SPAMoE framework enhances full-waveform inversion with spectrum-aware neural operators

Researchers have developed SPAMoE, a novel framework designed to improve the efficiency and accuracy of full-waveform inversion (FWI) for subsurface velocity model reconstruction. This approach addresses the challenge of frequency entanglement in multi-scale geological features by incorporating a spectral-preserving encoder and a dynamic routing mechanism for a Mixture-of-Experts ensemble. Experiments on the OpenFWI datasets demonstrated that SPAMoE significantly reduces the mean absolute error compared to existing baselines, establishing a new architectural framework for learning-based FWI. AI

Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →

IMPACT Introduces a novel framework for inverse problems, potentially improving subsurface imaging accuracy and efficiency.

RANK_REASON This is a research paper detailing a new framework for a specific scientific problem. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• SPAMoE
• OpenFWI
• CNN
• MoE
• FNO
• MNO
• LNO
• arXiv
• Zhenyu Wang

• paper
• other