Topology-aware Embedding Memory for Continual Learning on Expanding Networks

Keyphrases

• Continual Learning 100%
• Topology-aware 100%
• Memory Replay 100%
• Decoupled Graph Neural Network 100%
• Topological Information 66%
• Tight 33%
• Training Effect 33%
• Memory Consumption 33%
• State-of-the-art Techniques 33%
• Receptive Field 33%
• Sampling Strategy 33%
• Compress 33%
• Training Parameters 33%
• Memory Space 33%
• Compact Vector 33%
• Neighborhood Structure 33%
• Class-incremental 33%
• Average Node Degree 33%
• Topological Neighborhood 33%
• Representative Node 33%
• Coverage Maximization 33%

Computer Science

• Graph Neural Network 100%
• Continual Learning 100%
• Subnetwork 50%
• Space Complexity 25%
• Memory Consumption 25%
• Receptive Field 25%
• Neighborhood Structure 25%
• Representative Node 25%