated in each group according to their adjacency relationship. The preprocessed dataset has fewer shards with any non-zero element and the sparsity is significantly decreased. The SVR algorithm can be used as a preprocessing extension to current GNN accelerators, improving performance without changing existing data flow. The proposed SVR is implemented to the baseline HyGCN and EnGN architecture for evaluation and we achieve an additional 1.43x memory access reduction on average in the aggregation phase and up to 2.52x sparsity reduction in the effectual shards.

关键词： Hardware Accelerator graph convolution neural network Vertex Reordering

来源：评论

学校读者我要写书评

暂无评论

AcaVis: A Visual Analytics Framework for Exploring Evolution of Dynamic Academic networks 16th

AcaVis: A Visual Analytics Framework for Exploring Evolution...

引用

16th CCF Conference on Computer Supported Cooperative Work and Social Computing (ChineseCSCW)

作者： Lu, Qiang Wen, Dajiu Huang, Wenjiao Lin, Tianyue Ma, Cheng Hefei Univ Technol Key Lab Knowledge Engn Big Data Minist Educ Hefei Anhui Peoples R China Hefei Univ Technol Sch Comp Sci & Informat Hefei Anhui Peoples R China Hefei Univ Technol Sch Foreign Studies Hefei Anhui Peoples R China Hefei Univ Technol Intelligent Interconnected Syst Lab Anhui Prov Hefei Anhui Peoples R China

ISBN: (纸本)9789811945496;9789811945489

In today's academic society, the trend of complexity, internationalization, diversification and interdisciplinary of scientific research activities is becoming more and more obvious. Although people are increasingly interested in academic cooperation in different fields and disciplines, people still know little about cross-field and interdisciplinary cooperation mechanism in academic big data. In order to further understand the topic and structure change pattern of academic big data, and explore the collaborative relationship, we propose a visual analytics framework to explore the time evolution of the dynamic academic network in the context of its structure in an interactive way. The first step is to learn the potential representation of the structure and attribute characteristics of nodes in the network based on graph convolution neural network (GCN). The second step is to optimize the learning process based on the end-to-end network combinatorial optimization algorithm, and then find communitieswith similar structures and similar attributes. The third step is to design AcaVis, a visual analytics framework to enable users to explore the topic and structure change characteristics of nodes, communities and the whole network in the academic social network. We demonstrated the effectiveness of our approach through three case studies conducted with real-world datasets and positive feedback from users. Code for our method is available at https://***/VIMLab-hfut/AcaVis.

关键词： Dynamic academic network graph convolution neural network Combinatorial optimization Visual analytics

来源：评论

学校读者我要写书评

暂无评论

A novel extended rule-based system based on K-Nearest Neighbor graph

引用

INFORMATION SCIENCES 2024年 662卷

作者： Fu, Yang-Geng Lin, Xin-Yi Fang, Geng-Chao Li, Jin Cai, Hong-Yi Gong, Xiao-Ting Wang, Ying-Ming Fuzhou Univ Coll Comp & Data Sci Fuzhou 350108 Peoples R China Fuzhou Univ Decis Sci Inst Fuzhou 350108 Peoples R China Wuchang Univ Technol Coll Business Wuhan 430223 Peoples R China

The Belief Rule -Based (BRB) system faces the rule combination explosion issue, making it challenging to construct the rule base efficiently. The Extended Belief Rule -Based (EBRB) system offers a solution to this problem by using data -driven methods. However, using EBRB system requires the traversal of the entire rule base, which can be time-consuming and result in the activation of many irrelevant rules, leading to an incorrect decision. Existing search optimization methods can somewhat solve this issue, but they have limitations. Moreover, the calculation of the rule activation weight only considers the similarity between input data and a single rule, ignoring the influence of the rule linkage. To address these problems, we propose a new EBRB system based on the K -Nearest Neighbor graph index (graph-EBRB). We introduce the Hierarchical Navigable Small World (HNSW) algorithm to create the K -Nearest Neighbor graph index of the EBRB system. This index allows us to efficiently search and activate a set of key rules. We also propose a new activation weight calculation method based on the graph convolution neural network (GCN), and we optimize the system performance using a parameter learning strategy. We conduct a comprehensive experiment on 14 commonly used public data sets, and the results show that graph-EBRB system significantly improves the reasoning efficiency and accuracy of the EBRB system. Finally, we apply the graph-EBRB system to tree disease identification and achieve excellent classification performance, identifying over 90% of the diseased trees on the complete dataset.

