Research map / Graph learning / GNN architectures
GT-SNT: A Linear-Time Transformer for Large-Scale Graphs via Spiking Node Tokenization
Huizhe Zhang, Jintang Li, Yuchang Zhu, Huazhen Zhong, Liang Chen
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 |
|---|---|---|---|
| use spiking node tokenization with codebook guided attention instead of Performer (ablation) | GT-SNT · ogbn-arxiv | accuracy (%): 71.2 -> 72.4 | Improved |
| use CGSA instead of Linformer (ablation) | GT-SNT · ogbn-arxiv | accuracy (%): 65.2 -> 72.4 | Improved |
| use spiking node tokenization instead of VQ-VAE codebook (ablation) | GT-SNT · PubMed | accuracy (%): 79.8 -> 80.6 | Improved |
| use spiking node tokenization instead of FSQ (ablation) | GT-SNT · PubMed | accuracy (%): 78.4 -> 80.6 | Improved |
| use spiking node tokenization instead of FSQ (ablation) | GT-SNT · Co-CS | accuracy (%): 93.7 -> 93.7 | No clear effect |
| use PLIF spiking neuron instead of IF (ablation) | GT-SNT · PubMed | accuracy (%): 81.6 -> 80.6 | Hurt |
| use PLIF spiking neuron instead of LIF (ablation) | GT-SNT · PubMed | accuracy (%): 79.6 -> 80.6 | Improved |
| use STFNorm instead of LayerNorm (ablation) | GT-SNT · PubMed | accuracy (%): 78.9 -> 80.6 | Improved |
| swap GT-SNT's normalization to STFNorm variant (ablation) | GT-SNT · PubMed | accuracy (%): 80.6 -> 82.6 | Improved |
| use spiking node tokenization instead of repeated-graph or retrieval-based tokenization | GT-SNT · Co-Physics inference | inference latency (s): 2.60 -> 0.02 | Improved |
| build tokens from spike counts instead of pre-defined VQ codebook | GT-SNT · Cora and CS codebook usage | codebook usage: GOAT 10.6% and VQGraph 16.9% (Cora) -> 100% | Improved |
| increase propagation steps T beyond 6 (ablation) | GT-SNT · Cora and CS | accuracy | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 5 on most shared results.
- SGFormer — SGFormer: Simplifying and Empowering Transformers for Large-Graph Representations (NeurIPS 2023) · better on 3 of 6 shared results
- VQGraph — VQGraph: Rethinking Graph Representation Space for Bridging GNNs and MLPs (ICLR 2024) · better on 1 of 3 shared results
- GOAT — GOAT: A Global Transformer on Large-scale Graphs (ICML 2023) · better on 2 of 3 shared results
- NAGphormer — NAGphormer: A Tokenized Graph Transformer for Node Classification in Large Graphs (ICLR 2023) · better on 1 of 1 shared result
- NodeFormer — NodeFormer: A Scalable Graph Structure Learning Transformer for Node Classification (NeurIPS 2022) · 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
- SGNNBench: A Holistic Evaluation of Spiking Graph Neural Networks on Large-scale Graphs (NeurIPS 2026 · similarity 0.92)
- Delay-DSGN: A Dynamic Spiking Graph Neural Network with Delay Mechanisms for Evolving Graph (ICML 2025 · similarity 0.90)
- A Graph is Worth 1-bit Spikes: When Graph Contrastive Learning Meets Spiking Neural Networks (ICLR 2024 · similarity 0.88)
- Dynamic Spiking Graph Neural Networks (AAAI 2024 · similarity 0.88)
- Self-Supervised Discovery of Neural Circuits in Spatially Patterned Neural Responses with Graph Neural Networks (NeurIPS 2025 · similarity 0.88)
- What Improves the Generalization of Graph Transformers? A Theoretical Dive into the Self-attention and Positional Encoding (ICML 2024 · similarity 0.88)
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.
