检索结果-内蒙古大学图书馆

Content and structure based attention for graph node classification

Journal of Intelligent and Fuzzy Systems 2024年第4期46卷 8329-8343页

作者： Chen, Yong Xie, Xiao-Zhu Weng, Wei College of Computer and Information Engineering Xiamen University of Technology Xiamen China Fujian Key Laboratory of Pattern Recognition and Image Understanding Xiamen China

Graph-structured data is ubiquitous in real-world applications, such as social networks, citation networks, and communication networks. Graph neural network (GNN) is the key to process them. In recent years, graph attention networks (GATs) have been proposed for node classification and achieved encouraging performance. It focuses on the content associated on nodes to evaluate the attention weights, and the rich structure information in the graph is almost ignored. Therefore, we propose a multi-head attention mechanism to fully employ node content and graph structure information. The core idea is to introduce the interactions in the topological structure into the existing GATs. This method can more accurately estimate the attention weights among nodes, thereby improving the convergence of GATs. Second, the mechanism is lightweight and efficient, requires no training to learn, can accurately analyze higher-order structural information, and can be strongly interpreted through heatmaps. We name the proposed model content- and structure-based graph attention network (CSGAT). Furthermore, our proposed model achieves state-of-the-art performance on a number of datasets in node classification. The code and data are available at https://***/CroakerShark/CSGAT. © 2024 - IOS Press. All rights reserved.

关键词： Classification (of information)

来源：评论

学校读者我要写书评

暂无评论

Graph convolutional network combined with random walks and graph attention network for node classification

引用

The Journal of China Universities of Posts and Telecommunications 2024年第3期31卷 1-14页

作者： Chen Yong Xie Xiaozhu Weng Wei College of Computer and Information Engineering Xiamen University of TechnologyXiamen 361024China Fujian Key Laboratory of Pattern Recognition and Image Understanding Xiamen University of TechnologyXiamen 361024China

Graph conjoint attention(CAT)network is one of the best graph convolutional networks(GCNs)frameworks,which uses a weighting mechanism to identify important neighbor ***,this weighting mechanism is learned based on static information,which means it is susceptible to noisy nodes and edges,resulting in significant *** this paper,a method is proposed to obtain context dynamically based on random walk,which allows the context-based weighting mechanism to better avoid noise ***,the proposed context-based weighting mechanism is combined with the node content-based weighting mechanism of the graph attention(GAT)network to form a model based on a mixed weighting *** model is named as the context-based and content-based graph convolutional network(CCGCN).CCGCN can better discover important neighbors,eliminate noise edges,and learn node embedding by message *** show that CCGCN achieves state-of-the-art performance on node classification tasks in multiple datasets.

关键词： graph neural network(GNN) graph convolutional network(GCN) semi-supervised classification graph analysis

来源：评论

学校读者我要写书评

暂无评论

GCAT: graph calibration attention transformer for robust object tracking

引用

Neural Computing and Applications 2024年第23期36卷 14151-14172页

作者： Chen, Si Hu, Xinxin Wang, Da-Han Yan, Yan Zhu, Shunzhi Fujian Key Laboratory of Pattern Recognition and Image Understanding School of Computer and Information Engineering Xiamen University of Technology Xiamen361024 China State Key Laboratory of Integrated Services Networks Xidian University Xi’an710071 China Fujian Key Laboratory of Sensing and Computing for Smart City School of Informatics Xiamen University Xiamen361005 China

Recent Siamese trackers have taken advantage of transformers to achieve impressive advancements. However, existing transformer trackers ignore considering the positional and structural information between tokens, and traditional template update strategies easily introduce noises to the dynamic templates during tracking. In order to alleviate this issue, this paper develops a novel end-to-end graph calibration attention transformer network (GCAT) to enhance tracking robustness and accuracy. A graph calibration attention mechanism is first designed to calibrate and aggregate template information, for effectively updating dynamic templates during the tracking process. Specifically, each token is considered as a node in the graph, and then, we calculate the weight relationships between each node and their adjacent nodes. Thus, this mechanism can aggregate the global context information of the template and search nodes and activate feature channels based on weights and biases to obtain more discriminative feature information. Moreover, we leverage a multi-level dropout mechanism to perform the data dropout, the layer dropout, and the feature dropout on the data, network, and attention levels, respectively, to avoid overfitting of local-specific information and improve the generalization ability. Extensive experiments show the proposed method achieves superior performance on seven challenging benchmark datasets, i.e., OTB100, OTB2013, UAV123, LaSOT, GOT10K, VOT2020, and TrackingNet. © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2024.

关键词： Aggregates

来源：评论

学校读者我要写书评

暂无评论

Multi-Zone Transformer Based on Self-Distillation for Facial Attribute recognition 17

Multi-Zone Transformer Based on Self-Distillation for Facial...