关键词： Extended belief rule-based system K-Nearest Neighbor graph graph convolution neural network Hierarchical navigable small world graph

来源：评论

学校读者我要写书评

暂无评论

Towards point cloud completion: Point Rank Sampling and Cross-Cascade graph CNN

引用

NEUROCOMPUTING 2021年 461卷 1-16页

作者： Zhu, Liping Wang, Bingyao Tian, Gangyi Wang, Wenjie Li, Chengyang China Univ Petr Coll Informat Sci & Engn Beijing Peoples R China China Univ Petr Beijing Key Lab Petr Data Min Beijing Peoples R China Peking Univ Sch Elect Engn & Comp Beijing Peoples R China

The Point Fractal network (PF-Net) is a seminal work with capability of completing the missing regions of point clouds. However, the multi-resolution structure of PF-Net neglects effective feature fusion between each resolution, which makes the resulting shape lack local geometric details. To tackle this problem, we design a novel shape completion network named PRSCN. We first present Point Rank Sampling to rate and sample feature points more objectively through local outline form. In this way, the sampled points can facilitate the downstream tasks. Subsequently, considering the correlation between features from different scales, we design a Cross-Cascade Module to combine features hierarchically. Moreover, we propose Leap-type EdgeConv to enlarge the receptive field while maintain the kernel size unchanged. These improvements together make our CD error 3% lower than that of state-of-the-art method on ShapeNet-part dataset. CO 2021 Published by Elsevier B.V.

关键词： 3D point cloud Shape completion Deep learning graph convolution neural network Down-sampling algorithm

来源：评论

学校读者我要写书评

暂无评论

Prediction method of pharmacokinetic parameters of small molecule drugs based on GCN network model

引用

JOURNAL OF MOLECULAR MODELING 2024年第8期30卷 264-264页

作者： Yang, Zhihua Wang, Ying Du, Getao Zhan, Yonghua Zhan, Wenhua Ningxia Med Univ Gen Hosp Dept Radiat Oncol Yinchuan 750004 Peoples R China Xidian Univ Minist Educ Engn Res Ctr Mol & Neuro Imaging Sch Life Sci & Technol Xian 710071 Shaanxi Peoples R China

ContextAccurately predicting plasma protein binding rate (PPBR) and oral bioavailability (OBA) helps to better reveal the absorption and distribution of drugs in the human body and subsequent drug design. Although machine learning models have achieved good results in prediction accuracy, they often suffer from insufficient accuracy when dealing with data with irregular topological *** view of this, this study proposes a pharmacokinetic parameter prediction framework based on graph convolutional networks (GCN), which predicts the PPBR and OBA of small molecule drugs. In the framework, GCN is first used to extract spatial feature information on the topological structure of drug molecules, in order to better learn node features and association information between nodes. Then, based on the principle of drug similarity, this study calculates the similarity between small molecule drugs, selects different thresholds to construct datasets, and establishes a prediction model centered on the GCN algorithm. The experimental results show that compared with traditional machine learning prediction models, the prediction model constructed based on the GCN method performs best on PPBR and OBA datasets with an inter-molecular similarity threshold of 0.25, with MAE of 0.155 and 0.167, respectively. In addition, in order to further improve the accuracy of the prediction model, GCN is combined with other algorithms. Compared to using a single GCN method, the distribution of the predicted values obtained by the combined model is highly consistent with the true values. In summary, this work provides a new method for improving the rate of early drug screening in the future.

关键词： Pharmacokinetic parameters graph convolution neural network Molecular similarity Drug screening

来源：评论

学校读者我要写书评

暂无评论

Research on early diagnosis of Alzheimer's disease based on dual fusion cluster graph convolutional network

引用

BIOMEDICAL SIGNAL PROCESSING AND CONTROL 2023年第PartA期86卷

作者： Meng, Lu Zhang, Qianqian Northeastern Univ Coll Informat Sci & Engn Shenyang Peoples R China

