A two-step Bayesian PINN that sets its likelihood variance to a residual error bound improves calibration on cosmological ODEs inside the training region, but is unstable outside it.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Improved Uncertainty Quantification in Physics-Informed Neural Networks Using Error Bounds and Solution Bundles
A two-step Bayesian PINN that sets its likelihood variance to a residual error bound improves calibration on cosmological ODEs inside the training region, but is unstable outside it.