引用

17th IEEE International Conference on Automatic Face and Gesture recognition, FG 2023

作者： Chen, Si Zhu, Xueyan Wang, Da-Han Zhu, Shunzhi Wu, Yun Fujian Key Laboratory of Pattern Recognition and Image Understanding School of Computer and Information Engineering Xiamen University of Technology Xiamen361024 China

ISBN: (纸本)9798350345445

Recently, transformers have shown great promising performance in various computer vision tasks. However, the current transformer based methods ignore the information exchanges between transformer blocks, and they have not been applied in the facial attribute recognition task. In this paper, we propose a multi-zone transformer based on self-distillation for FAR, termed MZTS, to predict the facial attributes. A multi-zone transformer encoder is firstly presented to achieve the interactions of the different transformer encoder blocks, thus avoiding forgetting the effective information between the transformer encoder block groups during the iteration process. Furthermore, we introduce a new self-distillation mechanism based on class tokens, which distills the class tokens obtained from the last transformer encoder block group to the other shallow groups by interacting with the significant information between the different transformer blocks through attention. Extensive experiments on the challenging CelebA and LFWA datasets have demonstrated the excellent performance of the proposed method for FAR. © 2023 IEEE.

关键词： Face recognition

来源：评论

学校读者我要写书评

暂无评论

A temporal-spatial feature coordination network for change detection of VHR images 3

A temporal-spatial feature coordination network for change d...

引用

3rd International Conference on Geographic Information and Remote Sensing Technology, GIRST 2024

作者： Zeng, Yuxian Zhang, Xinxin College of Computer and Information Engineering Xiamen University of Technology Xiamen361024 China Fujian Key Laboratory of Pattern Recognition and Image Understanding Xiamen University of Technology Xiamen361024 China

ISBN: (纸本)9781510689077

Deep learning techniques have significantly improved the accuracy and efficiency of change detection of very high resolution (VHR) images. However, many current models ignore the inherent heterogeneity of bi-temporal remote sensing images, thus making it difficult to distinguish between target changes and background variations. To address this problem, we propose a deep learning (DL) framework, namely temporal-spatial feature coordination network (TSFCN), to improve the accuracy of detection by enhancing the differentiation of change pixel features and improving the extraction and processing of change information. Further comparison and ablation experiments on the SVCD dataset show that the proposed model, compared with other models, improves the F1 scores and IoU by 2.09% and 3.81%, respectively. The results demonstrate that our model can achieve better accuracy and performance improvement. © 2025 SPIE.

关键词： Change detection

来源：评论

学校读者我要写书评

暂无评论

Collaborative Weighting for Graph Convolutional Networks

Journal of Network Intelligence

引用

Journal of Network Intelligence 2023年第2期8卷 432-447页

作者： Chen, Yong Xie, Xiao-Zhu Weng, Wei Zhang, Shan-Dan Li, Tong College of Computer and Information Engineering Xiamen University of Technology Xiamen361024 China College of Computer and Information Engineering Xiamen University of Technology Fujian Key Laboratory of Pattern Recognition and Image Understanding Xiamen361024 China

Graph neural network (GNN), as a powerful method for graph representation, has attracted extensive research interest. Recently, Graph Convolutional Network (GCN) and Graph Attention Network (GAT) have shown superior performance on graph node classification and are considered as the most promising frameworks. Both GAT and GCN use a weighting mechanism in the information aggregation process. However, GCN is weighted by topological structure, while the node content is still unemployed. GAT is weighted by node content, while the topological structure remains being ignored. These weighting mechanism, which does not make full use of topology and node content, is not conducive to graph representation. In this paper, we propose a "Collaborative Weighting For Graph Convolutional Network" (CWGCN) that combines the advantages of GCN and GAT. Specifically, its weighting mechanism not only considers the topology of the graph, but also the node content. With this weighting method for representation learning, CWGCN achieves encouraging performance on a number of citation datasets in node classification. © 2023, Taiwan Ubiquitous Information CO LTD. All rights reserved.

关键词： Classification (of information)

来源：评论

学校读者我要写书评

暂无评论

Adjustable Gating Prompt Transformer for Facial Attribute recognition with Limited Labeled Data 27th

Adjustable Gating Prompt Transformer for Facial Attribute R...

引用

27th International Conference on pattern recognition, ICPR 2024

作者： Ye, Qinxian Chen, Si Wang, Da-Han Jiang, Nanfeng Su, Yanfei Yan, Yan Fujian Key Laboratory of Pattern Recognition and Image Understanding School of Computer and Information Engineering Xiamen University of Technology Xiamen361024 China School of Informatics Xiamen University Xiamen361005 China

ISBN: (纸本)9783031781032

Existing supervised facial attribute recognition (FAR) methods that rely on large labeled datasets can pose a challenge in real-world scenarios. In the case of limited labeled data, the current methods that introduce auxiliary tasks with a large number of parameters are not conducive to the embedded applications of FAR. To overcome these challenges, this paper develops an adjustable gating prompt Transformer that can handle the limited labeled FAR task with a small number of training parameters. Specifically, we employ an effective image-guided prompt tuning, where the image-related prompt sequence is first generated by feeding image tokens into an image-guided prompt generation network (IPG-Net). Then, the prompt sequence can learn facial image information and guide the frozen pre-trained Transformer to fine-tune the model. In addition, dynamically adjustable gating is applied to the prompt sequence to adaptively adjust the contribution of the prompts from different encoder layers, which enhances the interaction between the different encoder layers and retains effective feature information during the iterative process. Experimental results on the CelebA and LFWA datasets demonstrate that our method outperforms competitive methods with a very small amount of training parameters when only limited labeled data are used. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

关键词： Signal encoding

来源：评论

学校读者我要写书评

暂无评论

Multi-Scale Channel Attention for Chinese Scene Text recognition 11

Multi-Scale Channel Attention for Chinese Scene Text Recogni...

