#GNN
2 related articles
·2 min
GNN Power Load Forecasting: A Diagnostic Protocol to Disentangle Parameter Sharing from Neighbor Access
In GNN power load forecasting, parameter sharing and neighbor access gains are often conflated. This article introduces a diagnostic protocol using per-node, shared-network, and edge-removed GNN comparisons plus degree-preserving rewiring, revealing that edge structure does not contribute the main advantage across multiple datasets, providing empirical evidence for whether maintaining graph structure is worthwhile.
Read more →
·2 min
GPEvac: Real-Time Evacuation Routing for Active Shooter Events Using GNN + PPO
GPEvac combines GNN and PPO reinforcement learning to compute real-time evacuation routes during active shooter events in just 14.73ms on local CPU, significantly reducing threat exposure.
Read more →