mlsanityData to 7 Oct 2026

Contrastive Neural Ratio Estimation for Simulation-based Inference

arXiv 2210.06170 ↗First posted 11 Oct 2022stat.ML astro-ph.IM cs.LGcited by 7

Not on the map. Why?

Abstract (arXiv)

Likelihood-to-evidence ratio estimation is usually cast as either a binary (NRE-A) or a multiclass (NRE-B) classification task. In contrast to the binary classification framework, the current formulation of the multiclass version has an intrinsic and unknown bias term, making otherwise informative diagnostics unreliable. We propose a multiclass framework free from the bias inherent to NRE-B at optimum, leaving us in the position to run diagnostics that practitioners depend on. It also recovers NRE-A in one corner case and NRE-B in the limiting case. For fair comparison, we benchmark the behavior of all algorithms in both familiar and novel training regimes: when jointly drawn data is unlimited, when data is fixed but prior draws are unlimited, and in the commonplace fixed data and parameters setting. Our investigations reveal that the highest performing models are distant from the competitors (NRE-A, NRE-B) in hyperparameter space. We make a recommendation for hyperparameters distinct from the previous models. We suggest two bounds on the mutual information as performance metrics for simulation-based inference methods, without the need for posterior samples, and provide experimental results. This version corrects a minor implementation error in γ, improving results.

Key facts. Cited by 7 papers in the map (data through 7 Oct 2026). Builds on 39 of its 89 references that are in the map. First posted on arXiv 11 Oct 2022.

Builds on

39 of 89 references.

References · 39

  1. 2015Deep Residual Learning for Image Recognitioncited by 51,809
  2. 2019PyTorch: An Imperative Style, High-Performance Deep Learning Librarycited by 20,160
  3. 2018Representation Learning with Contrastive Predictive Codingcited by 7,662
  4. 2015Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shiftcited by 6,979
  5. 2022Generative Adversarial Networkscited by 5,603
  6. 2013Distributed Representations of Words and Phrases and their Compositionalitycited by 3,462
  7. 2019Normalizing Flows for Probabilistic Modeling and Inferencecited by 1,315
  8. 2017Masked Autoregressive Flow for Density Estimationcited by 680
  9. 2019On Variational Bounds of Mutual Informationcited by 550
  10. 2019Quantifying the Carbon Emissions of Machine Learningcited by 516
  11. 2019The frontier of simulation-based inferencecited by 490
  12. 2016Revisiting Classifier Two-Sample Testscited by 258
27 more
  1. 2018Sequential Neural Likelihood: Fast Likelihood-free Inference with Autoregressive Flowscited by 217
  2. 2019Automatic Posterior Transformation for Likelihood-Free Inferencecited by 216
  3. 2017Flexible statistical inference for mechanistic models of neural dynamicscited by 162
  4. 2021Benchmarking Simulation-Based Inferencecited by 160
  5. 2016Learning in Implicit Generative Modelscited by 124
  6. 2018Validating Bayesian Inference Algorithms with Simulation-Based Calibrationcited by 118
  7. 2018Noise Contrastive Estimation and Negative Sampling for Conditional Models: Consistency and Statistical Efficiencycited by 103
  8. 2020On Contrastive Learning for Likelihood-free Inferencecited by 96
  9. 2015Approximating Likelihood Ratios with Calibrated Discriminative Classifierscited by 87
  10. 2016Likelihood-free inference by ratio estimationcited by 80
  11. 2015Bayesian Optimization for Likelihood-Free Inference of Simulator-Based Statistical Modelscited by 61
  12. 2018Likelihood-free inference with emulator networkscited by 61
  13. 2018Mining gold from implicit models to improve likelihood-free inferencecited by 60
  14. 2022Variational methods for simulation-based inferencecited by 43
  15. 2022GATSBI: Generative Adversarial Training for Simulation-Based Inferencecited by 36
  16. 2018A Likelihood-Free Inference Framework for Population Genetic Data using Exchangeable Neural Networkscited by 30
  17. 2021Truncated Marginal Neural Ratio Estimationcited by 30
  18. 2018Constraining Effective Field Theories with Machine Learningcited by 26
  19. 2020Solving high-dimensional parameter inference: marginal posterior densities & Moment Networkscited by 22
  20. 2014High-Dimensional Density Ratio Estimation with Extensions to Approximate Likelihood Computationcited by 19
  21. 2019Parallel Gaussian process surrogate Bayesian inference with noisy likelihood evaluationscited by 18
  22. 2020Confidence Sets and Hypothesis Testing in a Likelihood-Free Inference Settingcited by 17
  23. 2020Towards constraining warm dark matter with stellar streams through neural simulation-based inferencecited by 14
  24. 2021Fast and Credible Likelihood-Free Cosmology with Truncated Marginal Neural Ratio Estimationcited by 13
  25. 2019Likelihood-free approximate Gibbs samplingcited by 7
  26. 2020Simulation-efficient marginal posterior estimation with swyft: stop wasting your precious timecited by 7
  27. 2022Statistical applications of contrastive learningcited by 5

Built on by

Cited by year

Citing papers per year: 2023: 6; 2024: 0; 2025: 1.

7 papers.

  1. 2025Simulation-Based Inference: A Practical Guidecited by 30
  2. 2023Learning Likelihood Ratios with Neural Network Classifierscited by 11
  3. 2023Balancing Simulation-based Inference for Conservative Posteriorscited by 10
  4. 2023Calibrating Neural Simulation-Based Inference with Differentiable Coverage Probabilitycited by 10
  5. 2023Hierarchical Neural Simulation-Based Inference Over Event Ensemblescited by 8
  6. 2023Simulation-based Inference with the Generalized Kullback-Leibler Divergencecited by 3
  7. 2023Pseudo-Likelihood Inferencecited by 0