ICML 2022spotlight17 citations
Combining Diverse Feature Priors
Saachi Jain, Dimitris Tsipras, Aleksander Madry
Abstract
To improve model generalization, model designers often restrict the features that their models use, either implicitly or explicitly. In this work, we explore the design space of leveraging such feature priors by viewing them as distinct perspectives on the data. Specifically, we find that models trained with diverse sets of explicit feature priors have less overlapping failure modes, and can thus be combined more effectively. Moreover, we demonstrate that jointly training such models on additional (unlabeled) data allows them to correct each other’s mistakes, which, in turn, leads to better generalization and resilience to spurious correlations.
BibTeX
@InProceedings{pmlr-v162-jain22b,
title = {Combining Diverse Feature Priors},
author = {Jain, Saachi and Tsipras, Dimitris and Madry, Aleksander},
booktitle = {Proceedings of the 39th International Conference on Machine Learning},
pages = {9802--9832},
year = {2022},
editor = {Chaudhuri, Kamalika and Jegelka, Stefanie and Song, Le and Szepesvari, Csaba and Niu, Gang and Sabato, Sivan},
volume = {162},
series = {Proceedings of Machine Learning Research},
month = {17--23 Jul},
publisher = {PMLR},
pdf = {https://proceedings.mlr.press/v162/jain22b/jain22b.pdf},
url = {https://proceedings.mlr.press/v162/jain22b.html},
abstract = {To improve model generalization, model designers often restrict the features that their models use, either implicitly or explicitly. In this work, we explore the design space of leveraging such feature priors by viewing them as distinct perspectives on the data. Specifically, we find that models trained with diverse sets of explicit feature priors have less overlapping failure modes, and can thus be combined more effectively. Moreover, we demonstrate that jointly training such models on additional (unlabeled) data allows them to correct each other’s mistakes, which, in turn, leads to better generalization and resilience to spurious correlations.}
}