Fair Generative Modeling via Weak Supervision

Kristy Choi, Aditya Grover, Trisha Singh, Rui Shu, Stefano Ermon
Proceedings of the 37th International Conference on Machine Learning, PMLR 119:1887-1898, 2020.

Abstract

Real-world datasets are often biased with respect to key demographic factors such as race and gender. Due to the latent nature of the underlying factors, detecting and mitigating bias is especially challenging for unsupervised machine learning. We present a weakly supervised algorithm for overcoming dataset bias for deep generative models. Our approach requires access to an additional small, unlabeled reference dataset as the supervision signal, thus sidestepping the need for explicit labels on the underlying bias factors. Using this supplementary dataset, we detect the bias in existing datasets via a density ratio technique and learn generative models which efficiently achieve the twin goals of: 1) data efficiency by using training examples from both biased and reference datasets for learning; and 2) data generation close in distribution to the reference dataset at test time. Empirically, we demonstrate the efficacy of our approach which reduces bias w.r.t. latent factors by an average of up to 34.6% over baselines for comparable image generation using generative adversarial networks.

Cite this Paper

BibTeX
@InProceedings{pmlr-v119-choi20a, title = {Fair Generative Modeling via Weak Supervision}, author = {Choi, Kristy and Grover, Aditya and Singh, Trisha and Shu, Rui and Ermon, Stefano}, booktitle = {Proceedings of the 37th International Conference on Machine Learning}, pages = {1887--1898}, year = {2020}, editor = {III, Hal Daumé and Singh, Aarti}, volume = {119}, series = {Proceedings of Machine Learning Research}, month = {13--18 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v119/choi20a/choi20a.pdf}, url = {https://proceedings.mlr.press/v119/choi20a.html}, abstract = {Real-world datasets are often biased with respect to key demographic factors such as race and gender. Due to the latent nature of the underlying factors, detecting and mitigating bias is especially challenging for unsupervised machine learning. We present a weakly supervised algorithm for overcoming dataset bias for deep generative models. Our approach requires access to an additional small, unlabeled reference dataset as the supervision signal, thus sidestepping the need for explicit labels on the underlying bias factors. Using this supplementary dataset, we detect the bias in existing datasets via a density ratio technique and learn generative models which efficiently achieve the twin goals of: 1) data efficiency by using training examples from both biased and reference datasets for learning; and 2) data generation close in distribution to the reference dataset at test time. Empirically, we demonstrate the efficacy of our approach which reduces bias w.r.t. latent factors by an average of up to 34.6% over baselines for comparable image generation using generative adversarial networks.} }
Endnote
%0 Conference Paper %T Fair Generative Modeling via Weak Supervision %A Kristy Choi %A Aditya Grover %A Trisha Singh %A Rui Shu %A Stefano Ermon %B Proceedings of the 37th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2020 %E Hal Daumé III %E Aarti Singh %F pmlr-v119-choi20a %I PMLR %P 1887--1898 %U https://proceedings.mlr.press/v119/choi20a.html %V 119 %X Real-world datasets are often biased with respect to key demographic factors such as race and gender. Due to the latent nature of the underlying factors, detecting and mitigating bias is especially challenging for unsupervised machine learning. We present a weakly supervised algorithm for overcoming dataset bias for deep generative models. Our approach requires access to an additional small, unlabeled reference dataset as the supervision signal, thus sidestepping the need for explicit labels on the underlying bias factors. Using this supplementary dataset, we detect the bias in existing datasets via a density ratio technique and learn generative models which efficiently achieve the twin goals of: 1) data efficiency by using training examples from both biased and reference datasets for learning; and 2) data generation close in distribution to the reference dataset at test time. Empirically, we demonstrate the efficacy of our approach which reduces bias w.r.t. latent factors by an average of up to 34.6% over baselines for comparable image generation using generative adversarial networks.
APA
Choi, K., Grover, A., Singh, T., Shu, R. & Ermon, S.. (2020). Fair Generative Modeling via Weak Supervision. Proceedings of the 37th International Conference on Machine Learning, in Proceedings of Machine Learning Research 119:1887-1898 Available from https://proceedings.mlr.press/v119/choi20a.html.

Related Material