Deepened Graph Auto-Encoders Help Stabilize and Enhance Link Prediction

Wu, Xinxing; Cheng, Qiang

doi:10.48550/arxiv.2103.11414

Cited by 2 publications

(6 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Over the last few years, the overall effectiveness of the GAE and VGAE paradigms at addressing link prediction has been widely confirmed experimentally [15,16,17,18,19,20,21,22,24,27,28,51,52,53]. Numerous research efforts proposed and evaluated variants of GAE and VGAE designed for this specific task, improving their performances by considering more refined encoders [17,21,51,54], decoders [18,20,21,22] or regularization techniques [15,19,28]. Other works also successfully addressed different downstream tasks that are closely related to link prediction, such as edge classification [52] or graph-based recommendation [51,53,55].…”

Section: Link Prediction With Gae Vgae and Extensionsmentioning

confidence: 99%

Modularity-Aware Graph Autoencoders for Joint Community Detection and Link Prediction

Salha-Galvan¹,

Lutzeyer²,

Dasoulas³

et al. 2022

Preprint

View full text Add to dashboard Cite

Graph autoencoders (GAE) and variational graph autoencoders (VGAE) emerged as powerful methods for link prediction. Their performances are less impressive on community detection problems where, according to recent and concurring experimental evaluations, they are often outperformed by simpler alternatives such as the Louvain method. It is currently still unclear to which extent one can improve community detection with GAE and VGAE, especially in the absence of node features. It is moreover uncertain whether one could do so while simultaneously preserving good performances on link prediction. In this paper, we show that jointly addressing these two tasks with high accuracy is possible. For this purpose, we introduce and theoretically study a community-preserving message passing scheme, doping our GAE and VGAE encoders by considering both the initial graph structure and modularity-based prior communities when computing embedding spaces. We also propose novel training and optimization strategies, including the introduction of a modularity-inspired regularizer complementing the existing reconstruction losses for joint link prediction and community detection. We demonstrate the empirical effectiveness of our approach, referred to as Modularity-Aware GAE and VGAE, through in-depth experimental validation on various real-world graphs.

show abstract

Section: Link Prediction With Gae Vgae and Extensionsmentioning

confidence: 99%

Modularity-Aware Graph Autoencoders for Joint Community Detection and Link Prediction

Salha-Galvan¹,

Lutzeyer²,

Dasoulas³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Wu et al [37] proposed the IMvGCN model, which uses reconstruction error and Laplace matrix learning tasks to better establish the connection between GCN and multi-view learning from a feature and structural perspective, and to better learn node features. Existing graph auto-encoder models all use only shallow structures [18], and there are limitations of non-Euclidean data on the link prediction task. The DGAE model proposed by Wu et al [18] for this problem can effectively solve the problem.…”

Section: Related Workmentioning

confidence: 99%

“…Existing graph auto-encoder models all use only shallow structures [18], and there are limitations of non-Euclidean data on the link prediction task. The DGAE model proposed by Wu et al [18] for this problem can effectively solve the problem. The graph auto-encoder of the shallow model has poor perceptual field and convergence and will lose some node features causing some limitations in node feature extraction [11].…”

Section: Related Workmentioning

confidence: 99%

“…Graph embedding is a process of converting highdimensional dense graph data into low-dimensional vectors through some mapping [5], [18], [19]. To obtain better graph embeddings, researchers have continuously proposed new models.…”

Section: Introductionmentioning

confidence: 99%

“…GCNs are only applicable to transductive and cannot be applied to inductive learning [9]. At the same time, GCN as an encoder only uses a shallow structure [18], which makes the convergence of the model poor and will lose some node features causing some limitations in node feature extraction [2], [10], [11], [33], [38].…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Random Walk Graph Auto-Encoders With Ensemble Networks in Graph Embedding

et al. 2023

View full text Add to dashboard Cite

Recently graph auto-encoders have received increasingly widespread attention as one of the important models in the field of deep learning. Existing graph auto-encoder models only use graph convolutional neural networks (GCNs) as encoders to learn the embedding representation of nodes. However, GCNs are only suitable for transductive learning, have poor scalability and shallow models with a poor perceptual field, and have limitations in node feature extraction. To alleviate these problems, we propose to use an adaptive weight integration graph attention network (GAT) and GCN's random walk graph autoencoder (EGRWR-GAE) to better learn the embedding representation of nodes. There is a large amount of noise in the graph data, which interferes with feature extraction and the GAT model is sensitive to noisy data, we propose a random walk graph auto-encoder (EGSRWR-GAE) that integrates GAT, GCN, and self-supervised graph attention networks (SuperGAT) using adaptive weights. The effectiveness of our model is well demonstrated by three publicly available datasets (Cora, Citeseer, and Pubmed) with optimizations of up to 2.2% on the link prediction task and up to 12.9% on the node clustering task.

show abstract

Deepened Graph Auto-Encoders Help Stabilize and Enhance Link Prediction

Cited by 2 publications

References 20 publications

Modularity-Aware Graph Autoencoders for Joint Community Detection and Link Prediction

Modularity-Aware Graph Autoencoders for Joint Community Detection and Link Prediction

Random Walk Graph Auto-Encoders With Ensemble Networks in Graph Embedding

Contact Info

Product

Resources

About