Enhancing Robustness of Graph Convolutional Networks via Dropping Graph Connections

Mathematics

• Attack 32%
• Biased 7%
• Design 5%
• Evaluate 6%
• Experimental Results 8%
• Graph in graph theory 37%
• Learning 14%
• Manipulation 16%
• Neural Networks 7%
• Observation 5%
• Performance 9%
• Real-world Applications 10%
• Robustness 67%
• Subgraph 6%
• Training 8%
• Vertex of a graph 4%

Engineering & Materials Science

• Graph structures 100%
• Neural networks 49%
• Sampling 56%