Pith. sign in

REVIEW 5 cited by

Learning Likelihood Ratios with Neural Network Classifiers

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 2305.10500 v2 pith:DBZVCNI3 submitted 2023-05-17 hep-ph physics.data-anstat.APstat.ML

classification hep-phphysics.data-anstat.APstat.ML
keywords likelihoodrationeuralapproximatingclassifiersdistributionsnetworkparametrizations
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

The likelihood ratio is a crucial quantity for statistical inference in science that enables hypothesis testing, construction of confidence intervals, reweighting of distributions, and more. Many modern scientific applications, however, make use of data- or simulation-driven models for which computing the likelihood ratio can be very difficult or even impossible. By applying the so-called ``likelihood ratio trick,'' approximations of the likelihood ratio may be computed using clever parametrizations of neural network-based classifiers. A number of different neural network setups can be defined to satisfy this procedure, each with varying performance in approximating the likelihood ratio when using finite training data. We present a series of empirical studies detailing the performance of several common loss functionals and parametrizations of the classifier output in approximating the likelihood ratio of two univariate and multivariate Gaussian distributions as well as simulated high-energy particle physics datasets.

Discussion (0). Sign in to comment.

Forward citations

Cited by 5 Pith papers

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

  1. HDSense: An efficient method for ranking observable sensitivity

    hep-ph 2026-02 conditional novelty 6.0 of 10

    HDSense ranks observable subsets by adding per-observable Fisher information and penalizing overlap, picking near-optimal sets for Pythia hadronization parameters in tested cases.

  2. Higgs Signal Strength Estimation with Machine Learning under Systematic Uncertainties

    hep-ph 2025-08 conditional novelty 6.0 of 10

    SAGE, a dual-branch GNN trained under nuisance fluctuations, estimates the Higgs signal strength with near-nominal coverage (0.662-0.683) but wider intervals than the top FAIR-HUC leaderboard methods.

  3. High-Dimensional Unfolding in Large Backgrounds

    hep-ph 2025-07 conditional novelty 6.0 of 10

    OmniFold-HI, an ML unfolding algorithm that handles large backgrounds and high-dimensional auxiliary observables, is derived, shown equivalent to iterative Bayesian unfolding, and demonstrated to improve jet-substruct...

  4. Analysis note: measurement of thrust in $e^{+}e^{-}$ collisions at $\sqrt{s}$ = 91 GeV with archived ALEPH data

    hep-ex 2025-07 conditional novelty 5.0 of 10

    Using archived ALEPH data, the authors produce a detector-corrected thrust distribution with machine-learning unbinned unfolding that matches the old ALEPH result and adds flexible per-event weights.

  5. An Introduction to Bayesian and Frequentist Simulation-Based Inference with Machine Learning

    cs.LG 2026-07 accept novelty 2.0 of 10

    A structured introduction to ML-based simulation-based inference, contrasting Bayesian and frequentist frameworks and covering parameter inference, unfolding, and validation.

Pith tools