Line graph contrastive learning for node classification
Existing graph contrastive learning methods often rely on differences in node features within subgraphs, lacking effective capture of the global structural information of the graph. To address this issue, we propose a novel graph contrastive learning method, Line Graph Contrastive Learning (LineGCL)...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2024-04-01
|
| Series: | Journal of King Saud University: Computer and Information Sciences |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S1319157824001009 |
| _version_ | 1797234016379731968 |
|---|---|
| author | Mingyuan Li Lei Meng Zhonglin Ye Yuzhi Xiao Shujuan Cao Haixing Zhao |
| author_facet | Mingyuan Li Lei Meng Zhonglin Ye Yuzhi Xiao Shujuan Cao Haixing Zhao |
| author_sort | Mingyuan Li |
| collection | DOAJ |
| description | Existing graph contrastive learning methods often rely on differences in node features within subgraphs, lacking effective capture of the global structural information of the graph. To address this issue, we propose a novel graph contrastive learning method, Line Graph Contrastive Learning (LineGCL), aimed at overcoming the deficiencies in understanding the overall characteristics and topological structures of graphs found in current methods. LineGCL utilizes the characteristics of line graphs to transform the original graph into corresponding line graphs, presenting edge information in the form of node features, effectively capturing the global structural information of the graph. Additionally, LineGCL adopts a novel multi-view contrastive learning approach, characterizing the similarity and differences between the original graph and line graph comprehensively from the perspectives of node features and spectral features, further enhancing the model's understanding and learning capabilities of the global structure of graphs. Experimental results on public datasets demonstrate that LineGCL outperforms baseline models, achieving performance improvements ranging from 0.5% to 28.5%. These results validate the effectiveness and superiority of LineGCL in capturing and understanding global structural information, surpassing the limitations of baseline methods in graph contrastive learning. |
| first_indexed | 2024-04-24T16:25:21Z |
| format | Article |
| id | doaj.art-100049d7ce83444baac040866c2318e8 |
| institution | Directory Open Access Journal |
| issn | 1319-1578 |
| language | English |
| last_indexed | 2024-04-24T16:25:21Z |
| publishDate | 2024-04-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Journal of King Saud University: Computer and Information Sciences |
| spelling | doaj.art-100049d7ce83444baac040866c2318e82024-03-31T04:37:11ZengElsevierJournal of King Saud University: Computer and Information Sciences1319-15782024-04-01364102011Line graph contrastive learning for node classificationMingyuan Li0Lei Meng1Zhonglin Ye2Yuzhi Xiao3Shujuan Cao4Haixing Zhao5College of Computer, Qinghai Normal University, Xining, Qinghai 810001, China; The State Key Laboratory of Tibetan lntelligent Information Processing and Application, Xining, Qinghai 810001, ChinaCollege of Computer, Qinghai Normal University, Xining, Qinghai 810001, China; The State Key Laboratory of Tibetan lntelligent Information Processing and Application, Xining, Qinghai 810001, ChinaCollege of Computer, Qinghai Normal University, Xining, Qinghai 810001, China; The State Key Laboratory of Tibetan lntelligent Information Processing and Application, Xining, Qinghai 810001, ChinaCollege of Computer, Qinghai Normal University, Xining, Qinghai 810001, China; The State Key Laboratory of Tibetan lntelligent Information Processing and Application, Xining, Qinghai 810001, ChinaCollege of Computer, Qinghai Normal University, Xining, Qinghai 810001, China; The State Key Laboratory of Tibetan lntelligent Information Processing and Application, Xining, Qinghai 810001, ChinaCorresponding author at: College of Computer, Qinghai Normal University, Xining, Qinghai 810001, China.; College of Computer, Qinghai Normal University, Xining, Qinghai 810001, China; The State Key Laboratory of Tibetan lntelligent Information Processing and Application, Xining, Qinghai 810001, ChinaExisting graph contrastive learning methods often rely on differences in node features within subgraphs, lacking effective capture of the global structural information of the graph. To address this issue, we propose a novel graph contrastive learning method, Line Graph Contrastive Learning (LineGCL), aimed at overcoming the deficiencies in understanding the overall characteristics and topological structures of graphs found in current methods. LineGCL utilizes the characteristics of line graphs to transform the original graph into corresponding line graphs, presenting edge information in the form of node features, effectively capturing the global structural information of the graph. Additionally, LineGCL adopts a novel multi-view contrastive learning approach, characterizing the similarity and differences between the original graph and line graph comprehensively from the perspectives of node features and spectral features, further enhancing the model's understanding and learning capabilities of the global structure of graphs. Experimental results on public datasets demonstrate that LineGCL outperforms baseline models, achieving performance improvements ranging from 0.5% to 28.5%. These results validate the effectiveness and superiority of LineGCL in capturing and understanding global structural information, surpassing the limitations of baseline methods in graph contrastive learning.http://www.sciencedirect.com/science/article/pii/S1319157824001009Graph contrastive learningLine graphSpectral featuresMulti-angle contrast loss |
| spellingShingle | Mingyuan Li Lei Meng Zhonglin Ye Yuzhi Xiao Shujuan Cao Haixing Zhao Line graph contrastive learning for node classification Journal of King Saud University: Computer and Information Sciences Graph contrastive learning Line graph Spectral features Multi-angle contrast loss |
| title | Line graph contrastive learning for node classification |
| title_full | Line graph contrastive learning for node classification |
| title_fullStr | Line graph contrastive learning for node classification |
| title_full_unstemmed | Line graph contrastive learning for node classification |
| title_short | Line graph contrastive learning for node classification |
| title_sort | line graph contrastive learning for node classification |
| topic | Graph contrastive learning Line graph Spectral features Multi-angle contrast loss |
| url | http://www.sciencedirect.com/science/article/pii/S1319157824001009 |
| work_keys_str_mv | AT mingyuanli linegraphcontrastivelearningfornodeclassification AT leimeng linegraphcontrastivelearningfornodeclassification AT zhonglinye linegraphcontrastivelearningfornodeclassification AT yuzhixiao linegraphcontrastivelearningfornodeclassification AT shujuancao linegraphcontrastivelearningfornodeclassification AT haixingzhao linegraphcontrastivelearningfornodeclassification |