Research map / Probabilistic and scientific modeling / Molecules and proteins
MOES-Pred: Molecular Structural Representation Learning by Adaptive Energy-Sentinel Vibration for Generalized Property Prediction
Zhiran Hou, Tinghuai Ma, Huan Rong, Li Jia, Anouar Imel, Heng Zhang, Ming Li
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 energy sentinel (ablation) | MOES-Pred · MD22 Buckyball force | MAE: 1.637 -> 0.421 | Improved |
| add molecule-specific perturbation schemes (ablation) | MOES-Pred · MD22 Buckyball force | MAE: 0.502 -> 0.421 | Improved |
| add representation structure enhancement (BRICS motifs, influence matrix) (ablation) | MOES-Pred · MD22 Buckyball force | MAE: 1.367 -> 0.421 | Improved |
| add representation structure enhancement (BRICS motifs, influence matrix) (ablation) | MOES-Pred · QM9 G | MAE: 9.51 -> 6.01 | Improved |
| add energy sentinel (ablation) | MOES-Pred · QM9 G | MAE: 8.32 -> 6.01 | Improved |
| add molecule-specific perturbation schemes (ablation) | MOES-Pred · QM9 Gap | MAE: 35.8 -> 27.7 | Improved |
| add energy sentinel (ablation) | MOES-Pred · ISO17 force | MAE: 1.62 -> 1.01 | Improved |
| replace uniform Gaussian noise (Coord) with molecule-specific vibrational noise (ablation) | TorchMD-NET denoising pre-training · MD22 Double-walled nanotube force | MAE: 2.261 -> 0.705 | Improved |
| replace Frad noise with MOES-Pred noise | denoising pre-training · MD22 Stack force | MAE: F-VRN 0.281 -> MOES-Pred 0.296 | Mixed |
| replace Frad noise with MOES-Pred noise | denoising pre-training · QM9 12 properties | MAE | Mixed |
| increase Coord noise amplitude (ablation) | Coord · force prediction | MAE | Hurt |
| add molecule-aware perturbation amplitude vs Frad | MOES-Pred · perturbation amplitude statistics | perturbation amplitude: 12.6% and 13.4% larger stretching and bending amplitude | Improved |
Closest papers by meaning
- Sliced Denoising: A Physics-Informed Molecular Pre-Training Method (ICLR 2024 · similarity 0.90)
- Learning 3D Anisotropic Noise Distributions Improves Molecular Force Fields (NeurIPS 2025 · similarity 0.88)
- 3D Denoisers Are Good 2D Teachers: Molecular Pretraining via Denoising and Cross-Modal Distillation (AAAI 2025 · similarity 0.87)
- MolSpectra: Pre-training 3D Molecular Representation with Multi-modal Energy Spectra (ICLR 2025 · similarity 0.86)
- Multimodal Molecular Pretraining via Modality Blending (ICLR 2024 · similarity 0.86)
- Beyond Atoms: Enhancing Molecular Pretrained Representations with 3D Space Modeling (ICML 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.
