Graph Neural Networks for Tabular Data Learning (GNN4TDL)

Table of contents

• ✨ News
• Intro: Graph Neural Networks for Tabular Data Learning
• Homogeneous Instance Graphs
• Homogeneous Feature Graphs
• Bipartite Graphs
• Heterogeneous Graphs
• Hypergraph GNNs
• Knowledge Graphs

👉 If you notice any errors or have suggestions, feel free to share them with us!

⭐️ If you find this resource helpful, please consider to star this repository and cite our survey paper:

@misc{li2024graph,
      title={Graph Neural Networks for Tabular Data Learning: A Survey with Taxonomy and Directions}, 
      author={Cheng-Te Li and Yu-Che Tsai and Chih-Yao Chen and Jay Chiehen Liao},
      year={2024},
      eprint={2401.02143},
      archivePrefix={arXiv},
      primaryClass={cs.LG}
}

✨ News

• [2023-12-16] We have released this repository that collects the resources related to GNNs for tabular data learning (GNN4TDL).

Intro: Graph Neural Networks for Tabular Data Learning

The deep learning-based approaches to Tabular Data Learning (TDL), classification and regression, have shown competing performance, compared to their conventional counterparts. However, the latent correlation among data instances and feature values is less modeled in deep neural TDL. Recently, graph neural networks (GNNs), which can enable modeling relations and interactions between different tabular data elements, has received tremendous attention across application domains including TDL. It turns out creating proper graph structures from the input tabular dataset, along with GNN learning, can improve the TDL performance. In this survey, we systematically review the methodologies of designing and applying GNNs for TDL (GNN4TDL). The topics to be covered include: (1) foundations and overview of GNN-based TDL methods; (2) a comprehensive taxonomy of constructing graph structures and representation learning in GNN-based TDL methods; (3) how to apply GNN to various TDL application scenarios and tasks; (4) limitations in current research and future directions.

This survey presents an in-depth exploration into the application of GNNs in tabular data learning. It starts by establishing the fundamental problem statement and introduces various graph types used to represent tabular data. The survey is structured around a detailed GNN-based learning pipeline, encompassing phases like Graph Formulation, where tabular elements are converted into graph nodes; Graph Construction, focusing on establishing connections within these elements; Representation Learning, highlighting how GNNs process these structures to learn data instance features; and Training Plans, discussing the integration of auxiliary tasks for enhanced predictive outcomes.

