Valid Inference with Synthetic Data via Task Exchangeability

arXiv:2606.13629 · stat.ME, cs.AI, cs.LG, stat.ML · Submitted 2026-08-21 · Read on arXiv

Lezhi Tan, Tijana Zrnic

stat.ME, cs.AI, cs.LG, stat.ML

Submitted: 2026-08-21

Updated: 2026-08-24

Terminology

• Exchangeability — 85× in this paper · explained in 2 episodes
• Conformal Prediction — 10× in this paper · explained in 5 episodes
• Calibration — 7× in this paper · explained in 7 episodes
• American National Election Studies — 4× in this paper
• LLM-as-a-Judge — 4× in this paper · explained in 3 episodes
• Generative Model — 3× in this paper · explained in 2 episodes
• Generative Models — 3× in this paper · explained in 2 episodes
• Adam — 2× in this paper · explained in 2 episodes
• Covariate Shift — 2× in this paper · explained in 2 episodes
• Adaptivity — 1× in this paper · explained in 1 episode
• Core — 1× in this paper · explained in 2 episodes
• LLM judge — 1× in this paper · explained in 2 episodes
• Pairwise Comparisons — 1× in this paper · explained in 2 episodes
• Pew American Trends Panel — 1× in this paper
• Quantifying Uncertainty — 1× in this paper · explained in 5 episodes
• Robustness — 1× in this paper · explained in 5 episodes

Sources

• PPI++: Efficient Prediction-Powered Inference
• Theoretical Foundations of Conformal Prediction
• General Synthetic-Powered Inference
• Power Analysis for Prediction-Powered Inference
• Multiple-Prediction-Powered Inference
• Prediction-Powered Inference Across Many Tasks for AI Evaluation & Social Science Research
• Predictions as Surrogates: Revisiting Surrogate Outcomes in the Age of AI
• GLM Inference with AI-Generated Synthetic Data Using Misspecified Linear Regression
• Prediction-Powered Inference with Imputed Covariates and Nonuniform Sampling
• Demystifying Prediction Powered Inference
• Imputation-Powered Inference

Related papers

• Doubly robust inference via calibration
• Bayesian Empirical Bayes: Simultaneous Inference from Probabilistic Symmetries
• Flexible Nonparametric Inference for Causal Effects under the Front-Door Model
• Deployment of AI-Assisted Interventions: Capacity Constraints and Noisy Compliance
• A Survey on Archetypal Analysis
• Dynamic Spatial Bayesian Machine Learning Model: Applications to Intergenerational Economic Mobility and Geographic Income Inequality in the United States