Fishr: Invariant Gradient Variances for Out-of-Distribution Generalization - Archive ouverte HAL Access content directly
Conference Papers Year : 2022

Fishr: Invariant Gradient Variances for Out-of-Distribution Generalization

Abstract

Learning robust models that generalize well under changes in the data distribution is critical for real-world applications. To this end, there has been a growing surge of interest to learn simultaneously from multiple training domains - while enforcing different types of invariance across those domains. Yet, all existing approaches fail to show systematic benefits under controlled evaluation protocols. In this paper, we introduce a new regularization - named Fishr - that enforces domain invariance in the space of the gradients of the loss: specifically, the domain-level variances of gradients are matched across training domains. Our approach is based on the close relations between the gradient covariance, the Fisher Information and the Hessian of the loss: in particular, we show that Fishr eventually aligns the domain-level loss landscapes locally around the final weights. Extensive experiments demonstrate the effectiveness of Fishr for out-of-distribution generalization. Notably, Fishr improves the state of the art on the DomainBed benchmark and performs consistently better than Empirical Risk Minimization. Our code is available at https://github.com/alexrame/fishr.
Fichier principal
Vignette du fichier
fishr-final.pdf (2.27 Mo) Télécharger le fichier
Origin Files produced by the author(s)

Dates and versions

hal-03944846 , version 1 (18-01-2023)

Identifiers

Cite

Alexandre Rame, Corentin Dancette, Matthieu Cord. Fishr: Invariant Gradient Variances for Out-of-Distribution Generalization. 39th International Conference on Machine Learning (ICML 2022), Apr 2022, Baltimore, MD, United States. pp.18347--18377. ⟨hal-03944846⟩
7 View
26 Download

Altmetric

Share

Gmail Mastodon Facebook X LinkedIn More