Filters








25,258 Hits in 5.4 sec

Joint Graph Learning and Matching for Semantic Feature Correspondence [article]

He Liu, Tao Wang, Yidong Li, Congyan Lang, Yi Jin, Haibin Ling
2021 arXiv   pre-print
In this paper, we propose a joint graph learning and matching network, named GLAM, to explore reliable graph structures for boosting graph matching.  ...  In recent years, powered by the learned discriminative representation via graph neural network (GNN) models, deep graph matching methods have made great progresses in the task of matching semantic features  ...  ACKNOWLEDGMENTS This work is supported by the National Nature Science Foundation of China (Nos. 62076021 and 61872032) and the Beijing Municipal Natural Science Foundation (Nos.4202060 and 4212041).  ... 
arXiv:2109.00240v2 fatcat:nnwiifijwfcopdyocm25wgi5gm

Learning Dual Semantic Relations with Graph Attention for Image-Text Matching

Keyu Wen, Xiaodong Gu, Qingrong Cheng
2020 IEEE transactions on circuits and systems for video technology (Print)  
and the joint semantic relations module.  ...  However, the lack of joint learning of regional features and global features will cause the regional features to lose contact with the global context, leading to the mismatch with those non-object words  ...  the later joint relations learning. 2) GATs in joint semantic relations module: In Section III-D, we learn the joint semantic relations by dint of GATs.  ... 
doi:10.1109/tcsvt.2020.3030656 fatcat:ymindb2imnbgnlmitnkziskkmi

High-Order Information Matters: Learning Relation and Topology for Occluded Person Re-Identification [article]

Guan'an Wang, Shuo Yang, Huanyu Liu, Zhicheng Wang, Yang Yang, Shuliang Wang, Gang Yu, Erjin Zhou, Jian Sun
2020 arXiv   pre-print
In this paper, we propose a novel framework by learning high-order relation and topology information for discriminative features and robust alignment.  ...  When aligning two groups of local features from two images, we view it as a graph matching problem and propose a cross-graph embedded-alignment (CGEA) layer to jointly learn and embed topology information  ...  It takes two graphs as inputs, learns correspondence of nodes across the two graphs using graph-matching strategy, and passes messages by viewing the learned correspondence as an adjacency matrix.  ... 
arXiv:2003.08177v4 fatcat:qebat4cnw5fr3auusgom3ibany

High-Order Information Matters: Learning Relation and Topology for Occluded Person Re-Identification

Guan'an Wang, Shuo Yang, Huanyu Liu, Zhicheng Wang, Yang Yang, Shuliang Wang, Gang Yu, Erjin Zhou, Jian Sun
2020 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)  
In this paper, we propose a novel framework by learning high-order relation and topology information for discriminative features and robust alignment.  ...  When aligning two groups of local features from two images, we view it as a graph matching problem and propose a cross-graph embeddedalignment (CGEA) layer to jointly learn and embed topology information  ...  Then we can extract semantic features of corresponding key-points. (2) In the R, we view the learned semantic features of an image as nodes of a graph and propose an adaptive-direction graph convolutional  ... 
doi:10.1109/cvpr42600.2020.00648 dblp:conf/cvpr/WangYLWYWYZS20 fatcat:gpvuahigazar3er5wo3wxhpg74

Fine-Grained Video-Text Retrieval With Hierarchical Graph Reasoning

Shizhe Chen, Yida Zhao, Qin Jin, Qi Wu
2020 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)  
Different levels of texts can guide the learning of diverse and hierarchical video representations for cross-modal matching to capture both global and local details.  ...  The current dominant approach is to learn a joint embedding space to measure cross-modal similarities.  ...  These features will be sent to the following matching module to match with their corresponding textual features at different levels, which guarantees different transformation weights can be learned to  ... 
doi:10.1109/cvpr42600.2020.01065 dblp:conf/cvpr/ChenZJW20 fatcat:brlrtsp7lre7bne7cm37esr5wu

A Knowledge-based Deep Heterogeneous Graph Matching Model for Multimodal RecipeQA

Yunjie Wu, Sai Zhang, Xiaowang Zhang, Zhiyong Feng, Liang Wan
2021 International Semantic Web Conference  
Secondly, we design a two-stage heterogeneous graph matching method to guarantee neighborhoods consensus.  ...  In this paper, we propose a Knowledge-based Deep Heterogeneous Graph Matching Model (DHGM) to model temporal structures of recipes.  ...  Acknowledgments The work described in this paper is partially supported by Shenzhen Science and Technology Foundation (JCYJ20170816093943197) and Key Research and Development Program of Hubei Province  ... 
dblp:conf/semweb/WuZZ0W21 fatcat:cupjgke4bvf6le3gr7zrcpd2jq

Using Text to Teach Image Retrieval [article]

Haoyu Dong, Ze Wang, Qiang Qiu, Guillermo Sapiro
2020 arXiv   pre-print
Building on the concept of image manifold, we first propose to represent the feature space of images, learned via neural networks, as a graph.  ...  When limited images are available, this manifold is sparsely sampled, making the geodesic computation and the corresponding retrieval harder.  ...  One is to match the corresponding images given sentence queries, or vice versa; this has been used as an evaluation metric in many works that learn visual-semantic embeddings [8, 35] .  ... 
arXiv:2011.09928v1 fatcat:laivg53j7nhnxmknlnmcq25swm

Joint Learning with BERT-GCN and Multi-Attention for Event Text Classification and Event Assignment

