SCOPE Before You Sample: Conformal Scores Guide Posterior Exploration for Fast and Calibrated Bayesian Predictive Inference

Key points

• Bayesian predictive regions account for parameter uncertainty but can be poorly calibrated under model misspecification, even with expensive full MCMC.1
• SCOPE gives nonconformity scores a new role in guiding posterior computation: score gradients on a held-out guide set, weighted toward the target boundary, identify the relevant directions.1
• We establish asymptotic coverage for SCOPE and provide a split variant with finite-sample validity.1
• Controlled simulations show that score-guided exploration yields more accurate calibrated regions than non-targeted direction choices at low active rank.1

Sources (1)

• [1]SCOPE Before You Sample: Conformal Scores Guide Posterior Exploration for Fast and Calibrated Bayesian Predictive Inference
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 12, 05:27 PM

  We propose SCOPE (Score-guided Conformal Posterior Exploration), which retains the predictive quality of full MCMC while achieving calibrated coverage at a fraction of its cost.

  Bayesian predictive regions account for parameter uncertainty but can be poorly calibrated under model misspecification, even with expensive full MCMC.

Extractive summary: sentences quoted from the sources.