Tag
The paper introduces OpDis-PS, a diffusion posterior sampler for electrocardiographic imaging that models structured, signal-dependent forward-operator discrepancies (e.g., lung conductivity changes and subcutaneous fat) instead of treating them as noise, outperforming classical regularization and learned predictors especially on held-out cardiac pathologies.
该论文研究了Hilbert尺度上线性逆问题的后验扩散先验,在显式分数放大条件下推导出稳定性理论与非渐近恢复界,给出由算子平滑性、真值正则性与分数误差共同决定的可恢复正则性阈值。
This paper introduces Christoffel-DPS, a distribution-free framework for optimal sensor placement in diffusion posterior sampling that outperforms classical Gaussian-based methods. It provides theoretical guarantees and practical improvements for reconstructing states from complex, non-Gaussian distributions using generative models.