PSEF trains a permutation-invariant transformer analysis map on synthetic state-observation pairs using strictly proper scoring rules to approximate the true Bayesian filter, with a proof under realizability.
Title resolution pending
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
years
2026 2verdicts
UNVERDICTED 2representative citing papers
Introduces power-law, logistic, and discrepancy-based tapers for correlation-based localization that suppress spurious correlations and often preserve more posterior ensemble variance than distance-based methods in synthetic reservoir assimilation tests.
citing papers explorer
-
Learning Probabilistic Filters with Strictly Proper Scoring Rules
PSEF trains a permutation-invariant transformer analysis map on synthetic state-observation pairs using strictly proper scoring rules to approximate the true Bayesian filter, with a proof under realizability.
-
Statistical Tapers for Correlation-Based Localization in Ensemble Data Assimilation
Introduces power-law, logistic, and discrepancy-based tapers for correlation-based localization that suppress spurious correlations and often preserve more posterior ensemble variance than distance-based methods in synthetic reservoir assimilation tests.