Research map / Graph learning / Graph self-supervised learning
Hard-Constrained Graph Generation with Discrete-Projection Diffusion
Xuesong Zhang, Haifeng Sun, Qi Qi, Shengkuan Li, Yuhao Li, Tianyi Kou, Zirui Zhuang, Bo He, Jianxin Liao, Jingyu Wang
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 SMT discrete projection to GDSS | GDSS · Community-small, edge count | VALc: 0.30 -> 1.00 | Improved |
| use SMT projection instead of PRODIGY on DruM | DruM · Community-small, edge count | VALc: 0.55 -> 1.00 | Improved |
| add SMT projection to LayerDAG | LayerDAG · LP dataset (rho = 0) | Validity: 0.56 ± 0.02 -> 0.99 ± 0.00 | Improved |
| replace SMT projector with auto-regressive neural projector (ablation) | NSPSG · LP dataset (rho = 0) | Validity: 0.99 -> 0.97 | No clear effect |
| replace auto-regressive neural projector with GNN projector (ablation) | NSPSG · LP dataset (rho = 0) | Validity: 0.97 ± 0.01 -> 0.76 ± 0.01 | Hurt |
| add SMT projection to DiGress | DiGress+ · Tree (structural constraint) | Valid (%): 91.6 ± 0.7 -> 100.0 ± 0.0 | Improved |
| project only in the late sampling steps instead of the final step only (ablation) | NSPSG · Planar, Tree, Lobster | distribution metrics | Improved |
| project at every sampling step instead of only late steps (ablation) | NSPSG · Planar, Tree, Lobster | distribution shift and cost | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 5 of 5 on most shared results.
- +PRODIGY — PRODIGY: Enabling In-context Learning Over Graphs (NeurIPS 2023) · better on 3 of 4 shared results
- DiGress+ — DiGress: Discrete Denoising diffusion for graph generation (ICLR 2023) · better on 3 of 3 shared results
- DeFoG — DeFoG: Discrete Flow Matching for Graph Generation (ICML 2025) · better on 2 of 2 shared results
- LayerDAG — LayerDAG: A Layerwise Autoregressive Diffusion Model for Directed Acyclic Graph Generation (ICLR 2025) · better on 2 of 2 shared results
- DruM — DRUM: End-To-End Differentiable Rule Mining On Knowledge Graphs (NeurIPS 2019) · 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
- Generative Modelling of Structurally Constrained Graphs (NeurIPS 2024 · similarity 0.90)
- Diffuse, Sample, Project: Plug-And-Play Controllable Graph Generation (ICML 2024 · similarity 0.88)
- SBGD: Improving Graph Diffusion Generative Model via Stochastic Block Diffusion (ICML 2025 · similarity 0.87)
- Discrete-state Continuous-time Diffusion for Graph Generation (NeurIPS 2024 · similarity 0.86)
- Constraints-Guided Diffusion Reasoner for Neuro-Symbolic Learning (AAAI 2026 · similarity 0.86)
- AT-SKM-Net: An Accelerated Trainable Sampling Kaczmarz-Motzkin Framework for Linear Hard-Constraint Feasibility on Dynamic Graphs (NeurIPS 2026 · similarity 0.86)
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.
