# Deep Graph Structure Learning for Robust Representations: A Survey

@article{Zhu2021DeepGS, title={Deep Graph Structure Learning for Robust Representations: A Survey}, author={Yanqiao Zhu and Weizhi Xu and Jinghao Zhang and Qiang Liu and Shu Wu and Liang Wang}, journal={ArXiv}, year={2021}, volume={abs/2103.03036} }

Graph Neural Networks (GNNs) are widely used for analyzing graph-structured data. Most GNN methods are highly sensitive to the quality of graph structures and usually require a perfect graph structure for learning informative embeddings. However, the pervasiveness of noise in graphs necessitates learning robust representations for real-world problems. To improve the robustness of GNN models, many studies have been proposed around the central concept of Graph Structure Learning (GSL), which aims… Expand

#### 11 Citations

Label-informed Graph Structure Learning for Node Classification

- Computer Science
- CIKM
- 2021

This paper proposes a novel label-informed graph structure learning framework which incorporates label information explicitly through a class transition matrix and shows that this method outperforms or matches the state-of-the-art baselines. Expand

An Empirical Study of Graph Contrastive Learning

- Computer Science
- ArXiv
- 2021

This work identifies several critical design considerations within a general GCL paradigm, including augmentation functions, contrasting modes, contrastive objectives, and negative mining techniques, and develops an easy-to-use library PyGCL, featuring modularized CL components, standardized evaluation, and experiment management. Expand

Attention-driven Graph Clustering Network

- Computer Science
- ACM Multimedia
- 2021

This work proposes a novel deep clustering method named Attention-driven Graph Clustering Network (AGCN), which exploits a heterogeneity-wise fusion module to dynamically fuse the node attribute feature and the topological graph feature and develops a scale-wise fuse module to adaptively aggregate the multi-scale features embedded at different layers. Expand

Latent Structures Mining with Contrastive Modality Fusion for Multimedia Recommendation

- Computer Science
- ArXiv
- 2021

In the proposed MICRO model, a novel modality-aware structure learning module is devised, which learns item-item relationships for each modality, and a novel multi-modal contrastive framework is designed to facilitate fine-grained multimodal fusion. Expand

Learnt Sparsification for Interpretable Graph Neural Networks

- Computer Science
- ArXiv
- 2021

This paper proposes a novel method called KEdge for explicitly sparsification using the Hard Kumaraswamy distribution that can be used in conjugation with any GNN model and effectively counters the over-smoothing phenomena in deep GNNs by maintaining good task performance with increasing GNN layers. Expand

Mining Cross Features for Financial Credit Risk Assessment

- Computer Science
- CIKM
- 2021

This work proposes a novel automatic feature crossing method called DNN2LR, which is a LR model empowered with cross features, generated by DNN1LR is a white-box model, and conducts experiments on both public and business datasets from real-world credit risk assessment applications, which show that it outperform both conventional models used for credit assessment and several feature crossing methods. Expand

Mining Latent Structures for Multimedia Recommendation

- Computer Science
- ACM Multimedia
- 2021

A novel modality-aware structure learning layer is devised, which learns item-item structures for each modality and aggregates multiple modalities to obtain latent item graphs and performs graph convolutions to explicitly inject high-order item affinities into item representations. Expand

Property-Aware Relation Networks for Few-Shot Molecular Property Prediction

- Computer Science
- 2021

This paper introduces a property-aware embedding function to transform the generic molecular embeddings to substructure-aware space relevant to the target property, and designs an adaptive relation graph learning module to jointly estimate molecular relation graph and refine molecularembeddings w.r.t. the targetproperty. Expand

Property-aware Adaptive Relation Networks for Molecular Property Prediction

- Computer Science
- ArXiv
- 2021

This paper proposes a property-aware adaptive relation networks (PAR) for the few-shot molecular property prediction problem, which leverages the facts that both substructures and relationships among molecules are different considering various molecular properties. Expand

SLAPS: Self-Supervision Improves Structure Learning for Graph Neural Networks

- Computer Science
- ArXiv
- 2021

The Simultaneous Learning of Adjacency and GNN Parameters with Self-supervision, or SLAPS, a method that provides more supervision for inferring a graph structure through selfsupervision and outperforms several models that have been proposed to learn a task-specific graph structure on established benchmarks. Expand

#### References

SHOWING 1-10 OF 73 REFERENCES

Graph Sparsification via Meta-Learning

- 2020

We present a novel graph sparsification approach for semisupervised learning on undirected attributed graphs. The main challenge is to retain few edges while minimize the loss of node classification… Expand

Iterative Deep Graph Learning for Graph Neural Networks: Better and Robust Node Embeddings

- Computer Science, Mathematics
- NeurIPS
- 2020

This paper proposes an end-to-end graph learning framework, namely Iterative Deep Graph Learning (IDGL), for jointly and iteratively learning graph structure and graph embedding and proposes a scalable version of IDGL, namely IDGL-ANCH, which significantly reduces the time and space complexity of ID GL without compromising the performance. Expand

Graph Attention Networks

- Mathematics, Computer Science
- ICLR
- 2018

We present graph attention networks (GATs), novel neural network architectures that operate on graph-structured data, leveraging masked self-attentional layers to address the shortcomings of prior… Expand

Data Augmentation for Graph Neural Networks

- Computer Science, Mathematics
- AAAI
- 2021

This work shows that neural edge predictors can effectively encode class-homophilic structure to promote intra- class edges and demote inter-class edges in given graph structure, and introduces the GAug graph data augmentation framework, which leverages these insights to improve performance in GNN-based node classification via edge prediction. Expand

Heterogeneous Graph Structure Learning for Graph Neural Networks

- Computer Science
- AAAI
- 2021

This work makes the first attempt towards learning an optimal heterogeneous graph structure for HGNNs and proposes a novel framework HGSL, which jointly performs Heterogeneous Graph Structure Learning and GNN parameter learning for classification. Expand

GNNGuard: Defending Graph Neural Networks against Adversarial Attacks

- Computer Science, Mathematics
- NeurIPS
- 2020

GNNGuard, a general defense approach against a variety of training-time attacks that perturb the discrete graph structure, is developed and can effectively restore the state-of-the-art performance of GNNs in the face of various adversarial attacks. Expand

Graph Structure Learning for Robust Graph Neural Networks

- Computer Science, Mathematics
- KDD
- 2020

A general framework Pro-GNN is proposed, which can jointly learn a structural graph and a robust graph neural network model from the perturbed graph guided by these properties, and achieves significantly better performance compared with the state-of-the-art defense methods, even when the graph is heavily perturbed. Expand

Graph-Revised Convolutional Network

- Computer Science, Mathematics
- ECML/PKDD
- 2020

A GCN-based graph revision module is introduced for predicting missing edges and revising edge weights w.r.t. downstream tasks via joint optimization, which shows that GRCN consistently outperforms strong baseline methods by a large margin. Expand

Hierarchical Graph Pooling with Structure Learning

- Computer Science, Mathematics
- AAAI 2020
- 2019

A novel graph pooling operator, called Hierarchical Graph Pooling with Structure Learning (HGP-SL), which can be integrated into various graph neural network architectures, and introduces a structure learning mechanism to learn a refined graph structure for the pooled graph at each layer. Expand

Topology Optimization based Graph Convolutional Network

- Computer Science
- IJCAI
- 2019

A novel Topology Optimization based Graph Convolutional Networks (TO-GCN) to fully utilize the potential information by jointly refining the network topology and learning the parameters of the FCN is proposed. Expand