Uniform Pooling for Graph Networks
The graph convolution network has received a lot of attention because it extends the convolution to non-Euclidean domains. However, the graph pooling method is still less concerned, which can learn coarse graph embedding to facilitate graph classification. Previous pooling methods were based on assi...
| 發表在: | Applied Sciences |
|---|---|
| Main Authors: | , , , |
| 格式: | Article |
| 語言: | 英语 |
| 出版: |
MDPI AG
2020-09-01
|
| 主題: | |
| 在線閱讀: | https://www.mdpi.com/2076-3417/10/18/6287 |
| _version_ | 1850408213784035328 |
|---|---|
| author | Jian Qin Li Liu Hui Shen Dewen Hu |
| author_facet | Jian Qin Li Liu Hui Shen Dewen Hu |
| author_sort | Jian Qin |
| collection | DOAJ |
| container_title | Applied Sciences |
| description | The graph convolution network has received a lot of attention because it extends the convolution to non-Euclidean domains. However, the graph pooling method is still less concerned, which can learn coarse graph embedding to facilitate graph classification. Previous pooling methods were based on assigning a score to each node and then pooling only the highest-scoring nodes, which might throw away whole neighbourhoods of nodes and therefore information. Here, we proposed a novel pooling method UGPool with a new point-of-view on selecting nodes. UGPool learns node scores based on node features and uniformly pools neighboring nodes instead of top nodes in the score-space, resulting in a uniformly coarsened graph. In multiple graph classification tasks, including the protein graphs, the biological graphs and the brain connectivity graphs, we demonstrated that UGPool outperforms other graph pooling methods while maintaining high efficiency. Moreover, we also show that UGPool can be integrated with multiple graph convolution networks to effectively improve performance compared to no pooling. |
| format | Article |
| id | doaj-art-0fe0e708564d44c681e52f6f02913a4a |
| institution | Directory of Open Access Journals |
| issn | 2076-3417 |
| language | English |
| publishDate | 2020-09-01 |
| publisher | MDPI AG |
| record_format | Article |
| spelling | doaj-art-0fe0e708564d44c681e52f6f02913a4a2025-08-19T22:47:42ZengMDPI AGApplied Sciences2076-34172020-09-011018628710.3390/app10186287Uniform Pooling for Graph NetworksJian Qin0Li Liu1Hui Shen2Dewen Hu3The College of Intelligence Science and Technology, National University of Defense Technology, Changsha 410073, ChinaThe College of System Engineering, National University of Defense Technology, Changsha 410073, ChinaThe College of Intelligence Science and Technology, National University of Defense Technology, Changsha 410073, ChinaThe College of Intelligence Science and Technology, National University of Defense Technology, Changsha 410073, ChinaThe graph convolution network has received a lot of attention because it extends the convolution to non-Euclidean domains. However, the graph pooling method is still less concerned, which can learn coarse graph embedding to facilitate graph classification. Previous pooling methods were based on assigning a score to each node and then pooling only the highest-scoring nodes, which might throw away whole neighbourhoods of nodes and therefore information. Here, we proposed a novel pooling method UGPool with a new point-of-view on selecting nodes. UGPool learns node scores based on node features and uniformly pools neighboring nodes instead of top nodes in the score-space, resulting in a uniformly coarsened graph. In multiple graph classification tasks, including the protein graphs, the biological graphs and the brain connectivity graphs, we demonstrated that UGPool outperforms other graph pooling methods while maintaining high efficiency. Moreover, we also show that UGPool can be integrated with multiple graph convolution networks to effectively improve performance compared to no pooling.https://www.mdpi.com/2076-3417/10/18/6287graph convolution networkgraph poolinggraph classificationnon-euclidean structured signal |
| spellingShingle | Jian Qin Li Liu Hui Shen Dewen Hu Uniform Pooling for Graph Networks graph convolution network graph pooling graph classification non-euclidean structured signal |
| title | Uniform Pooling for Graph Networks |
| title_full | Uniform Pooling for Graph Networks |
| title_fullStr | Uniform Pooling for Graph Networks |
| title_full_unstemmed | Uniform Pooling for Graph Networks |
| title_short | Uniform Pooling for Graph Networks |
| title_sort | uniform pooling for graph networks |
| topic | graph convolution network graph pooling graph classification non-euclidean structured signal |
| url | https://www.mdpi.com/2076-3417/10/18/6287 |
| work_keys_str_mv | AT jianqin uniformpoolingforgraphnetworks AT liliu uniformpoolingforgraphnetworks AT huishen uniformpoolingforgraphnetworks AT dewenhu uniformpoolingforgraphnetworks |