Research map / Graph learning / Graph self-supervised learning
Fast Graph Generation via Spectral Diffusion
Tianze Luo, Zhanfeng Mo, Sinno Jialin Pan
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 |
|---|---|---|---|
| diffuse in graph spectrum space instead of whole adjacency matrix (GSDM vs GDSS) | Graph diffusion model · Community-small | Avg. MMD: 0.046 -> 0.009 | Improved |
| diffuse in graph spectrum space instead of whole space | Graph diffusion model · Grid | Avg. MMD: 0.062 -> 0.0007 | Improved |
| diffuse in graph spectrum space instead of whole space | Graph diffusion model · Enzymes | Avg. MMD: 0.046 -> 0.037 | Improved |
| spectral diffusion instead of GDSS on QM9 | Graph diffusion model · QM9 | FCD: 2.900 -> 2.650 | Improved |
| spectral diffusion instead of GDSS on ZINC250k | Graph diffusion model · ZINC250k | Validity w/o correction (%): 97.01 -> 92.70 | Hurt |
| spectral diffusion instead of GDSS | Graph diffusion model · QM9 | Generation time (s): 1.06e^3 -> 18.02 | Improved |
| spectral diffusion instead of GDSS | Graph diffusion model · ZINC250k | Generation time (s): 2.11e^3 -> 45.91 | Improved |
| keep only top 30% eigenvalues instead of all (ablation) | GSDM · Enzymes | Avg. MMD: 0.039 vs 0.038 full | No clear effect |
| keep only top 10% eigenvalues instead of all (ablation) | GSDM · Community-small | Avg. MMD: 0.009 -> 0.781 | Hurt |
| keep only top 60% eigenvalues instead of all (ablation) | GSDM · Grid | Avg. MMD: 0.001 -> 0.001 | No clear effect |
| vary number of diffusion steps (GSDM vs GDSS) (ablation) | GSDM · Grid and Community-small | performance | Improved |
| vary diffusion schedule (VE/VP, six noise schedules) (ablation) | GSDM · Ego-small and Grid | Avg. MMD | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- GraphDF — GraphDF: A Discrete Flow Model for Molecular Graph Generation (ICML 2021) · better on 2 of 2 shared results
- GraphRNN — GraphRNN: Generating Realistic Graphs with Deep Auto-regressive Models (ICML 2018) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Generating Graphs via Spectral Diffusion (ICLR 2025 · similarity 0.93)
- Discrete-state Continuous-time Diffusion for Graph Generation (NeurIPS 2024 · similarity 0.90)
- Graph Generation with Diffusion Mixture (ICML 2024 · similarity 0.90)
- SBGD: Improving Graph Diffusion Generative Model via Stochastic Block Diffusion (ICML 2025 · similarity 0.90)
- Advancing Graph Generation through Beta Diffusion (ICLR 2025 · similarity 0.90)
- Unifying Generation and Prediction on Graphs with Latent Graph Diffusion (NeurIPS 2024 · similarity 0.89)
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.
