Research map / Graph learning / Graph self-supervised learning
Mixed-Curvature Geometric Latent Diffusion Model for Graph Generation
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Proposes a mixed-curvature geometric latent diffusion model for graph generation that uses heterogeneous latent geometries to capture structurally diverse graph patterns.
Problem
Existing latent graph diffusion models learn graph representations and generative priors within a single homogeneous latent geometry, which is restrictive when graphs mix hierarchical, cyclic, locally regular, and densely connected structures.
Approach
Replaces the homogeneous latent space of graph diffusion with a mixed-curvature geometric latent space so different structural patterns can be embedded in geometries suited to them before diffusion modeling.
Experimental setup
Builds on: latent graph diffusion models
Closest papers by meaning
- Hyperbolic Geometric Latent Diffusion Model for Graph Generation (ICML 2024 · similarity 0.92)
- Toward a Unified Geometry Understanding : Riemannian Diffusion Framework for Graph Generation and Prediction (NeurIPS 2025 · similarity 0.91)
- Graph Generation with Diffusion Mixture (ICML 2024 · similarity 0.88)
- Unifying Generation and Prediction on Graphs with Latent Graph Diffusion (NeurIPS 2024 · similarity 0.88)
- When Edge Independence Fails: Joint Graph Diffusion with Latent Sociability Priors (NeurIPS 2026 · similarity 0.88)
- Generative Modelling of Structurally Constrained Graphs (NeurIPS 2024 · 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.