Research map / Graph learning / Graph self-supervised learning
Cost-aware Graph Generation: A Deep Bayesian Optimization Approach
Jiaxu Cui, Bo Yang, Bingyi Sun, Jiming Liu
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 |
|---|---|---|---|
| Use CAGG instead of DGBO | Molecular optimization · QM9, logP−SA | # evaluations to find optimum: 189 -> 128 | Improved |
| Use CAGG instead of DGBO | Molecular optimization · QM9, 5×QED−SA | # evaluations to find optimum: 448 -> 144 | Improved |
| Use CAGG instead of DGBO | Molecular optimization · QM9, logP−SA | total cost (hours): 189.3 -> 129.2 | Improved |
| Use CAGG instead of DGBO | Molecular optimization · QM9, 5×QED−SA | total cost (hours): 449.0 -> 145.3 | Improved |
| Use CAGG with a 50.3-hour budget instead of REA with a 205-hour budget | Cell-based NAS · CIFAR100 | classification accuracy (%): 73.09±0.25 -> 72.87±0.27 | Hurt |
| Use CAGG with a 50.3-hour budget instead of REA with a 205-hour budget | Cell-based NAS · ImageNet16-120 | classification accuracy (%): 46.12±0.67 -> 46.13±0.46 | No clear effect |
| Increase CAGG budget from 50.3 to 140.9 hours | Cell-based NAS · CIFAR100 | classification accuracy (%): 72.87±0.27 -> 73.25±0.42 | Improved |
| Increase CAGG budget from 50.3 to 140.9 hours | Cell-based NAS · ImageNet16-120 | classification accuracy (%): 46.13±0.46 -> 46.34±0.27 | Improved |
| Use CAGG instead of NASGBO | Multi-branch NAS · Indoor | mean squared error: 0.090±0.012 -> 0.063±0.001 | Improved |
| Use CAGG instead of NASGBO | Multi-branch NAS · Slice | mean squared error: 0.560±0.046 -> 0.433±0.010 | Improved |
| Use CAGG with a 4-hour budget instead of the 12-hour baselines | Multi-branch NAS · Indoor | mean squared error: best baseline 0.090±0.012; CAGG 0.072±0.003 | Improved |
| Use CAGG with an 8-hour budget instead of the 12-hour baselines | Multi-branch NAS · Slice | mean squared error: 0.615±0.044 -> 0.625±0.001 | No clear effect |
Closest papers by meaning
- On the Scalability of GNNs for Molecular Graphs (NeurIPS 2024 · similarity 0.86)
- Controllable 3D Molecular Generation for Structure-Based Drug Design Through Bayesian Flow Networks and Gradient Integration (NeurIPS 2025 · similarity 0.86)
- Position: Graph Learning Will Lose Relevance Due To Poor Benchmarks (ICML 2025 · similarity 0.86)
- Empowering Active Learning for 3D Molecular Graphs with Geometric Graph Isomorphism (NeurIPS 2024 · similarity 0.85)
- Smooth Interpolation for Improved Discrete Graph Generative Models (ICML 2025 · similarity 0.85)
- Inexact Column Generation for Bayesian Network Structure Learning via Difference-of-Submodular Optimization (NeurIPS 2025 · similarity 0.85)
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.
