Research map / Graph learning / Graph self-supervised learning
Data-Efficient Graph Grammar Learning for Molecular Generation
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| learn a substructure-based grammar and optimize it for class membership | DEG · Isocyanates | Membership: 79.8% -> 96.3% | Helped |
| optimize grammar construction for Retro* score | DEG · Isocyanates | Retro* Score: 5.63% -> 27.2% | Helped |
| learn a substructure-based grammar and optimize it for class membership | DEG · Acrylates | Membership: 48.64% -> 69.6% | Helped |
| optimize grammar construction for Retro* score | DEG · Acrylates | Retro* Score: 36.8% -> 43.9% | Helped |
| learn a substructure-based grammar and optimize it for class membership | DEG · Chain Extenders | Membership: 79.6% -> 93.5% | Helped |
| optimize grammar construction for Retro* score | DEG · Chain Extenders | Retro* Score: 50.6% -> 67.5% | Helped |
| increase training data from 0.15% to 0.3% (ablation) | DEG on the large polymer dataset · Large polymer dataset | Chamfer: 0.62 -> 0.63 | Helped |
| use the DeepChem feature extractor instead of the pretrained GNN (ablation) | DEG · Isocyanates | optimization performance | Mixed |
| vary the balance weights for diversity and Retro* score (ablation) | DEG · Isocyanates and Acrylates | Diversity and Retro* score | Mixed |
| increase the probability of terminal-only production rules during generation (ablation) | DEG grammar sampling · Molecule generation | generation time and diversity | Mixed |
| sample rules uniformly rather than increasing terminal-only rule probability over iterations (ablation) | DEG grammar sampling · Molecule generation | termination | Hurt |
Closest papers by meaning
- Foundation Molecular Grammar: Multi-Modal Foundation Models Induce Interpretable Molecular Graph Languages (ICML 2025 · similarity 0.91)
- Reinforced Molecular Optimization with Neighborhood-Controlled Grammars (NeurIPS 2020 · similarity 0.90)
- Domain-Agnostic Molecular Generation with Chemical Feedback (ICLR 2024 · similarity 0.86)
- Towards Foundational Models for Molecular Learning on Large-Scale Multi-Task Datasets (ICLR 2024 · similarity 0.86)
- Lift Your Molecules: Molecular Graph Generation in Latent Euclidean Space (ICLR 2025 · similarity 0.86)
- LLaMo: Large Language Model-based Molecular Graph Assistant (NeurIPS 2024 · 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.