On Disentangled Representations Learned from Correlated Data
Frederik Träuble, Elliot Creager, Niki Kilbertus, Francesco Locatello, Andrea Dittadi, Anirudh Goyal, Bernhard Schölkopf, Stefan Bauer
Abstract
The focus of disentanglement approaches has been on identifying independent factors of variation in data. However, the causal variables underlying real-world observations are often not statistically independent. In this work, we bridge the gap to real-world scenarios by analyzing the behavior of the most prominent disentanglement approaches on correlated data in a large-scale empirical study (including 4260 models). We show and quantify that systematically induced correlations in the dataset are being learned and reflected in the latent representations, which has implications for downstream applications of disentanglement such as fairness. We also demonstrate how to resolve these latent correlations, either using weak supervision during training or by post-hoc correcting a pre-trained model with a small number of labels.
BibTeX
@InProceedings{pmlr-v139-trauble21a,
title = {On Disentangled Representations Learned from Correlated Data},
author = {Tr{\"a}uble, Frederik and Creager, Elliot and Kilbertus, Niki and Locatello, Francesco and Dittadi, Andrea and Goyal, Anirudh and Sch{\"o}lkopf, Bernhard and Bauer, Stefan},
booktitle = {Proceedings of the 38th International Conference on Machine Learning},
pages = {10401--10412},
year = {2021},
editor = {Meila, Marina and Zhang, Tong},
volume = {139},
series = {Proceedings of Machine Learning Research},
month = {18--24 Jul},
publisher = {PMLR},
pdf = {http://proceedings.mlr.press/v139/trauble21a/trauble21a.pdf},
url = {https://proceedings.mlr.press/v139/trauble21a.html},
abstract = {The focus of disentanglement approaches has been on identifying independent factors of variation in data. However, the causal variables underlying real-world observations are often not statistically independent. In this work, we bridge the gap to real-world scenarios by analyzing the behavior of the most prominent disentanglement approaches on correlated data in a large-scale empirical study (including 4260 models). We show and quantify that systematically induced correlations in the dataset are being learned and reflected in the latent representations, which has implications for downstream applications of disentanglement such as fairness. We also demonstrate how to resolve these latent correlations, either using weak supervision during training or by post-hoc correcting a pre-trained model with a small number of labels.}
}