引用

11th International Conference on Computing and pattern recognition, ICCPR 2022

作者： Liao, Hai-Qing Du, Xia Wu, Yun Wang, Da-Han Fujian Key Laboratory of Pattern Recognition and Image Understanding Xiamen University of Technology Xiamen361024 China

ISBN: (纸本)9781450397056

Scene text recognition have proven to be highly effective in solving various computer vision tasks. Recently, numerous recognition algorithms based on the encoder-decoder framework have been proposed for handling scene texts with perspective distortion and curve shape. Nevertheless, most of these methods only consider single-scale features while not taking multi-scale features into account. Meanwhile, the existing text recognition methods are mainly used for English texts, whereas ignoring Chinese texts' pivotal role. In this paper, we proposed an end-to-end method to integrate multi-scale features for Chinese scene text recognition (CSTR). Specifically, we adopted and customized the Dense Atrous Spatial Pyramid Pooling (DenseASPP) to our backbone network to capture multi-scale features of the input image while simultaneously extending the receptive fields. Moreover, we added Squeeze-and-Excitation Networks (SE) to capture attentional features with global information to improve the performance of CSTR further. The experimental results of the Chinese scene text datasets demonstrate that the proposed method can efficiently mitigate the impacts of the loss of contextual information caused by the text scale varying and outperforms the state-of-the-art approaches. © 2022 ACM.

关键词： Character recognition

来源：评论

学校读者我要写书评

暂无评论

A Lightweight Brain Tumor Segmentation Network Based on 3D Inverted Residual Modules 11

A Lightweight Brain Tumor Segmentation Network Based on 3D I...

引用

11th International Conference on Computing and pattern recognition, ICCPR 2022

作者： Liu, Yuchao Du, Xia Wang, Da-Han Zhu, Shunzhi Fujian Key Laboratory of Pattern Recognition and Image Understanding Xiamen University of Technology Xiamen361024 China

ISBN: (纸本)9781450397056

Semantic segmentation technology based on deep learning has played an important role for doctors in identifying brain tumor regions and formulating treatment plans. Popular automated segmentation methods for brain tumors include 2D and 3D convolution networks. The 3D networks give better results but lead to a significant increase in parameters and computational cost. In this paper, we propose a lightweight brain tumor segmentation network composed of 3D inverted residual modules, which can significantly reduce the computational complexity of 3D models. Based on a lightweight depthwise separable convolution, our 3D inverted residual module extracts high-dimensional brain tumor features through an intermediate expansion layer, thus improving performance. On the brain tumor dataset BraTS 2018, our network achieves dice scores of 80.8%, 90.7%, and 84.3% (for ET, WT, and TC, respectively) with only 0.68M parameters and 51.46G FLOPs. The results show that our method can significantly reduce the complexity of the 3D model and achieve very competitive performance. © 2022 ACM.

关键词： Semantics

来源：评论

学校读者我要写书评

暂无评论

Multi-Scale Features Integration for Handwritten Mathematical Expression recognition 11

Multi-Scale Features Integration for Handwritten Mathematica...

引用

11th International Conference on Computing and pattern recognition, ICCPR 2022

作者： Liu, Xianghao Wang, Da-Han Zhu, Shunzhi Fujian Key Laboratory of Pattern Recognition and Image Understanding Xiamen University of Technology Xiamen361024 China

ISBN: (纸本)9781450397056

Handwritten mathematical expression recognition (HMER) is a challenging task due to the complex two-dimensional structure of mathematical expressions and the similarity of handwritten texts. Most existing methods for HMER only consider single-scale features while ignoring multi-scale features that are very important to HMER. Few works have explored the fusion of multi-scale features in HMER, but exhibited an extra branch that brings more parameters and computation. In this paper, we propose an end-to-end method to integrate multi-scale features using a unified model. Specifically, we customized the Dense Atrous Spatial Pyramid Pooling (DenseASPP) to our backbone network to capture the multi-scale features of the input image meanwhile expanding the receptive fields. Moreover, we added a symbol classifier using focal loss to better discriminate and recognize similar symbols, to further improve the performance of HMER. Experiments on the Competition on recognition of Online Handwritten Mathematical Expressions (CROHME) 2014, 2016 and 2019 shows that the proposed method achieves superior performance to most state-of-the-art methods, demonstrating the effectiveness of the proposed method. © 2022 ACM.

关键词： Character recognition

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

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

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：