Research map / Image generation / Fast diffusion sampling
MAST: Motif-Augmented Diffusion with Search Tree for Spectroscopic Molecular Structure Elucidation
Chenghao Jia, Mengdi Liu, Hong Chang, Shiguang Shan, Xilin 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 |
|---|---|---|---|
| add motif prior (ablation) | MAST (w/o MCTS) · QM9S | Acc@1: 69.94% -> 79.12% | Improved |
| add motif prior (ablation) | MAST (w/o MCTS) · QM9S | MCES: 0.9783 -> 0.6296 | Improved |
| add motif prior (ablation) | MAST (w/o MCTS) · QM9S | Motif macro-F1: 0.9087 -> 0.9430 | Improved |
| add MCTS tree search (ablation) | MAST · QM9S | Acc@1: 79.12% -> 94.89% | Improved |
| add MCTS tree search (ablation) | MAST · QM9S | MCES: 0.6296 -> 0.1048 | Improved |
| fixed guidance only instead of MCTS (ablation) | MAST · QM9S | Acc@1: 79.12% -> 79.32% | No clear effect |
| scheduled guidance only instead of MCTS (ablation) | MAST · QM9S | Acc@1: 79.12% -> 79.96% | No clear effect |
| use softmax backup instead of max backup (ablation) | MCTS · QM9S | Acc@1: 94.92% -> 95.70% | Improved |
| use softmax backup instead of mean backup (ablation) | MCTS · QM9S | Acc@1: 94.53% -> 95.70% | Improved |
| MCTS vs Best-of-N at about 2000 denoising steps | MAST · QM9S | Acc@1: 86.13% -> 94.34% | Improved |
| add Gaussian noise SNR=30 dB to input spectra | MAST (w/o MCTS) · QM9S | Acc@1: 79.12% -> 75.98% | Hurt |
| add baseline drift=5 to input spectra | MAST (w/o MCTS) · QM9S | Acc@1: 79.12% -> 74.61% | Hurt |
Closest papers by meaning
- DiffMS: Diffusion Generation of Molecules Conditioned on Mass Spectra (ICML 2025 · similarity 0.89)
- Atomic Diffusion Models for Small Molecule Structure Elucidation from NMR Spectra (NeurIPS 2025 · similarity 0.87)
- MS-BART: Unified Modeling of Mass Spectra and Molecules for Structure Elucidation (NeurIPS 2025 · similarity 0.87)
- Unraveling Molecular Structure: A Multimodal Spectroscopic Dataset for Chemistry (NeurIPS 2024 · similarity 0.87)
- DMol: A Highly Efficient and Chemical Motif-Preserving Molecule Generation Platform (NeurIPS 2025 · similarity 0.86)
- Equivariant Blurring Diffusion for Hierarchical Molecular Conformer Generation (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.
