Pith. sign in

REVIEW 4 cited by

Bayesian Experimental Design via Contrastive Diffusions

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2410.11826 v2 pith:SWULWTAT submitted 2024-10-15 stat.ML cs.LG

Bayesian Experimental Design via Contrastive Diffusions

classification stat.ML cs.LG
keywords boeddesignmaximizationposteriorbayesiancontrastexpectedexperimental
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
read the original abstract

Bayesian Optimal Experimental Design (BOED) is a powerful tool to reduce the cost of running a sequence of experiments. When based on the Expected Information Gain (EIG), design optimization corresponds to the maximization of some intractable expected contrast between prior and posterior distributions. Scaling this maximization to high dimensional and complex settings has been an issue due to BOED inherent computational complexity. In this work, we introduce a pooled posterior distribution with cost-effective sampling properties and provide a tractable access to the EIG contrast maximization via a new EIG gradient expression. Diffusion-based samplers are used to compute the dynamics of the pooled posterior and ideas from bi-level optimization are leveraged to derive an efficient joint sampling-optimization loop. The resulting efficiency gain allows to extend BOED to the well-tested generative capabilities of diffusion models. By incorporating generative models into the BOED framework, we expand its scope and its use in scenarios that were previously impractical. Numerical experiments and comparison with state-of-the-art methods show the potential of the approach.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 4 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Bayesian Experimental Design via Score Matching

    stat.ML 2026-07 conditional novelty 7.0

    SCOREBED isolates EIG double intractability in a policy-independent score-matching stage, then trains design policies with a singly intractable gradient estimator, enabling cheap multi-policy selection.

  2. Action-BED: Task-Driven Bayesian Experimental Design with Singly Intractable Objectives

    stat.ML 2026-06 unverdicted novelty 7.0

    Action-BED recasts BED as expected future loss on actions, producing singly intractable objectives jointly optimized for design and action policies via stochastic gradients without explicit posterior estimation.

  3. FairBED: A Bayesian Experimental Design Approach to Gathering Fairer Data

    stat.ML 2026-06 unverdicted novelty 7.0

    FairBED quantifies dataset fairness as uninformative about sensitive attributes and uses fairness-aware BED to gather data yielding better fairness-accuracy trade-offs than random or standard BED acquisition.

  4. Bayesian experimental design: grouped geometric pooled posterior via ensemble Kalman methods

    cs.IT 2026-04 unverdicted novelty 6.0

    A grouped pooling strategy with ensemble Kalman inversion improves accuracy of expected information gain estimators in Bayesian experimental design at amortized computational cost.