Graph Neural Networks (GNNs) are revolutionizing the way we handle data that involves relationships between entities. From social networks to molecular structures, GNNs are proving to be effective tools for a variety of applications. In this article, we will explore a comprehensive collection of resources related to GNNs, guiding you through crucial papers, surveys, applications, and libraries.

Contents

• Survey Papers
• Papers
  • Recurrent Graph Neural Networks
  • Convolutional Graph Neural Networks
  • Graph Autoencoders
    • Network Embedding
    • Graph Generation
  • Spatial-Temporal Graph Neural Networks
  • Applications
    • Computer Vision
    • Natural Language Processing
    • Internet
    • Recommender Systems
    • Healthcare
    • Chemistry
    • Physics
    • Others
• Libraries

Survey Papers

Survey papers provide excellent insights into the GNN landscape. Here are some noteworthy ones:

• A Comprehensive Survey on Graph Neural Networks. Zonghan Wu, et al. 2019 [paper]
• Deep learning on graphs: A survey. Ziwei Zhang, et al. 2018 [paper]
• Graph Neural Networks: A Review of Methods and Applications. Jie Zhou, et al. 2018 [paper]

Papers

When diving into GNNs, the following papers are key resources for understanding both foundational concepts and advanced methodologies.

Recurrent Graph Neural Networks

• Supervised neural networks for the classification of structures. A. Sperduti and A. Starita. 1997 [paper]
• The graph neural network model. Franco Scarselli, et al. 2009 [paper]

Convolutional Graph Neural Networks

The world of Convolutional Graph Neural Networks is vast and complex. Think of it as navigating a spider’s web. The nodes are like junctions where decisions are made, and the edges represent the pathways connecting those decisions. The intricate patterns help us see how different elements in our data relate to each other.

• Spectral networks and locally connected networks on graphs. Joan Bruna, et al. 2014 [paper]
• Semi-supervised classification with graph convolutional networks. Thomas N. Kipf and Max Welling. 2017 [paper]

Graph Autoencoders

Graph autoencoders are crucial for representation learning in GNNs. They compress information while maintaining critical relationships. Imagine them as a set of efficient librarians who categorize and summarize information from a massive library into succinct books. Users can quickly find what they need without sifting through every tome.

Network Embedding

• Variational graph auto-encoders. Thomas N. Kipf, Max Welling. 2016 [paper]

Graph Generation

• MolGAN: An implicit generative model for small molecular graphs. Nicola De Cao, Thomas Kipf. 2018 [paper]

Spatial-Temporal Graph Neural Networks

When dealing with temporal data, Spatial-Temporal Graph Neural Networks shine. They can be likened to GPS systems that adjust your path based on traffic conditions, constantly reevaluating routes based on new real-time data.

• Deep multi-view spatial-temporal network for taxi. Huaxiu Yao, et al. 2018 [paper]

Applications

GNNs extend their limbs into various applications:

• Computer Vision
• Natural Language Processing
• Internet
• Recommender Systems
• Healthcare
• Chemistry
• Physics
• Others

Computer Vision

• 3D Graph Neural Networks for RGBD Semantic Segmentation. Xiaojuan Qi et al. CVPR 2017 [paper]

Natural Language Processing

• Graph Convolutional Encoders for Syntax-aware Neural Machine Translation. Joost Bastings et al. EMNLP 2017 [paper]

Internet

• Adversarial attacks on neural networks for graph data. Daniel Zügner et al. KDD 2018 [paper]

Recommender Systems

• Graph Convolutional Matrix Completion. Rianne van den Berg et al. 2017 [paper]

Healthcare

• GRAM: Graph-based Attention Model for Healthcare Representation Learning. Edward Choi et al. KDD 2017 [paper]

Chemistry

• Protein interface prediction using graph convolutional networks. Alex Fout et al. NIPS 2017 [paper]

Physics

• Interaction Networks for Learning about Objects, Relations and Physics. Peter Battaglia et al. NIPS 2016 [paper]

Others

• Learning to represent programs with graphs. Miltiadis Allamanis et al. ICLR 2017 [paper]

Libraries

When building GNNs, these libraries are invaluable:

• PyTorch Geometric
• Deep Graph Library
• Graph Nets Library

Troubleshooting

If you encounter challenges, here are some troubleshooting tips:

• Ensure you have all necessary dependencies installed for libraries.
• Check for any conflicts between different versions of libraries.
• Review documentation for specific error messages you might encounter.

For more insights, updates, or to collaborate on AI development projects, stay connected with fxis.ai.

At fxis.ai, we believe that such advancements are crucial for the future of AI, as they enable more comprehensive and effective solutions. Our team is continually exploring new methodologies to push the envelope in artificial intelligence, ensuring that our clients benefit from the latest technological innovations.