arXiv：cs.LG· Shiqin Zeng, Yunlin Zeng, Abhinav Prakash Gahlot, Zijun Deng, Felix J. Herrmann·· 4 小时前AI 评分24
自注意力摘要网络：从共成像点道集构建地下速度模型
Self-attention summary networks for subsurface velocity-model building from common-image gathers
AI 导读
研究提出一种多尺度自注意力摘要网络，将高维 3D 共成像点道集（CIG）体映射为紧凑条件嵌入，用于概率性地下速度反演。该嵌入保留随偏移距变化的运动学结构与空间相干性，配合 flow-matching 模型从高斯分布输运至速度场后验分布。数值实验中，相比直接以原始 CIG 为条件，该网络提升了后验速度推断，多尺度注意力设计在背景速度模型失配下更稳健，后验重建更准确且预测不确定性更低。
正文
Abstract:Common-image gathers (CIGs) contain physically meaningful information about velocity-model errors through reflector focusing and residual moveout, but in conventional imaging workflows they are typically used only as diagnostic tools. In this work, we propose a multiscale self-attention summary network that maps high-dimensional 3D CIG volumes into compact conditioning embeddings for probabilistic subsurface velocity inversion. These learned embeddings preserve offset-dependent kinematic structure and spatial coherence while reducing variability caused by background-velocity mismatch. Conditioned on these summary embeddings, a flow-matching model learns a transport from a Gaussian source distribution to the posterior distribution of plausible velocity fields. Numerical experiments show that, compared with direct conditioning on raw CIGs, the proposed summary network improves posterior velocity inference. In particular, the multiscale attention design provides greater robustness to background-model mismatch, yielding more accurate posterior reconstructions and lower predictive uncertainty.
| Subjects: | Machine Learning (cs.LG); Geophysics (physics.geo-ph) |
| Cite as: | arXiv:2610.09282 [cs.LG] |
| (or arXiv:2610.09282v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2610.09282 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Shiqin Zeng [view email]
[v1]
Wed, 7 Oct 2026 01:31:15 UTC (6,871 KB)
来源：arXiv：cs.LG · arxiv.org