Year: 2021
Type: preprint
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 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. Code available at https://github.com/MadryLab/copriors
Source: arXiv (Cornell University)
Institution: Massachusetts Institute of Technology
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W3207460859