Research map / Image generation / Autoregressive and latent generation
Learning Flexible Forward Trajectories for Masked Molecular Diffusion
Hyunjin Seo, Taewon Kim, Sihyun Yu, Sungsoo Ahn
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 learnable element-wise noise schedule | masked diffusion (power-law) · ZINC250K unconditional | FCD: 26.09 -> 1.51 | Improved |
| add learnable element-wise noise schedule | masked diffusion (power-law) · ZINC250K unconditional | scaffold similarity: 0.001 -> 0.560 | Improved |
| add learnable element-wise noise schedule | masked diffusion (power-law) · QM9 unconditional | NSPDK: 0.007 -> 0.0002 | Improved |
| add learnable element-wise noise schedule | masked diffusion (power-law) · Polymers property-conditioned | validity (%): 17.31 -> 99.10 | Improved |
| add learnable element-wise noise schedule | masked diffusion (power-law) · Polymers property-conditioned | property MAE: 1.620 -> 0.798 | Improved |
| use MELD instead of GraphDiT | GraphDiT · Polymers property-conditioned | property MAE: 0.921 -> 0.798 | Improved |
| use MELD instead of GraphDiT | GraphDiT · Polymers property-conditioned | diversity: 87.12 -> 85.91 | Hurt |
| learn node-wise schedule only vs full node+edge schedule (ablation) | MELD · ZINC250K | FCD: 1.63 -> 1.51 | Improved |
| learn edge-wise schedule only vs full node+edge schedule (ablation) | MELD · ZINC250K | FCD: 1.73 -> 1.51 | Improved |
| element-wise schedule vs class-wise learnable schedule (GenMD4) (ablation) | MELD · ZINC250K | FCD: 3.19 -> 1.51 | Improved |
| element-wise schedule vs fixed spindle schedule (DiffusionBERT) (ablation) | MELD · ZINC250K | FCD: 1.95 -> 1.51 | Improved |
| element-wise learnable schedule increases distinct intermediate states (ablation) | MELD · ZINC250K (131 molecules) | unique graph states at T-25: 126.0 -> 131.0 | Improved |
Closest papers by meaning
- DiffMS: Diffusion Generation of Molecules Conditioned on Mass Spectra (ICML 2025 · similarity 0.90)
- Geometric-Facilitated Denoising Diffusion Model for 3D Molecule Generation (AAAI 2024 · similarity 0.90)
- Graph Diffusion that can Insert and Delete (NeurIPS 2025 · similarity 0.90)
- All-atom Diffusion Transformers: Unified generative modelling of molecules and materials (ICML 2025 · similarity 0.89)
- Graph Diffusion Transformers for Multi-Conditional Molecular Generation (NeurIPS 2024 · similarity 0.89)
- Steering Masked Discrete Diffusion Models via Discrete Denoising Posterior Prediction (ICLR 2025 · similarity 0.89)
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.
