Revisiting precision recall definition for generative modeling

Loic Simon, Ryan Webster, Julien Rabin
Proceedings of the 36th International Conference on Machine Learning, PMLR 97:5799-5808, 2019.

Abstract

In this article we revisit the definition of Precision-Recall (PR) curves for generative models proposed by (Sajjadi et al., 2018). Rather than providing a scalar for generative quality, PR curves distinguish mode-collapse (poor recall) and bad quality (poor precision). We first generalize their formulation to arbitrary measures hence removing any restriction to finite support. We also expose a bridge between PR curves and type I and type II error (a.k.a. false detection and rejection) rates of likelihood ratio classifiers on the task of discriminating between samples of the two distributions. Building upon this new perspective, we propose a novel algorithm to approximate precision-recall curves, that shares some interesting methodological properties with the hypothesis testing technique from (Lopez-Paz & Oquab, 2017). We demonstrate the interest of the proposed formulation over the original approach on controlled multi-modal datasets.

Cite this Paper


BibTeX
@InProceedings{pmlr-v97-simon19a, title = {Revisiting precision recall definition for generative modeling}, author = {Simon, Loic and Webster, Ryan and Rabin, Julien}, booktitle = {Proceedings of the 36th International Conference on Machine Learning}, pages = {5799--5808}, year = {2019}, editor = {Chaudhuri, Kamalika and Salakhutdinov, Ruslan}, volume = {97}, series = {Proceedings of Machine Learning Research}, month = {09--15 Jun}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v97/simon19a/simon19a.pdf}, url = {https://proceedings.mlr.press/v97/simon19a.html}, abstract = {In this article we revisit the definition of Precision-Recall (PR) curves for generative models proposed by (Sajjadi et al., 2018). Rather than providing a scalar for generative quality, PR curves distinguish mode-collapse (poor recall) and bad quality (poor precision). We first generalize their formulation to arbitrary measures hence removing any restriction to finite support. We also expose a bridge between PR curves and type I and type II error (a.k.a. false detection and rejection) rates of likelihood ratio classifiers on the task of discriminating between samples of the two distributions. Building upon this new perspective, we propose a novel algorithm to approximate precision-recall curves, that shares some interesting methodological properties with the hypothesis testing technique from (Lopez-Paz & Oquab, 2017). We demonstrate the interest of the proposed formulation over the original approach on controlled multi-modal datasets.} }
Endnote
%0 Conference Paper %T Revisiting precision recall definition for generative modeling %A Loic Simon %A Ryan Webster %A Julien Rabin %B Proceedings of the 36th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2019 %E Kamalika Chaudhuri %E Ruslan Salakhutdinov %F pmlr-v97-simon19a %I PMLR %P 5799--5808 %U https://proceedings.mlr.press/v97/simon19a.html %V 97 %X In this article we revisit the definition of Precision-Recall (PR) curves for generative models proposed by (Sajjadi et al., 2018). Rather than providing a scalar for generative quality, PR curves distinguish mode-collapse (poor recall) and bad quality (poor precision). We first generalize their formulation to arbitrary measures hence removing any restriction to finite support. We also expose a bridge between PR curves and type I and type II error (a.k.a. false detection and rejection) rates of likelihood ratio classifiers on the task of discriminating between samples of the two distributions. Building upon this new perspective, we propose a novel algorithm to approximate precision-recall curves, that shares some interesting methodological properties with the hypothesis testing technique from (Lopez-Paz & Oquab, 2017). We demonstrate the interest of the proposed formulation over the original approach on controlled multi-modal datasets.
APA
Simon, L., Webster, R. & Rabin, J.. (2019). Revisiting precision recall definition for generative modeling. Proceedings of the 36th International Conference on Machine Learning, in Proceedings of Machine Learning Research 97:5799-5808 Available from https://proceedings.mlr.press/v97/simon19a.html.

Related Material