Xiangrong She, Jianpeng Chen, Gang Chen
2022 IEEE Access  
To address these problems, we propose a joint learning method for event text classification and event assignment for Chinese government hotline.  ...  Firstly, graph convolution network (GCN) and BERT are used to process the event text respectively to obtain the corresponding representation vector.  ...  Based on above, we proposed a joint learning method for event text classification and event assignment.  ... 
doi:10.1109/access.2022.3156918 fatcat:chp3j5f4c5afta42q7kp6bu7gm

Similarity Reasoning and Filtration for Image-Text Matching [article]

Haiwen Diao, Ying Zhang, Lin Ma, Huchuan Lu
2021 arXiv   pre-print
In this paper, we propose a novel Similarity Graph Reasoning and Attention Filtration (SGRAF) network for image-text matching.  ...  Specifically, the vector-based similarity representations are firstly learned to characterize the local and global alignments in a more comprehensive manner, and then the Similarity Graph Reasoning (SGR  ...  Similarity Prediction Most existing works ) for image-text matching learned the joint embedding and the similarity measures for cross-modal matching.  ... 
arXiv:2101.01368v1 fatcat:wl5nnrea4rfyzcie55ju7jgq44

Unsupervised Open Domain Recognition by Semantic Discrepancy Minimization

Junbao Zhuo, Shuhui Wang, Shuhao Cui, Qingming Huang
2019 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)  
UODTN better preserves the semantic structure and enforces the consistency between the learned domain invariant visual features and the semantic embeddings.  ...  To measure the domain discrepancy for asymmetric label space between S and T , we propose Semantic-Guided Matching Discrepancy (SGMD), which first employs instance matching between S and T , and then the  ...  Compared to multistage learning paradigms [38, 14] that perform GCN-based classification model propagation and visual feature learning step-by-step, the joint classification network and GCN learning  ... 
doi:10.1109/cvpr.2019.00084 dblp:conf/cvpr/ZhuoWCH19 fatcat:6xfpdkhp3rgfnpclnjkcsqx24e

Unsupervised Open Domain Recognition by Semantic Discrepancy Minimization [article]

Junbao Zhuo, Shuhui Wang, Shuhao Cui, Qingming Huang
2019 arXiv   pre-print
UODTN better preserves the semantic structure and enforces the consistency between the learned domain invariant visual features and the semantic embeddings.  ...  To measure the domain discrepancy for asymmetric label space between S and T, we propose Semantic-Guided Matching Discrepancy (SGMD), which first employs instance matching between S and T, and then the  ...  Compared to multistage learning paradigms [38, 14] that perform GCN-based classification model propagation and visual feature learning step-by-step, the joint classification network and GCN learning  ... 
arXiv:1904.08631v1 fatcat:uqurk2ceevg4dhvg2kdvcibrly

Unsupervised Vision-Language Parsing: Seamlessly Bridging Visual Scene Graphs with Language Structures via Dependency Relationships [article]

Chao Lou, Wenjuan Han, Yuhuan Lin, Zilong Zheng
2022 arXiv   pre-print
Moreover, we benchmark our dataset by proposing a contrastive learning (CL)-based framework VLGAE, short for Vision-Language Graph Autoencoder.  ...  Previous works have shown compelling comprehensive results by building hierarchical structures for visual scenes (e.g., scene graphs) and natural languages (e.g., dependency trees), individually.  ...  Refer to Section 5.2 for the learning process. Cross-modality Matching We employ cross-modality matching to align vision and language features in different levels.  ... 
arXiv:2203.14260v3 fatcat:65xsxe3cdzefdf3uadnj3d27dm

SVGraph: Learning Semantic Graphs from Instructional Videos [article]

Madeline C. Schiappa, Yogesh S. Rawat
2022 arXiv   pre-print
We attempt to overcome "black box" learning limitations by presenting Semantic Video Graph or SVGraph, a multi-modal approach that utilizes narrations for semantic interpretability of the learned graphs  ...  We perform experiments on multiple datasets and demonstrate the interpretability of SVGraph in semantic graph learning.  ...  Semantic Attention In order to make our graph interpretable without annotations, we learn semantically relevant features for our nodes.  ... 
arXiv:2207.08001v1 fatcat:546obyqea5d4zbc7xrdygskfde

Hypergraph Attention Networks for Multimodal Learning

Eun-Sol Kim, Woo Young Kang, Kyoung-Woon On, Yu-Jung Heo, Byoung-Tak Zhang
2020 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)  
between the graphs in the semantic space, and integrating the multimodal inputs using the co-attention maps to get the final joint representation.  ...  HANs follow the process: constructing the common semantic space with symbolic graphs of each modality, matching the semantics between sub-structures of the symbolic graphs, constructing co-attention maps  ...  [21] proposed Graph Matching Networks (GMNs) to learn the similarity between two graphs.  ... 
doi:10.1109/cvpr42600.2020.01459 dblp:conf/cvpr/KimKOHZ20 fatcat:mrul5ses6jbe5csufbqlx3xvda

Question Answering over Linked Data Using First-order Logic

Shizhu He, Kang Liu, Yuanzhe Zhang, Liheng Xu, Jun Zhao
2014 Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP)  
All clauses can then produce interacted effects in a unified framework and can jointly resolve all ambiguities. Moreover, our method adopts a pattern-learning strategy for semantic item grouping.  ...  We formulate the knowledge for resolving the ambiguities in the main three steps of QALD (phrase detection, phrase-tosemantic-item mapping and semantic item grouping) as first-order logic clauses in a  ...  Acknowledgments The authors are grateful to the anonymous reviewers for their constructive comments.  ... 
doi:10.3115/v1/d14-1116 dblp:conf/emnlp/HeLZX014 fatcat:ku5k5ks5dfc5pjxh2zfbbjfzry
« Previous Showing results 1 — 15 out of 25,258 results