Khái niệm cốt lõi
Proposing NAS-Bench-Graph as a benchmark for fair, reproducible, and efficient evaluations in GraphNAS research.
Thống kê
"Specifically, we construct a unified, expressive yet compact search space, covering 26,206 unique graph neural network (GNN) architectures and propose a principled evaluation protocol."
"All the code and evaluation results have been open-sourced."
"The total time cost of creating our benchmark is approximately 8,000 GPU hours."
Trích dẫn
"Graph neural architecture search (GraphNAS) has recently aroused considerable attention in both academia and industry."
"Our proposed benchmark can enable fair, fully reproducible, and efficient comparisons for different GraphNAS methods."