CLB-Defense: based on contrastive learning defense for graph neural network against backdoor attack
For the problem that the existing backdoor attack defense methods are difficult to deal with irregular and unstructured discrete graph data to alleviate the threat of backdoor attacks, a backdoor attack defense method for GNN based on contrastive learning was proposed, namely CLB-Defense.Specificall...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | zho |
| Published: |
Editorial Department of Journal on Communications
2023-04-01
|
| Series: | Tongxin xuebao |
| Subjects: | |
| Online Access: | http://www.joconline.com.cn/zh/article/doi/10.11959/j.issn.1000-436x.2023074/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1841540008767389696 |
|---|---|
| author | Jinyin CHEN Haiyang XIONG Haonan MA Yayu ZHENG |
| author_facet | Jinyin CHEN Haiyang XIONG Haonan MA Yayu ZHENG |
| author_sort | Jinyin CHEN |
| collection | DOAJ |
| description | For the problem that the existing backdoor attack defense methods are difficult to deal with irregular and unstructured discrete graph data to alleviate the threat of backdoor attacks, a backdoor attack defense method for GNN based on contrastive learning was proposed, namely CLB-Defense.Specifically, a contrastive model was built by using contrastive learning in an unsupervised way, which searches suspicious backdoored samples.Then the suspicious backdoored samples were reshaped by using the graph importance indexes and the label smoothing strategy, and the defense against graph backdoor attack was realized.Finally, extensive experimental results show that CLB-Defense realizes the effect of defense performance on four public datasets and five popular graph backdoor attacks, e.g., CLB-Defense can reduce the attack success rate by an average of 75.66% (compared with the baselines, an improvement of 54.01%). |
| format | Article |
| id | doaj-art-0eec1139c0354ccbbda74af36a43913d |
| institution | Kabale University |
| issn | 1000-436X |
| language | zho |
| publishDate | 2023-04-01 |
| publisher | Editorial Department of Journal on Communications |
| record_format | Article |
| series | Tongxin xuebao |
| spelling | doaj-art-0eec1139c0354ccbbda74af36a43913d2025-01-14T06:28:28ZzhoEditorial Department of Journal on CommunicationsTongxin xuebao1000-436X2023-04-014415416659390382CLB-Defense: based on contrastive learning defense for graph neural network against backdoor attackJinyin CHENHaiyang XIONGHaonan MAYayu ZHENGFor the problem that the existing backdoor attack defense methods are difficult to deal with irregular and unstructured discrete graph data to alleviate the threat of backdoor attacks, a backdoor attack defense method for GNN based on contrastive learning was proposed, namely CLB-Defense.Specifically, a contrastive model was built by using contrastive learning in an unsupervised way, which searches suspicious backdoored samples.Then the suspicious backdoored samples were reshaped by using the graph importance indexes and the label smoothing strategy, and the defense against graph backdoor attack was realized.Finally, extensive experimental results show that CLB-Defense realizes the effect of defense performance on four public datasets and five popular graph backdoor attacks, e.g., CLB-Defense can reduce the attack success rate by an average of 75.66% (compared with the baselines, an improvement of 54.01%).http://www.joconline.com.cn/zh/article/doi/10.11959/j.issn.1000-436x.2023074/graph neural networkbackdoor attackrobustnessdefensecontrastive learning |
| spellingShingle | Jinyin CHEN Haiyang XIONG Haonan MA Yayu ZHENG CLB-Defense: based on contrastive learning defense for graph neural network against backdoor attack Tongxin xuebao graph neural network backdoor attack robustness defense contrastive learning |
| title | CLB-Defense: based on contrastive learning defense for graph neural network against backdoor attack |
| title_full | CLB-Defense: based on contrastive learning defense for graph neural network against backdoor attack |
| title_fullStr | CLB-Defense: based on contrastive learning defense for graph neural network against backdoor attack |
| title_full_unstemmed | CLB-Defense: based on contrastive learning defense for graph neural network against backdoor attack |
| title_short | CLB-Defense: based on contrastive learning defense for graph neural network against backdoor attack |
| title_sort | clb defense based on contrastive learning defense for graph neural network against backdoor attack |
| topic | graph neural network backdoor attack robustness defense contrastive learning |
| url | http://www.joconline.com.cn/zh/article/doi/10.11959/j.issn.1000-436x.2023074/ |
| work_keys_str_mv | AT jinyinchen clbdefensebasedoncontrastivelearningdefenseforgraphneuralnetworkagainstbackdoorattack AT haiyangxiong clbdefensebasedoncontrastivelearningdefenseforgraphneuralnetworkagainstbackdoorattack AT haonanma clbdefensebasedoncontrastivelearningdefenseforgraphneuralnetworkagainstbackdoorattack AT yayuzheng clbdefensebasedoncontrastivelearningdefenseforgraphneuralnetworkagainstbackdoorattack |