Data to 14 Oct 2026

Revisiting Classifier Two-Sample Tests

arXiv 1610.06545 ↗First posted 20 Oct 2016stat.MLcited by 268

Not on the map. Why?

Cited by 32 in Diffusion models, but placed by its own abstract, which didn’t fit.

Abstract (arXiv)

The goal of two-sample tests is to assess whether two samples, SP ~ Pn and SQ ~ Qm, are drawn from the same distribution. Perhaps intriguingly, one relatively unexplored method to build two-sample tests is the use of binary classifiers. In particular, construct a dataset by pairing the n examples in SP with a positive label, and by pairing the m examples in SQ with a negative label. If the null hypothesis "P = Q" is true, then the classification accuracy of a binary classifier on a held-out subset of this dataset should remain near chance-level. As we will show, such Classifier Two-Sample Tests (C2ST) learn a suitable representation of the data on the fly, return test statistics in interpretable units, have a simple null distribution, and their predictive uncertainty allow to interpret where P and Q differ. The goal of this paper is to establish the properties, performance, and uses of C2ST. First, we analyze their main theoretical properties. Second, we compare their performance against a variety of state-of-the-art alternatives. Third, we propose their use to evaluate the sample quality of generative models with intractable likelihoods, such as Generative Adversarial Networks (GANs). Fourth, we showcase the novel application of GANs together with C2ST for causal discovery.

Key facts. Cited by 268 papers in the map (data through 14 Oct 2026). First posted on arXiv 20 Oct 2016.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2021: 32; 2022: 27; 2023: 44; 2024: 35; 2025: 52; 2026: 78.

268 papers.

Sort:

By area:

2021MAUVE: Measuring the Gap Between Neural Text and Human Text using Divergence Frontierscited by 470
2021Large Scale Image Completion via Co-Modulated Generative Adversarial Networkscited by 220
2021Benchmarking Simulation-Based Inferencecited by 162
2024GPT-4V(ision) is a Human-Aligned Evaluator for Text-to-3D Generationcited by 153
2021Testing for Outliers with Conformal p-valuescited by 134
2023Flow Matching for Scalable Simulation-Based Inferencecited by 101
2021Ensembling Off-the-shelf Models for GAN Trainingcited by 86
2023Membership Inference Attacks against Synthetic Data through Overfitting Detectioncited by 66
2023SALAD: Part-Level Latent Diffusion for 3D Shape Generation and Manipulationcited by 66
2021On Memorization in Probabilistic Deep Generative Modelscited by 60
2021On Aliased Resizing and Surprising Subtleties in GAN Evaluationcited by 51
2022Sequential Neural Score Estimation: Likelihood-Free Inference with Conditional Score Based Diffusion Modelscited by 50

38 more

2021CaloFlow: Fast and Accurate Generation of Calorimeter Showers with Normalizing Flowscited by 46
2022Robust Neural Posterior Estimation and Statistical Model Criticismcited by 46
2022Variational methods for simulation-based inferencecited by 43
2024Membership Inference on Text-to-Image Diffusion Models via Conditional Likelihood Discrepancycited by 38
2024Lorentz-Equivariant Geometric Algebra Transformers for High-Energy Physicscited by 38
2022Autoregressive 3D Shape Generation via Canonical Mappingcited by 37
2022MAUVE Scores for Generative Models: Theory and Practicecited by 36
2021Inspect, Understand, Overcome: A Survey of Practical Methods for AI Safetycited by 35
2023MMD-FUSE: Learning and Combining Kernels for Two-Sample Testing Without Data Splittingcited by 33
2025Simulation-Based Inference: A Practical Guidecited by 32
2022Evaluating generative models in high energy physicscited by 31
2021Truncated Marginal Neural Ratio Estimationcited by 30
2023HAAR: Text-Conditioned Generative Model of 3D Strand-based Human Hairstylescited by 30
2021A Trust Crisis In Simulation-Based Inference? Your Posterior Approximations Can Be Unfaithfulcited by 28
2022SPAGHETTI: Editing Implicit Shapes Through Part Aware Generationcited by 28
2023Free-form Flows: Make Any Architecture a Normalizing Flowcited by 26
20253D Scene Generation: A Surveycited by 24
2021A Witness Two-Sample Testcited by 23
2021Learning to Generate 3D Shapes with Generative Cellular Automatacited by 22
2021Regularising Inverse Problems with Generative Machine Learning Modelscited by 22
2023Inductive Simulation of Calorimeter Showers with Normalizing Flowscited by 22
2023Sequential Predictive Two-Sample and Independence Testingcited by 21
2022AutoML Two-Sample Testcited by 20
2024Detecting Machine-Generated Texts by Multi-Population Aware Optimization for Maximum Mean Discrepancycited by 20
2022A Learning Based Hypothesis Test for Harmful Covariate Shiftcited by 19
2021Meta Two-Sample Testing: Learning Kernels for Testing with Limited Datacited by 18
2021Divergence Frontiers for Generative Models: Sample Complexity, Quantization Effects, and Frontier Integralscited by 18
2023L-C2ST: Local Diagnostics for Posterior Approximations in Simulation-Based Inferencecited by 18
2023Deep Generative Models for Detector Signature Simulation: A Taxonomic Reviewcited by 18
2024A Survey on Quality Metrics for Text-to-Image Generationcited by 18
2021GUIGAN: Learning to Generate GUI Designs Using Generative Adversarial Networkscited by 17
2023Generalized Bayesian Inference for Scientific Simulators via Amortized Cost Estimationcited by 17
2025E-valuator: Reliable Agent Verifiers with Sequential Hypothesis Testingcited by 17
2021Group equivariant neural posterior estimationcited by 16
2022New directions for surrogate models and differentiable programming for High Energy Physics detector simulationcited by 16
2021Neural Tangent Kernel Maximum Mean Discrepancycited by 15
2022A survey on GANs for computer vision: Recent research, analysis and taxonomycited by 15
2022Random Forests for Change Point Detectioncited by 14

50 most cited of 268 All 268 (JSON)