Homogeneous Instance Graphs

Year	Title	Venue	Paper	Code
2023	EGG-GAE: Scalable Graph Neural Networks for Tabular Data Imputation	ICML 2023	Link	Link
2023	Graph Neural Networks for Missing Value Classification in a Task-driven Metric Space	TKDE 2023	Link
2023	Look Around! A Neighbor Relation Graph Learning Framework for Real Estate Appraisal	AAAI 2023	Link
2023	Homophily-Enhanced Self-Supervision for Graph Structure Learning: Insights and Directions	TNNLS 2023	Link	Link
2023	TabGSL: Graph Structure Learning for Tabular Data Prediction	arXiv 2023	Link
2023	Oversmoothing Relief Graph Convolutional Network-Based Fault Diagnosis Method with Application to the Rectifier of High-Speed Trains	IEEE TII 2023	Link
2022	Self-Adaptation Graph Attention Network via Meta-Learning for Machinery Fault Diagnosis with Few Labeled Data	IEEE TIM 2022	Link
2023	A Graph Neural Network-based Bearing Fault Detection Method	Scientific Reports 2023	Link
2023	Latent Graphs for Semi-Supervised Learning on Biomedical Tabular Data	IJCLR 2023	Link
2023	Interpretable Graph Neural Networks for Tabular Data	arXiv 2023	Link
2022	Differentiable graph module (dgm) for graph convolutional networks	IEEE TPAMI 2022	Link	Link
2022	Towards Unsupervised Deep Graph Structure Learning	WWW 2022	Link	Link
2022	LUNAR: Unifying Local Outlier Detection Methods via Graph Neural Networks	AAAI 2022	Link	Link
2022	Graph Regularized Autoencoder and its Application in Unsupervised Anomaly Detection	IEEE TPAMI 2022	Link
2022	Deep Unsupervised Active Learning on Learnable Graphs	IEEE TNNLS 2022	Link
2022	Graph autoencoder-based unsupervised outlier detection	Information Sciences 2022	Link	Link
2022	Structure-Aware Siamese Graph Neural Networks for Encounter-Level Patient Similarity Learning	Journal of Biomedical Informatics 2022	Link
2022	GEDI: A Graph-based End-to-end Data Imputation Framework	arXiv 2022	Link
2021	Self-supervision improves structure learning for graph neural networks.	NeurIPS 2021	Link	Link
2021	Predicting Patient Outcomes with Graph Representation Learning	DLG-AAAI 2021	Link	Link
2021	A Weighted Patient Network-based Framework for Predicting Chronic Diseases Using Graph Neural Networks	Scientific Reports 2021	Link
2021	k-Nearest Neighbor Learning with Graph Neural Networks	Mathematics 2021	Link
2020	Iterative Deep Graph Learning for Graph Neural Networks: Better and Robust Node Embeddings	NeurIPS 2020	Link	Link
2020	Missing Data Imputation with Adversarially-trained Graph Convolutional Networks	Neural Networks 2020	Link	Link
2019	Learning Discrete Structures for Graph Neural Networks	ICML 2019	Link	Link

Homogeneous Feature Graphs

Year	Title	Venue	Paper	Code
2024	Graph Neural Network Contextual Embedding for Deep Learning on Tabular Data	Neural Networks 2024	Link	Link
2023	T2G-Former: Organizing Tabular Features into Relation Graphs Promotes Heterogeneous Feature Interaction	AAAI 2023	Link	Link
2023	Deep Tabular Data Modeling With Dual-Route Structure-Adaptive Graph Networks	IEEE TKDE 2023	Link
2023	Causality-based CTR Prediction using Graph Neural Networks	Information Processing & Management 2023	Link
2023	FT-GAT: Graph Neural Network for Predicting Spontaneous Breathing Trial Success in Patients with Mechanical Ventilation	Computer Methods and Programs in Biomedicine 2023	Link
2022	Table2Graph: Transforming Tabular Data to Unifed Weighted Graph	IJCAI 2022	Link
2022	Local Contrastive Feature Learning for Tabular Data	CIKM 2022	Link
2021	FIVES: Feature Interaction Via Edge Search for Large-Scale Tabular Data	KDD 2021	Link
2021	TabularNet: A Neural Network Architecture for Understanding Semantic Structures of Tabular Data	KDD 2021	Link
2021	GCN-Int: A Click-Through Rate Prediction Model Based on Graph Convolutional Network Interaction	IEEE Access 2021	Link
2019	Fi-GNN: Modeling Feature Interactions via Graph Neural Networks for CTR Prediction	CIKM 2019	Link	Link

Bipartite Graphs

Year	Title	Venue	Paper	Code
2023	Data Imputation with Iterative Graph Reconstruction	AAAI 2023	Link
2022	Learning Enhanced Representations for Tabular Data via Neighborhood Propagation	NeurIPS 2022	Link	Link
2022	Relational Multi-Task Learning: Modeling Relations between Data and Tasks	ICLR 2022	Link	Link
2022	Predicting the Survival of Cancer Patients with Multimodal Graph Neural Network	IEEE TCBB 2022	Link
2021	Towards Open-World Feature Extrapolation: An Inductive Graph Learning Approach	NeurIPS 2021	Link	Link
2021	MedGraph: Structural and Temporal Representation Learning of Electronic Medical Records	ECAI 2021	Link	Link
2021	Disease Prediction via Graph Neural Networks	IEEE JBHI 2021	Link	Link
2020	Handling Missing Data with Graph Representation Learning	NeurIPS 2020	Link	Link
2019	Large-Scale Heterogeneous Feature Embedding	AAAI 2019	Link	Link

