Research map / Probabilistic and scientific modeling / Molecules and proteins
$\mathbf{\mathbb{E}^{FWI}}$: Multiparameter Benchmark Datasets for Elastic Full Waveform Inversion of Geophysical Properties
Shihang Feng, Hanchen Wang, Chengyuan Deng, Yinan Feng, Yanhua Liu, Min Zhu, Peng Jin, Yinpeng Chen, Youzuo Lin
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 |
|---|---|---|---|
| predict Vp and Vs independently with two InversionNets instead of jointly (ablation) | InversionNet · EFWI E^FVA | Pr SSIM: 2% less (l1), 6% less (l2) | Hurt |
| predict Vp and Vs independently instead of jointly (ablation) | InversionNet · EFWI E^FVA | Pr MAE and MSE: 60%, 72% | Hurt |
| test Vp-only InversionNet on data with altered Vs structure (ablation) | InversionNet · EFWI E^FVA | MAE, RMSE, SSIM: 343% MAE, 296% RMSE, 7% SSIM drop | Hurt |
| test Vs-only InversionNet on data with altered Vp structure (l1) (ablation) | InversionNet · EFWI E^CFB | MAE, RMSE, SSIM: 120% MAE, 89% RMSE, 7% SSIM drop | Hurt |
| test Vs-only InversionNet on data with altered Vp structure (l2) (ablation) | InversionNet · EFWI E^CFB | MAE, RMSE, SSIM: 115% MAE, 88% RMSE, 7% SSIM drop | Hurt |
| move from version A to harder version B dataset | ElasticNet · EFWI Vel family | Vp MAE (l1): 0.0308 -> 0.0668 | Hurt |
| move from version A to harder version B dataset | ElasticNet · EFWI CurveFault family | Vp MAE (l1): 0.0551 -> 0.1535 | Hurt |
| use l2 loss instead of l1 loss (ablation) | ElasticNet · EFWI E^FVA | Vp MAE: 0.0308 -> 0.0235 | Improved |
| use l2 loss instead of l1 loss (ablation) | ElasticNet · EFWI E^FVB | Vp MAE: 0.0668 -> 0.1016 | Hurt |
| use l2 loss instead of l1 loss (ablation) | ElasticNet · EFWI E^CVB | Vp MAE: 0.1722 -> 0.1682 | No clear effect |
| use l2 loss instead of l1 loss (ablation) | ElasticNet · EFWI E^CFB | Pr SSIM: 0.3576 -> 0.4490 | Improved |
| use ElasticTransformer instead of ElasticNet (training cost) | ElasticNet · EFWI E^Vel family | training time per epoch: 4m15s -> 1m15s | Improved |
Closest papers by meaning
- A Unified Framework for Forward and Inverse Problems in Subsurface Imaging using Latent Space Translations (ICLR 2025 · similarity 0.85)
- Auto-Linear Phenomenon in Subsurface Imaging (ICML 2024 · similarity 0.84)
- A Novel Diffusion Model for Pairwise Geoscience Data Generation with Unbalanced Training Dataset (AAAI 2025 · similarity 0.82)
- Broadband Ground Motion Synthesis by Diffusion Model with Minimal Condition (ICML 2025 · similarity 0.82)
- InverseBench: Benchmarking Plug-and-Play Diffusion Priors for Inverse Problems in Physical Sciences (ICLR 2025 · similarity 0.81)
- Symmetric Basis Convolutions for Learning Lagrangian Fluid Mechanics (ICLR 2024 · similarity 0.81)
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.
