FedGCN: Federated Learning-Based Graph Convolutional Networks for Non-Euclidean Spatial Data
Federated Learning (FL) can combine multiple clients for training and keep client data local, which is a good way to protect data privacy. There are many excellent FL algorithms. However, most of these can only process data with regular structures, such as images and videos. They cannot process non-...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2022-03-01
|
| Series: | Mathematics |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2227-7390/10/6/1000 |
| _version_ | 1827628071836975104 |
|---|---|
| author | Kai Hu Jiasheng Wu Yaogen Li Meixia Lu Liguo Weng Min Xia |
| author_facet | Kai Hu Jiasheng Wu Yaogen Li Meixia Lu Liguo Weng Min Xia |
| author_sort | Kai Hu |
| collection | DOAJ |
| description | Federated Learning (FL) can combine multiple clients for training and keep client data local, which is a good way to protect data privacy. There are many excellent FL algorithms. However, most of these can only process data with regular structures, such as images and videos. They cannot process non-Euclidean spatial data, that is, irregular data. To address this problem, we propose a Federated Learning-Based Graph Convolutional Network (FedGCN). First, we propose a Graph Convolutional Network (GCN) as a local model of FL. Based on the classical graph convolutional neural network, TopK pooling layers and full connection layers are added to this model to improve the feature extraction ability. Furthermore, to prevent pooling layers from losing information, cross-layer fusion is used in the GCN, giving FL an excellent ability to process non-Euclidean spatial data. Second, in this paper, a federated aggregation algorithm based on an online adjustable attention mechanism is proposed. The trainable parameter <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mi>ρ</mi></semantics></math></inline-formula> is introduced into the attention mechanism. The aggregation method assigns the corresponding attention coefficient to each local model, which reduces the damage caused by the inefficient local model parameters to the global model and improves the fault tolerance and accuracy of the FL algorithm. Finally, we conduct experiments on six non-Euclidean spatial datasets to verify that the proposed algorithm not only has good accuracy but also has a certain degree of generality. The proposed algorithm can also perform well in different graph neural networks. |
| first_indexed | 2024-03-09T13:25:38Z |
| format | Article |
| id | doaj.art-c5c5bec22abc40b8a3d7701df6c9f067 |
| institution | Directory Open Access Journal |
| issn | 2227-7390 |
| language | English |
| last_indexed | 2024-03-09T13:25:38Z |
| publishDate | 2022-03-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Mathematics |
| spelling | doaj.art-c5c5bec22abc40b8a3d7701df6c9f0672023-11-30T21:25:16ZengMDPI AGMathematics2227-73902022-03-01106100010.3390/math10061000FedGCN: Federated Learning-Based Graph Convolutional Networks for Non-Euclidean Spatial DataKai Hu0Jiasheng Wu1Yaogen Li2Meixia Lu3Liguo Weng4Min Xia5School of Automation, Nanjing University of Information Science and Technology, Nanjing 210044, ChinaSchool of Automation, Nanjing University of Information Science and Technology, Nanjing 210044, ChinaSchool of Automation, Nanjing University of Information Science and Technology, Nanjing 210044, ChinaSchool of Automation, Nanjing University of Information Science and Technology, Nanjing 210044, ChinaSchool of Automation, Nanjing University of Information Science and Technology, Nanjing 210044, ChinaSchool of Automation, Nanjing University of Information Science and Technology, Nanjing 210044, ChinaFederated Learning (FL) can combine multiple clients for training and keep client data local, which is a good way to protect data privacy. There are many excellent FL algorithms. However, most of these can only process data with regular structures, such as images and videos. They cannot process non-Euclidean spatial data, that is, irregular data. To address this problem, we propose a Federated Learning-Based Graph Convolutional Network (FedGCN). First, we propose a Graph Convolutional Network (GCN) as a local model of FL. Based on the classical graph convolutional neural network, TopK pooling layers and full connection layers are added to this model to improve the feature extraction ability. Furthermore, to prevent pooling layers from losing information, cross-layer fusion is used in the GCN, giving FL an excellent ability to process non-Euclidean spatial data. Second, in this paper, a federated aggregation algorithm based on an online adjustable attention mechanism is proposed. The trainable parameter <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mi>ρ</mi></semantics></math></inline-formula> is introduced into the attention mechanism. The aggregation method assigns the corresponding attention coefficient to each local model, which reduces the damage caused by the inefficient local model parameters to the global model and improves the fault tolerance and accuracy of the FL algorithm. Finally, we conduct experiments on six non-Euclidean spatial datasets to verify that the proposed algorithm not only has good accuracy but also has a certain degree of generality. The proposed algorithm can also perform well in different graph neural networks.https://www.mdpi.com/2227-7390/10/6/1000federated learninggraph convolutional neural networknon-Euclidean spatial dataattention mechanism |
| spellingShingle | Kai Hu Jiasheng Wu Yaogen Li Meixia Lu Liguo Weng Min Xia FedGCN: Federated Learning-Based Graph Convolutional Networks for Non-Euclidean Spatial Data Mathematics federated learning graph convolutional neural network non-Euclidean spatial data attention mechanism |
| title | FedGCN: Federated Learning-Based Graph Convolutional Networks for Non-Euclidean Spatial Data |
| title_full | FedGCN: Federated Learning-Based Graph Convolutional Networks for Non-Euclidean Spatial Data |
| title_fullStr | FedGCN: Federated Learning-Based Graph Convolutional Networks for Non-Euclidean Spatial Data |
| title_full_unstemmed | FedGCN: Federated Learning-Based Graph Convolutional Networks for Non-Euclidean Spatial Data |
| title_short | FedGCN: Federated Learning-Based Graph Convolutional Networks for Non-Euclidean Spatial Data |
| title_sort | fedgcn federated learning based graph convolutional networks for non euclidean spatial data |
| topic | federated learning graph convolutional neural network non-Euclidean spatial data attention mechanism |
| url | https://www.mdpi.com/2227-7390/10/6/1000 |
| work_keys_str_mv | AT kaihu fedgcnfederatedlearningbasedgraphconvolutionalnetworksfornoneuclideanspatialdata AT jiashengwu fedgcnfederatedlearningbasedgraphconvolutionalnetworksfornoneuclideanspatialdata AT yaogenli fedgcnfederatedlearningbasedgraphconvolutionalnetworksfornoneuclideanspatialdata AT meixialu fedgcnfederatedlearningbasedgraphconvolutionalnetworksfornoneuclideanspatialdata AT liguoweng fedgcnfederatedlearningbasedgraphconvolutionalnetworksfornoneuclideanspatialdata AT minxia fedgcnfederatedlearningbasedgraphconvolutionalnetworksfornoneuclideanspatialdata |