Objective: Mild Cognitive Impairment (MCI) is an early stage of Alzheimer's Disease (AD), often mistaken for natural aging. Early detection and treatment of MCI are crucial for effective treatment, but the condition can be difficult to diagnose. In recent years, multi-modal data and deep learning methods have shown promise in this field. The objective of this study is to develop a computer-aided MCI diagnosis system that effectively processes multi-modal data using deep learning ***: We proposed a Dual Fusion Cluster graph convolution network (DFCGCN) model, which combines two channels of feature extraction, one adjacency matrix, and the Cluster GCN in series. Brain imaging is downsampled using graph pooling and flattened into sparse vectors, from which advanced features are extracted. Similarity between connectivity matrices is calculated using the Gaussian kernel function and combined with non-imaging details to construct a population graph that better represents inter-subject variability. Finally, features are assigned to subjects in the population graph, and node embeddings are learned using Cluster GCN to output diagnostic ***: We tested the proposed algorithm on the public Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset, achieving an accuracy, sensitivity, and specificity of 90.7%, 91.1%, and 94.0%, ***: The DFCGCN model presented in this study enhances the diagnosis of MCI and outperforms other state-of-the-art algorithms. This approach has potential to be a valuable tool for early detection and treatment of MCI.

关键词： graph convolution neural network Multi-modal Feature extraction Adjacency matrix ADNI dataset

来源：评论

学校读者我要写书评

暂无评论

Fusing the spatial structure of electroencephalogram channels can increase the individualization of the functional connectivity network

引用

FRONTIERS IN COMPUTATIONAL NEUROSCIENCE 2023年 17卷 1263710页

作者： Li, Ming Yang, Jun Tian, Wenli Ju, Xiangyu Natl Univ Def Technol Coll Intelligence Sci & Technol Changsha Peoples R China

An electroencephalogram (EEG) functional connectivity (FC) network is individualized and plays a significant role in EEG-based person identification. Traditional FC networks are constructed by statistical dependence and correlation between EEG channels, without considering the spatial relationships between the channels. The individual identification algorithm based on traditional FC networks is sensitive to the integrity of channels and crucially relies on signal preprocessing;therefore, finding a new presentation for FC networks may help increase the performance of the identification algorithms. EEG signals are smooth across space owing to the volume conduction effect. Considering such spatial relationships among channels can provide a more accurate representation of FC networks. In this study, we propose an EEG FC network with virtual nodes that combines the spatial relationships and functional connectivity of channels. The comparison results for individual identification show that the novel EEG network is more individualized and achieves an accuracy of 98.64% for data without preprocessing. Furthermore, our algorithm is more robust in reducing the number of channels and can perform well even when a large area of channels is removed.

关键词： virtual node EEG individual difference graph convolution neural network functional connectivity

来源：评论

学校读者我要写书评

暂无评论

A long-term water quality prediction model for marine ranch based on time-graph convolutional neural network

引用

ECOLOGICAL INDICATORS 2023年第1期154卷

作者： Li, Dashe Zhao, Weijie Hu, Jingzhe Zhao, Siwei Liu, Shue Shandong Technol & Business Univ Sch Comp Sci & Technol Yantai 264005 Shandong Peoples R China Shandong Technol & Business Univ Key Lab Intelligent Informat Proc Yantai 264005 Shandong Peoples R China Shandong Technol & Business Univ Univ Future Intelligent Comp Shandong Coll Coinnovat Ctr Yantai 264005 Shandong Peoples R China Binzhou Med Univ Yantai 264003 Shandong Peoples R China

Water quality prediction is important for maintaining water stability and control in marine environments. However, water quality parameters are affected by complex environmental *** time series data exhibit distribution shift and nonstationarity problems, owing to environmental *** is still difficult to obtain the spatial and time dependence of time series data with existing models and the model prediction accuracy cannot be guaranteed. Therefore, this paper proposes a graph-based convolutional neural network model, allowing networks and sequences to interact smoothly. The Time-graph convolutional fusion network(TGCFN) consists of a graph convolutional fusion network(GCFN) and Time-pyramidal fusion attention(T-PFA). To obtain the spatial dependence of the original sequence, the original sequence is first decomposed into three subsequences by STL, and then the GCFN obtains the original sequence and trinomial subsequence through a multilayer *** topology information is also obtained. Second, to eliminate the distribution differences between the training and test sets, the T-PFA is weakened using *** address the distribution shift problem of the sequence and the nonstationarity problem of the destationarized attention attenuation sequence, a higher weight is assigned to the sequence. Finally, depending on the long and short time series of the obtained sequence, the subsequence is downsampled by cross-learning, and the local features are extracted. Experiments on dissolved oxygen, salinity and temperature at six marine ranches on the Shandong peninsula were carried out using the T-GCFN, and the results were compared with those of other deep learning models. The experimental results show that the T-GCFN has better prediction performance and can achieve high-precision predictions of ocean chemistry parameters in the next three days.

