Research map / Probabilistic and scientific modeling / Molecules and proteins
Molecular Property Prediction: A Multilevel Quantum Interactions Modeling Perspective
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 |
|---|---|---|---|
| Model molecular interactions level by level rather than blending levels into one embedding (ablation) | MGCN · QM9, property U | MAE: 0.03683 | Helped |
| Use MGCN instead of SchNet | Molecular property prediction · QM9, εLUMO | MAE: 0.0372 -> 0.0574 | Hurt |
| Use MGCN instead of enn-s2s | Molecular property prediction · QM9, μ | MAE: 0.030 -> 0.056 | Hurt |
| Use MGCN instead of the strongest listed baseline | Molecular property prediction · QM9, the other 11 properties | MAE | Helped |
| Use MGCN instead of SchNet or DTNN | Molecular property prediction · ANI-1 | MAE: 0.108 -> 0.078; 0.113 -> 0.078 | Helped |
| Train MGCN with fewer labeled molecules rather than use the listed baselines | MGCN · QM9, U0; 50,000 training molecules | MAE: 0.0249 -> 0.0229 | Helped |
| Train MGCN with fewer labeled molecules rather than use the listed baselines | MGCN · QM9, U0; 100,000 training molecules | MAE: 0.0147 -> 0.0142 | Helped |
| Use fewer or more than four interaction layers (ablation) | MGCN · QM9, U0 | MAE | Hurt |
| Use small-molecule embeddings to initialize training on large molecules (ablation) | MGCN · QM9, large molecules | MAE | Helped |
| Train on small molecules and apply directly to large molecules (ablation) | MGCN · QM9, large molecules | MAE | Helped |
Closest papers by meaning
- Learning the Electronic Hamiltonian of Large Atomic Structures (ICML 2025 · similarity 0.88)
- $\nabla^2$DFT: A Universal Quantum Chemistry Dataset of Drug-Like Molecules and a Benchmark for Neural Network Potentials (NeurIPS 2024 · similarity 0.87)
- MIPT: Multilevel Informed Prompt Tuning for Robust Molecular Property Prediction (ICML 2025 · similarity 0.87)
- Learning Over Molecular Conformer Ensembles: Datasets and Benchmarks (ICLR 2024 · similarity 0.87)
- Neural Atoms: Propagating Long-range Interaction in Molecular Graphs through Efficient Communication Channel (ICLR 2024 · similarity 0.87)
- REBIND: Enhancing Ground-state Molecular Conformation Prediction via Force-Based Graph Rewiring (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.