Research map / Graph learning / Graph self-supervised learning
Disentangling Homophilic and Heterophilic Patterns for Multi-Domain Graph Foundation Models
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 multi-domain graph foundation model that disentangles homophilic and heterophilic connectivity patterns instead of relying on a single domain-invariant representation space.
Problem
Existing graph foundation models use a single domain-invariant representation space, which blurs homophily and heterophily patterns that govern node relations across multi-domain graphs.
Approach
Disentangles homophilic and heterophilic patterns into separate representation spaces within a multi-domain graph foundation model, rather than forcing one domain-invariant space.
Closest papers by meaning
- What Is Missing For Graph Homophily? Disentangling Graph Homophily For Graph Neural Networks (NeurIPS 2024 · similarity 0.88)
- Unraveling the Impact of Heterophilic Structures on Graph Positive-Unlabeled Learning (ICML 2024 · similarity 0.87)
- PC-Conv: Unifying Homophily and Heterophily with Two-Fold Filtering (AAAI 2024 · similarity 0.87)
- HeroFilter: Adaptive Spectral Graph Filter for Varying Heterophilic Relations (NeurIPS 2025 · similarity 0.87)
- DUPLEX: Dual GAT for Complex Embedding of Directed Graphs (ICML 2024 · similarity 0.87)
- HeterSEED: Semantics–Structure Decoupling for Heterogeneous Graph Learning under Heterophily (NeurIPS 2026 · 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.