Research map / Graph learning / GNN architectures
Neural Graph Embedding for Neural Architecture Search
Wei Li, Shaogang Gong, Xiatian Zhu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add plain learned embedding for operation probabilities (ablation) | DARTS-style baseline · CIFAR-10 | test error (%): 3.44 -> 3.14 | Improved |
| add RNN modelling of node relationships on top of plain embedding (ablation) | plain embedding · CIFAR-10 | test error (%): 3.14 -> 2.71 | Improved |
| replace RNN with 3-layer GCN neural graph embedding (ablation) | NGE · CIFAR-10 | test error (%): 2.71 -> 2.60 | Improved |
| replace RNN with GCN neural graph embedding (ablation) | NGE · CIFAR-100 | test error (%): 16.97 -> 16.53 | Improved |
| add plain embedding to baseline (ablation) | DARTS-style baseline · CIFAR-100 | test error (%): 17.90 -> 17.60 | Improved |
| use NGE instead of GHN graph-based search | graph-based NAS · CIFAR-10 | test error (%): 2.84 -> 2.60 | Improved |
| NGE cell searched on CIFAR-10 transferred to ImageNet vs DARTS | NGE · ImageNet mobile | top-1 error (%): 26.7 -> 25.3 | Improved |
| NGE vs ProxylessNAS-G on CIFAR-10 | NGE · CIFAR-10 | test error (%): 2.08 -> 2.60 | Hurt |
Closest papers by meaning
- NASGEM: Neural Architecture Search via Graph Embedding Method (AAAI 2021 · similarity 0.91)
- Rethinking Graph Neural Architecture Search From Message-Passing (CVPR 2021 · similarity 0.89)
- Neural Architecture Retrieval (ICLR 2024 · similarity 0.89)
- One-shot Graph Neural Architecture Search with Dynamic Search Space (AAAI 2021 · similarity 0.89)
- Towards characterizing the value of edge embeddings in Graph Neural Networks (ICML 2025 · similarity 0.88)
- NN-Former: Rethinking Graph Structure in Neural Architecture Representation (CVPR 2025 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
