Multi-constraints in deep graph convolutional networks with initial residual

Chen, Hui; Li, Yuancheng

doi:10.1007/s10489-022-04222-8

Cited by 11 publications

(18 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…APPNP [31], JK [32], Geom‐GCN [16], SimP‐GCN [45], and CPGNN [18] aim to improve the feature propagation scheme within layers of the model. More recently, researchers are proposing to make GNN models deeper [27, 29, 30]. However, deeper models suffer from over‐smoothing, where after stacking many GNN layers, features of the node become indistinguishable from each other, and there is a drop in the performance of the model.…”

Section: Related Workmentioning

confidence: 99%

“…DropEdge [46] proposes to drop a certain number of edges to reduce the speed of convergence of over‐smoothing and relieves the information loss. GCNII [27] use residual connections and identity mapping in GNN layers to enable deeper networks. RevGNN [29] uses deep reversible architectures and [30] uses noise regularisation to train deep GNN models.…”

Section: Related Workmentioning

confidence: 99%

“…We compare our model to 13 different GNN baselines and use the published results as the best performance of these models. GCNII [27], TDGNN [47], H2GCN [17] and GloGNN [49] have proposed multiple variants of their model. We have chosen the variant with the best performance on most datasets.…”

Section: Gnn Baselinesmentioning

confidence: 99%

“…There are many different ways of aggregation schemes proposed in current GNN literature. Many GNN models use simple indiscriminate aggregation [19–21], some use random sampling of neighbours [22, 23], attention‐based aggregation [24, 25], filter coefficient based aggregation [26–28], etc.…”

Section: Introductionmentioning

confidence: 99%

“…Thus, any feature aggregation by GNN with strict homophily assumption may often cause the addition of noisy features and may reduce the signal present in node features to learn effectively. Researchers have explored many ways to tackle this challenge by proposing many improvements in model design, for example, making models deeper [27, 29, 30], using attention layers [24], residual layers [27, 31, 32], etc. However, most of these solutions attempt to fit the model to the noisy data in an effective way [33, 34].…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Feature selection: Key to enhance node classification with graph neural networks

Maurya

Liu²,

Murata

2023

CAAI Trans on Intel Tech

View full text Add to dashboard Cite

Graphs help to define the relationships between entities in the data. These relationships, represented by edges, often provide additional context information which can be utilised to discover patterns in the data. Graph Neural Networks (GNNs) employ the inductive bias of the graph structure to learn and predict on various tasks. The primary operation of graph neural networks is the feature aggregation step performed over neighbours of the node based on the structure of the graph. In addition to its own features, for each hop, the node gets additional combined features from its neighbours. These aggregated features help define the similarity or dissimilarity of the nodes with respect to the labels and are useful for tasks like node classification. However, in real-world data, features of neighbours at different hops may not correlate with the node's features. Thus, any indiscriminate feature aggregation by GNN might cause the addition of noisy features leading to degradation in model's performance. In this work, we show that selective aggregation of node features from various hops leads to better performance than default aggregation on the node classification task. Furthermore, we propose a Dual-Net GNN architecture with a classifier model and a selector model. The classifier model trains over a subset of input node features to predict node labels while the selector model learns to provide optimal input subset to the classifier for the best performance. These two models are trained jointly to learn the best subset of features that give higher accuracy in node label predictions. With extensive experiments, we show that our proposed model outperforms both feature selection methods and state-of-the-art GNN models with remarkable improvements up to 27.8%.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Section: Gnn Baselinesmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Feature selection: Key to enhance node classification with graph neural networks

Maurya

Liu²,

Murata

2023

CAAI Trans on Intel Tech

View full text Add to dashboard Cite

show abstract

Assessing retinal vein occlusion based on color fundus photographs using neural understanding network (NUN)

et al. 2022

View full text Add to dashboard Cite

Objective To develop and validate a novel deep learning architecture to classify retinal vein occlusion (RVO) on color fundus photographs (CFPs) and reveal the image features contributing to the classification. Methods The neural understanding network (NUN) is formed by two components: (1) convolutional neural network (CNN)‐based feature extraction and (2) graph neural networks (GNN)‐based feature understanding. The CNN‐based image features were transformed into a graph representation to encode and visualize long‐range feature interactions to identify the image regions that significantly contributed to the classification decision. A total of 7062 CFPs were classified into three categories: (1) no vein occlusion (“normal”), (2) central RVO, and (3) branch RVO. The area under the receiver operative characteristic (ROC) curve (AUC) was used as the metric to assess the performance of the trained classification models. Results The AUC, accuracy, sensitivity, and specificity for NUN to classify CFPs as normal, central occlusion, or branch occlusion were 0.975 (± 0.003), 0.911 (± 0.007), 0.983 (± 0.010), and 0.803 (± 0.005), respectively, which outperformed available classical CNN models. Conclusion The NUN architecture can provide a better classification performance and a straightforward visualization of the results compared to CNNs.

show abstract

Intention-aware denoising graph neural network for session-based recommendation

Hua

Gan

2023

Appl Intell

View full text Add to dashboard Cite

Multi-constraints in deep graph convolutional networks with initial residual

Cited by 11 publications

References 27 publications

Feature selection: Key to enhance node classification with graph neural networks

Feature selection: Key to enhance node classification with graph neural networks

Assessing retinal vein occlusion based on color fundus photographs using neural understanding network (NUN)

Intention-aware denoising graph neural network for session-based recommendation

Contact Info

Product

Resources

About