Pith. sign in

REVIEW 7 cited by

Machine learning assisted Bayesian model comparison: learnt harmonic mean estimator

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 2111.12720 v3 pith:TSOGHXH5 submitted 2021-11-24 stat.ME astro-ph.IMstat.CO

Machine learning assisted Bayesian model comparison: learnt harmonic mean estimator

classification stat.ME astro-ph.IMstat.CO
keywords estimatorharmonicmeanlearntoriginalvariancebayesianlarge
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
Share X Bluesky LinkedIn Reddit HN
abstract

We resurrect the infamous harmonic mean estimator for computing the marginal likelihood (Bayesian evidence) and solve its problematic large variance. The marginal likelihood is a key component of Bayesian model selection to evaluate model posterior probabilities; however, its computation is challenging. The original harmonic mean estimator, first proposed by Newton and Raftery in 1994, involves computing the harmonic mean of the likelihood given samples from the posterior. It was immediately realised that the original estimator can fail catastrophically since its variance can become very large (possibly not finite). A number of variants of the harmonic mean estimator have been proposed to address this issue although none have proven fully satisfactory. We present the \emph{learnt harmonic mean estimator}, a variant of the original estimator that solves its large variance problem. This is achieved by interpreting the harmonic mean estimator as importance sampling and introducing a new target distribution. The new target distribution is learned to approximate the optimal but inaccessible target, while minimising the variance of the resulting estimator. Since the estimator requires samples of the posterior only, it is agnostic to the sampling strategy used. We validate the estimator on a variety of numerical experiments, including a number of pathological examples where the original harmonic mean estimator fails catastrophically. We also consider a cosmological application, where our approach leads to $\sim$ 3 to 6 times more samples than current state-of-the-art techniques in 1/3 of the time. In all cases our learnt harmonic mean estimator is shown to be highly accurate. The estimator is computationally scalable and can be applied to problems of dimension $O(10^3)$ and beyond. Code implementing the learnt harmonic mean estimator is made publicly available

discussion (0)

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

Forward citations

Cited by 7 Pith papers

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

  1. Improving exoplanet mass characterisation with Bayesian model selection using the Learned Harmonic Mean Estimator

    astro-ph.EP 2026-06 unverdicted novelty 7.0

    First use of the learned harmonic mean estimator for Bayesian model selection across circular/eccentric, white-noise/GP, and trend variants in radial velocity exoplanet analyses.

  2. Sample-efficient evidence estimation of score based priors for model selection

    cs.LG 2026-02 unverdicted novelty 7.0

    DiME estimates model evidence for diffusion priors by integrating time-marginals from posterior sampling, enabling efficient prior selection and misfit diagnosis in ill-posed inverse problems.

  3. General Relativistic Entropic Acceleration at the perturbation level: a CLASS implementation and first Boltzmann-code constraints

    gr-qc 2026-07 conditional novelty 6.0

    First full Boltzmann-code implementation of entropic dark energy, with MCMC constraints from CMB+BAO+SN, yields α≈1 and a fit statistically indistinguishable from ΛCDM.

  4. Evolving Dark Energy Is Vacuum Energy After All

    astro-ph.CO 2026-06 unverdicted novelty 6.0

    A QCD-vacuum-based model of dynamical dark energy fits Planck+ACT+SPT, DESI DR2, and supernova data while reproducing the late-time evolution favored by DESI.

  5. A search for periodic AGN variability in $\textit{Gaia}$ Data Release 3

    astro-ph.HE 2025-05 accept novelty 6.0

    Systematic search of 377k Gaia DR3 AGN light curves finds no reliable periodic SMBHB candidates after red-noise modeling and empirical false-alarm testing; all survivors lie in the few-cycle regime.

  6. Evolving Dark Energy Is Vacuum Energy After All

    astro-ph.CO 2026-06 conditional novelty 5.0

    A QCD-vacuum dark-energy model with a two-parameter switch reproduces the DESI-preferred late-time expansion and is mildly favoured over ΛCDM by Bayesian evidence (lnB≈2.7).

  7. Negative neutrino mass or negative dark energy?

    astro-ph.CO 2026-05 unverdicted novelty 5.0

    A sign-switching dark energy model (Λ_s CDM) recovers positive effective neutrino masses (0.055 ± 0.050 eV) consistent with oscillation data, unlike ΛCDM which prefers negative values (-0.075 eV), for DESI DR2 + CMB +...