关键词： graph convolution neural network Reversible normalization Time series Sequence decomposition Marine ranch

来源：评论

学校读者我要写书评

暂无评论

Orthognathic surgical planning using graph CNN with dual embedding module: External validations with multi-hospital datasets

引用

COMPUTER METHODS AND PROGRAMS IN BIOMEDICINE 2023年 242卷 107853-107853页

作者： Kim, In-Hwan Kim, Jun-Sik Jeong, Jiheon Park, Jae-Woo Park, Kanggil Cho, Jin-Hyoung Hong, Mihee Kang, Kyung-Hwa Kim, Minji Kim, Su-Jung Kim, Yoon-Ji Sung, Sang-Jin Kim, Young Ho Lim, Sung-Hoon Baek, Seung-Hak Kim, Namkug Univ Ulsan Asan Med Inst Convergence Sci & Technol Asan Med Ctr Dept Biomed EngnColl Med Seoul South Korea Univ Ulsan Coll Med Asan Med Ctr Dept Convergence Med 88Olymp Ro 43 Gil Seoul 05505 South Korea Chonnam Natl Univ Sch Dent Dept Orthodont Gwangju South Korea Kyungpook Natl Univ Sch Dent Dept Orthodont Daegu South Korea Wonkwang Univ Sch Dent Dept Orthodont Iksan Si South Korea Ewha Womans Univ Coll Med Dept Orthodont Seoul South Korea Kyung Hee Univ Sch Dent Dept Orthodont Seoul South Korea Univ Ulsan Coll Med Asan Med Ctr Dept Orthodont Seoul South Korea Ajou Univ Sch Med Inst Oral Hlth Sci Dept Orthodont Suwon Gyeonggi Do South Korea Chosun Univ Coll Dent Dept Orthodont Gwangju South Korea Seoul Natl Univ Dent Res Inst Sch Dent Dept Orthodont Daehak Ro 101 Seoul 03080 South Korea Univ Ulsan Coll Med Asan Med Ctr Dept Radiol Seoul South Korea

Background and objective: Despite recent development of AI, prediction of the surgical movement in the maxilla and mandible by OGS might be more difficult than that of tooth movement by orthodontic treatment. To evaluate the prediction accuracy of the surgical movement using pairs of pre-(T0) and post-surgical (T1) lateral cephalograms (latceph) of orthognathic surgery (OGS) patients and dual embedding module-graph convolution neural network (DEM-GCNN) ***: 599 pairs from 3 institutions were used as training, internal validation, and internal test sets and 201 pairs from other 6 institutions were used as external test set. DEM-GCNN model (IEM, learning the lat-ceph images;LTEM, learning the landmarks) was developed to predict the amount and direction of surgical movement of ANS and PNS in the maxilla and B-point and Md1crown in the mandible. The distance between T1 landmark coordinates actually moved by OGS (ground truth) and predicted by DEM-GCNN model and preexisted CNN-based Model-C (learning the lat-ceph images) was ***: In both internal and external tests, DEM-GCNN did not exhibit significant difference from ground truth in all landmarks (ANS, PNS, B-point, Md1crown, all P > 0.05). When the accumulated successful detection rate for each landmark was compared, DEM-GCNN showed higher values than Model-C in both the internal and external tests. In violin plots exhibiting the error distribution of the prediction results, both internal and external tests showed that DEM-GCNN had significant performance improvement in PNS, ANS, B-point, Md1crown than Model-C. DEM-GCNN showed significantly lower prediction error values than Model-C (one-jaw surgery, B-point, Md1crown, all P < 0.005;two-jaw surgery, PNS, ANS, all P < 0.05;B point, Md1crown, all P < 0.005).Conclusion: We developed a robust OGS planning model with maximized generalizability despite diverse qualities of lat-cephs from 9 institutions.

关键词： Cephalometry Deep learning Dual embedding module graph convolution neural network Maxillofacial surgery Multicenter study Orthognathic surgery Surgical prediction

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：