Heterogeneous Graphs

Year	Title	Venue	Paper	Code
2023	Relational Deep Learning: Graph Representation Learning on Relational Databases	arXiv 2023	Link	Link
2023	GCondNet: A Novel Method for Improving Neural Networks on Small High-Dimensional Tabular Data	NeurIPS 2023 TRL Workshop	Link
2023	Lifelong Property Price Prediction: A Case Study for the Toronto Real Estate Market	IEEE TKDE 2023	Link	Link
2023	GraphFC: Customs Fraud Detection with Label Scarcity	CIKM 2023	Link	Link
2023	GFS: Graph-based Feature Synthesis for Prediction over Relational Databases	arXiv 2023	Link
2022	GCF-RD: A Graph-based Contrastive Framework for Semi-Supervised Learning on Relational Databases	ACM CIKM 2022	Link	Link
2022	Reinforced Neighborhood Selection Guided Multi-Relational Graph Neural Networks	ACM TOIS 2022	Link	Link
2022	H2-Fdetector: A GNN-based Fraud Detector with Homophilic and Heterophilic Connections	WWW 2022	Link
2022	xFraud: Explainable Fraud Transaction Detection	VLDB 2022	Link	Link
2022	AUC-oriented Graph Neural Network for Fraud Detection	WWW 2022	Link
2022	Hierarchical Multi-Modal Fusion on Dynamic Heterogeneous Graph for Health Insurance Fraud Detection	IEEE ICME 2022	Link
2021	Pick and Choose: A GNN-based Imbalanced Learning Approach for Fraud Detection	WWW 2021	Link	Link
2021	Learning Graph Meta Embeddings for Cold-Start Ads in Click-Through Rate Prediction	SIGIR 2021	Link	Link
2021	Modeling Heterogeneous Graph Network on Fraud Detection: A Community-based Framework with Attention Mechanism	CIKM 2021	Link
2021	Towards Consumer Loan Fraud Detection: Graph Neural Networks with Role-Constrained Conditional Random Field	AAAI 2021	Link
2021	Intention-Aware Heterogeneous Graph Attention Networks for Fraud Transactions Detection	KDD 2021	Link
2021	TabGNN: Multiplex Graph Neural Network for Tabular Data Prediction	DLP-KDD 2021	Link	Link
2020	Enhancing Graph Neural Network-based Fraud Detectors against Camouflaged Fraudsters	CIKM 2020	Link	Link
2020	Loan Default Analysis with Multiplex Graph Learning	ACM CIKM 2020	Link
2020	Financial Risk Analysis for SMEs with Graph-based Supply Chain Mining	IJCAI 2020	Link
2020	Heterogeneous Similarity Graph Neural Network on Electronic Health Records	IEEE BigData 2020	Link
2020	Supervised Learning on Relational Databases with Graph Neural Networks	ICLR 2020	Link	Link
2020	Learning the Graphical Structure of Electronic Health Records with Graph Convolutional Transformer	AAAI 2020	Link	Link

Hypergraphs

Year	Title	Venue	Paper	Code
2023	HYTREL: Hypergraph-enhanced Tabular Data Representation Learning	NeurIPS 2023	Link	Link
2022	Learning Enhanced Representations for Tabular Data via Neighborhood Propagation	NeurIPS 2022	Link	Link
2022	Hypergraph Contrastive Learning for Electronic Health Records	SDM 2022	Link

Knowledge Graphs

Year	Title	Venue	Paper	Code
2023	High dimensional, tabular deep learning with an auxiliary knowledge graph	NeurIPS 2023	Link	link