Bayesian workflow diagnostics outperform unit tests for detecting and repairing statistically misspecified LLM-generated probabilistic programs across benchmarks and real generation tasks.
hub
Journal of the American statistical Association , volume=
13 Pith papers cite this work. Polarity classification is still indexing.
13
Pith papers citing it
hub tools
years
2026 13representative citing papers
Learned functional perturbations plus CRPS training convert deterministic ML interatomic potentials into probabilistic ones, improving CRPS by 19-32% on N-body benchmarks and uncertainty-error correlation from 0.75 to 0.84 on silica.
A Bregman divergence approach yields a general calibeating framework that achieves U-calibration with logarithmic regret for Tsallis losses and a new regret equality for Be The Regularized Leader.
Risk-controlled post-processing yields a threshold-structured policy that follows the baseline except where an oracle fallback sharply reduces conditional violation risk, achieving O(log n/n) expected excess risk in i.i.d. settings and exact risk control under exchangeability.
PODiff performs conditional diffusion in a fixed, variance-ordered POD latent space to enable efficient probabilistic super-resolution of high-dimensional scientific fields with lower memory and better-calibrated uncertainty than pixel-space or dropout baselines.
A new evaluation framework shows that blood glucose forecasting models with high overall accuracy often fail at timely hypoglycemia detection in high-risk periods and at predicting effects of changed insulin doses.
Laplace approximation framework for quantile regression with mixed-effects and Gaussian processes using Fisher information and population curvature of expected loss instead of observed Hessian.
DCMA learns conditional generative models from observational data to reconstruct interventional outcome distributions via Monte Carlo simulation and derive error bounds for distributional mediation effects.
Predictive Bayesian inference posteriors concentrate onto a forward-model-dependent quantity and produce miscalibrated credible sets unless the predictive model contains the true data-generating process.
REX-SUB combines a randomized exchange algorithm with Vecchia approximation to choose subsamples that minimize mean squared prediction error and interval scores in large-scale spatial GPs.
Quantile-based trading strategies for battery arbitrage fail to incentivize honest probabilistic forecasts and ignore price dependence, while stochastic programs using full distributions better connect forecast accuracy to economic value.
AlphaEarth embeddings improve out-of-region EMS point-process forecasts 2-6x at 1-2 week histories and 10-20% at longer histories compared to event-only baselines.
The work introduces WaLeF/FIDLAr for flood forecasting, CoDiCast for probabilistic weather, and Hypercube-RAG for explainable environmental QA, claiming superior accuracy, efficiency, and interpretability over baselines.
citing papers explorer
-
Calibration, Not Compilation: Detecting and Repairing Misspecified Probabilistic Programs Written by Language Models
Bayesian workflow diagnostics outperform unit tests for detecting and repairing statistically misspecified LLM-generated probabilistic programs across benchmarks and real generation tasks.
-
Uncertainty-aware Machine Learning Interatomic Potentials via Learned Functional Perturbations
Learned functional perturbations plus CRPS training convert deterministic ML interatomic potentials into probabilistic ones, improving CRPS by 19-32% on N-body benchmarks and uncertainty-error correlation from 0.75 to 0.84 on silica.
-
Calibeating for general proper losses: A Bregman divergence approach
A Bregman divergence approach yields a general calibeating framework that achieves U-calibration with logarithmic regret for Tsallis losses and a new regret equality for Be The Regularized Leader.
-
Risk-Controlled Post-Processing of Decision Policies
Risk-controlled post-processing yields a threshold-structured policy that follows the baseline except where an oracle fallback sharply reduces conditional violation risk, achieving O(log n/n) expected excess risk in i.i.d. settings and exact risk control under exchangeability.
-
PODiff: Latent Diffusion in Proper Orthogonal Decomposition Space for Scientific Super-Resolution
PODiff performs conditional diffusion in a fixed, variance-ordered POD latent space to enable efficient probabilistic super-resolution of high-dimensional scientific fields with lower memory and better-calibrated uncertainty than pixel-space or dropout baselines.
-
From Prediction to Practice: A Task-Aware Evaluation Framework for Blood Glucose Forecasting
A new evaluation framework shows that blood glucose forecasting models with high overall accuracy often fail at timely hypoglycemia detection in high-risk periods and at predicting effects of changed insulin doses.
-
Laplace Approximations for Mixed-Effects and Gaussian Process Quantile Regression
Laplace approximation framework for quantile regression with mixed-effects and Gaussian processes using Fisher information and population curvature of expected loss instead of observed Hessian.
-
Distributional Causal Mediation via Conditional Generative Modeling
DCMA learns conditional generative models from observational data to reconstruct interventional outcome distributions via Monte Carlo simulation and derive error bounds for distributional mediation effects.
-
Concentration and Calibration in Predictive Bayesian Inference
Predictive Bayesian inference posteriors concentrate onto a forward-model-dependent quantity and produce miscalibrated credible sets unless the predictive model contains the true data-generating process.
-
REX-SUB: A Scalable Subsampling Strategy for Modeling Large Spatial Datasets
REX-SUB combines a randomized exchange algorithm with Vecchia approximation to choose subsamples that minimize mean squared prediction error and interval scores in large-scale spatial GPs.
-
Probabilistic Forecasting for Day-ahead Electricity Prices, Battery Trading Strategies and the Economic Evaluation of Predictive Accuracy
Quantile-based trading strategies for battery arbitrage fail to incentivize honest probabilistic forecasts and ignore price dependence, while stochastic programs using full distributions better connect forecast accuracy to economic value.
-
When Context Compensates for Sparse Event History: AlphaEarth for Spatio-Temporal Point-Process Forecasting
AlphaEarth embeddings improve out-of-region EMS point-process forecasts 2-6x at 1-2 week histories and 10-20% at longer histories compared to event-only baselines.
-
Accurate, Efficient, and Explainable Deep Learning Approaches for Environmental Science Problems
The work introduces WaLeF/FIDLAr for flood forecasting, CoDiCast for probabilistic weather, and Hypercube-RAG for explainable environmental QA, claiming superior accuracy, efficiency, and interpretability over baselines.