IMA-GNN: In-Memory Acceleration of Centralized and Decentralized Graph Neural Networks at the Edge

Keyphrases

• Graph Neural Network 100%
• In-memory 100%
• Decentralized Setting 100%
• Power Consumption 50%
• Edge Devices 50%
• Communication Latency 50%
• Taxi Demand 50%
• Large Graphs 50%
• Computation Latency 50%
• Neural Network Inference 50%
• Edge Computation 50%
• Communication Speed 50%
• Cross-layer Model 50%
• Graph Datasets 50%
• Practical Case Study 50%

Computer Science

• Graph Neural Network 100%
• Speed-up 50%
• Case Study 50%
• Power Consumption 50%
• Communication Latency 50%
• Network Inference 50%
• Accelerator Memory 50%

Neuroscience

• Neural Network 100%