Revisiting classifier two-sample tests - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Conference Papers Year : 2017

Revisiting classifier two-sample tests


The goal of two-sample tests is to assess whether two samples, $S P ∼ P n$ and $S Q ∼ Q m$ , 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 S P with a positive label, and by pairing the m examples in $S Q$ 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.
Fichier principal
Vignette du fichier
classifier_tests.pdf (5.76 Mo) Télécharger le fichier
Origin Files produced by the author(s)

Dates and versions

hal-01862834 , version 1 (27-08-2018)


  • HAL Id : hal-01862834 , version 1


David Lopez-Paz, Maxime Oquab. Revisiting classifier two-sample tests. International Conference on Learning Representations, Apr 2017, Toulon, France. ⟨hal-01862834⟩
138 View
167 Download


Gmail Mastodon Facebook X LinkedIn More