Research map / Graph learning / GNN architectures
Hierarchical Multi Scale Graph Neural Networks: Scalable Heterophilous Learning with Oversmoothing and Oversquashing Mitigation
MD SAZZAD Hossen, Avimanyu Sahoo
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 adaptive signed affinity instead of fixed adjacency (ablation) | HMH · Actor | accuracy (%): 39.8 -> 43.3 | Improved |
| add multi-level hierarchy instead of Haar filtering at level 0 only (ablation) | HMH · Actor | accuracy (%): 37.5 -> 43.3 | Improved |
| use learned localized Haar basis instead of Chebyshev global basis (ablation) | HMH · Actor | accuracy (%): 38.1 -> 43.3 | Improved |
| add adaptive signed affinity instead of fixed adjacency (ablation) | HMH · Chameleon (non-filtered) | accuracy (%): 37.0 -> 41.8 | Improved |
| add multi-level hierarchy instead of Haar filtering at level 0 only (ablation) | HMH · Chameleon (non-filtered) | accuracy (%): 34.2 -> 41.8 | Improved |
| use learned localized Haar basis instead of Chebyshev global basis (ablation) | HMH · Chameleon (non-filtered) | accuracy (%): 37.2 -> 41.8 | Improved |
| add adaptive signed affinity instead of fixed adjacency (ablation) | HMH · Squirrel (non-filtered) | accuracy (%): 47.0 -> 50.9 | Improved |
| add multi-level hierarchy instead of Haar filtering at level 0 only (ablation) | HMH · Squirrel (non-filtered) | accuracy (%): 45.4 -> 50.9 | Improved |
| use learned localized Haar basis instead of Chebyshev global basis (ablation) | HMH · Squirrel (non-filtered) | accuracy (%): 48.3 -> 50.9 | Improved |
| add adaptive signed affinity instead of fixed adjacency (ablation) | HMH · PubMed | accuracy (%): 87.3 -> 91.4 | Improved |
| use HMH instead of GCN | HMH · Penn94 spokes | accuracy (%): 71.35 -> 82.15 | Improved |
| use HMH instead of GIN/ChebNet | HMH · Tree-NeighborsMatch depth 10 | accuracy (%): 50% -> >98% | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- Specformer — Specformer: Spectral Graph Neural Networks Meet Transformers (ICLR 2023) · better on 0 of 2 shared results
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
- Unifying Homophily and Heterophily for Spectral Graph Neural Networks via Triple Filter Ensembles (NeurIPS 2024 · similarity 0.90)
- LOHA: Direct Graph Spectral Contrastive Learning Between Low-Pass and High-Pass Views (AAAI 2025 · similarity 0.90)
- HeroFilter: Adaptive Spectral Graph Filter for Varying Heterophilic Relations (NeurIPS 2025 · similarity 0.89)
- Homophily-Related: Adaptive Hybrid Graph Filter for Multi-View Graph Clustering (AAAI 2024 · similarity 0.89)
- How Universal Polynomial Bases Enhance Spectral Graph Neural Networks: Heterophily, Over-smoothing, and Over-squashing (ICML 2024 · similarity 0.89)
- Graph Neural Networks Are More Than Filters: Revisiting and Benchmarking from A Spectral Perspective (ICLR 2025 · 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.
