Supervised Guidance Training enables conditioning of infinite-dimensional diffusion models via an extended Doob h-transform so that fine-tuned models accurately sample from posteriors in function space.
Springer, Cham, 2017
10 Pith papers cite this work, alongside 230 external citations. Polarity classification is still indexing.
10
Pith papers citing it
230
external citations · Crossref
citation-role summary
background 1
citation-polarity summary
roles
background 1polarities
background 1representative citing papers
A finite-sample perspective reveals that inexact likelihood approximations cause under- or over-estimation of posterior spread at intermediate timesteps, leading to early-stopping sensitivity, mode weighting errors, and hallucinations even from multimodal priors alone.
Presents a likelihood-free transport map learned by minimizing an averaged energy-distance objective to amortize Bayesian inference for inverse problems, including PDE-constrained cases with neural operator representations.
GRIFDIR proposes graph resolution-invariant FEM diffusion models that maintain resolution invariance and high fidelity on complex irregular domains.
Augmented Krylov subspaces jointly approximate quadratic forms and log-dets for faster MLE-based hyperparameter tuning in kernel-based linear system identification.
An amortized variational framework jointly targets the posterior and posterior-predictive distributions via a KL upper bound and moment regularization, yielding more accurate predictions at lower online cost than two-stage variational inference.
A PINN transfer learning framework for coal methane sorption reaches R²=0.932 on held-out data with 227% improvement over classical isotherms and identifies Monte Carlo Dropout as the best uncertainty method while ensembles degrade under shared physics constraints.
An infinite-dimensional Bayesian framework estimates seabed topography and roughness simultaneously from acoustic data by assuming statistical isotropy and using fractional differentiability.
New dimension and model reduction techniques for linear Bayesian inverse problems with rank-deficient priors, with approximation guarantees and efficiency demonstrations for high-dimensional inference.
Interpolation-based ROM techniques with Q-DEIM hyper-reduction are applied to reduce computational cost and memory use of stochastic integrals in the SFV method for high-dimensional stochastic spaces.
citing papers explorer
-
Supervised Guidance Training for Infinite-Dimensional Diffusion Models
Supervised Guidance Training enables conditioning of infinite-dimensional diffusion models via an extended Doob h-transform so that fine-tuned models accurately sample from posteriors in function space.
-
When, why, and how do diffusion posterior samplers fail? A finite-sample lens
A finite-sample perspective reveals that inexact likelihood approximations cause under- or over-estimation of posterior spread at intermediate timesteps, leading to early-stopping sensitivity, mode weighting errors, and hallucinations even from multimodal priors alone.
-
Amortized Energy-Based Bayesian Inference
Presents a likelihood-free transport map learned by minimizing an averaged energy-distance objective to amortize Bayesian inference for inverse problems, including PDE-constrained cases with neural operator representations.
-
GRIFDIR: Graph Resolution-Invariant FEM Diffusion Models in Function Spaces over Irregular Domains
GRIFDIR proposes graph resolution-invariant FEM diffusion models that maintain resolution invariance and high fidelity on complex irregular domains.
-
Kernel-based linear system identification using augmented Krylov subspaces
Augmented Krylov subspaces jointly approximate quadratic forms and log-dets for faster MLE-based hyperparameter tuning in kernel-based linear system identification.
-
Amortized Variational Inference for Joint Posterior and Predictive Distributions in Bayesian Uncertainty Quantification
An amortized variational framework jointly targets the posterior and posterior-predictive distributions via a KL upper bound and moment regularization, yielding more accurate predictions at lower online cost than two-stage variational inference.
-
Physics-Informed Neural Networks for Methane Sorption: Cross-Gas Transfer Learning, Ensemble Collapse Under Physics Constraints, and Monte Carlo Dropout Uncertainty Quantification
A PINN transfer learning framework for coal methane sorption reaches R²=0.932 on held-out data with 227% improvement over classical isotherms and identifies Monte Carlo Dropout as the best uncertainty method while ensembles degrade under shared physics constraints.
-
Simultaneous Estimation of Seabed and Its Roughness With Longitudinal Waves
An infinite-dimensional Bayesian framework estimates seabed topography and roughness simultaneously from acoustic data by assuming statistical isotropy and using fractional differentiability.
-
Dimension and model reduction approaches for linear Bayesian inverse problems with rank-deficient prior covariances
New dimension and model reduction techniques for linear Bayesian inverse problems with rank-deficient priors, with approximation guarantees and efficiency demonstrations for high-dimensional inference.
-
Model Order Reduction Techniques for the Stochastic Finite Volume Method
Interpolation-based ROM techniques with Q-DEIM hyper-reduction are applied to reduce computational cost and memory use of stochastic integrals in the SFV method for high-dimensional stochastic spaces.