Research map / Graph learning / Graph self-supervised learning
Periodic Graph Transformers for Crystal Material Property Prediction
Keqiang Yan, Yi Liu, Yuchao Lin, Shuiwang Ji
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 |
|---|---|---|---|
| periodic invariant radius multi-edge graph instead of Graphormer OCgraph (ablation) | Matformer · JARVIS formation energy | test MAE: 0.0530 -> 0.0348 | Improved |
| add periodic pattern encoding (self-connecting edges) (ablation) | Matformer · JARVIS formation energy | test MAE: 0.0337 -> 0.0325 | Improved |
| radius-based graph instead of fully connected T-fully graph (ablation) | Matformer · JARVIS formation energy | test MAE: 0.0402 -> 0.0325 | Improved |
| add angular information via SBF (ablation) | Matformer · JARVIS formation energy | test MAE | No clear effect |
| add angular information via RBF (ablation) | Matformer · JARVIS formation energy | test MAE: 0.0325 -> 0.0325 | No clear effect |
| sigmoid and normalization instead of softmax-scalar attention (ablation) | Matformer message passing · JARVIS formation energy | test MAE: 0.0376 -> 0.0325 | Improved |
| sigmoid and normalization instead of softmax-vector attention (ablation) | Matformer message passing · JARVIS formation energy | test MAE: 0.0347 -> 0.0325 | Improved |
| larger configuration (5 layers, 4 heads, batch 64) vs 3 layers, 1 head, batch 32 for radius graph without PE (ablation) | Matformer · JARVIS formation energy | test MAE: 0.0348 -> 0.0337 | Improved |
| Matformer instead of ALIGNN | crystal property prediction · JARVIS formation energy efficiency | training time per epoch: 327 s -> 64 s | Improved |
Closest papers by meaning
- Complete and Efficient Graph Transformers for Crystal Material Property Prediction (ICLR 2024 · similarity 0.94)
- Conformal Crystal Graph Transformer with Robust Encoding of Periodic Invariance (AAAI 2024 · similarity 0.93)
- Crystalformer: Infinitely Connected Attention for Periodic Structure Encoding (ICLR 2024 · similarity 0.89)
- Invariant Tokenization of Crystalline Materials for Language Model Enabled Generation (NeurIPS 2024 · similarity 0.86)
- Code-Generated Graph Representations Using Multiple LLM Agents for Material Properties Prediction (ICML 2025 · similarity 0.86)
- Periodic Materials Generation using Text-Guided Joint Diffusion Model (ICLR 2025 · 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.
