LatentPDE reconstructs sparse scientific measurements by representing the latent space of a diffusion model as the coefficients and source terms of an assumed governing PDE.
We initialize the background xb ∈R H×W using MAP or encoder prediction, matching the 3D-Var setup
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Learning Interpretable PDE Representations for Generative Reconstructions with Structured Sparsity
LatentPDE reconstructs sparse scientific measurements by representing the latent space of a diffusion model as the coefficients and source terms of an assumed governing PDE.