Discovering environments with XRM

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Pezeshki, Mohammad, Bouchacourt, Diane, Ibrahim, Mark, Ballas, Nicolas, Vincent, Pascal, Lopez-Paz, David
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909262249197568
author Pezeshki, Mohammad
Bouchacourt, Diane
Ibrahim, Mark
Ballas, Nicolas
Vincent, Pascal
Lopez-Paz, David
author_facet Pezeshki, Mohammad
Bouchacourt, Diane
Ibrahim, Mark
Ballas, Nicolas
Vincent, Pascal
Lopez-Paz, David
contents Environment annotations are essential for the success of many out-of-distribution (OOD) generalization methods. Unfortunately, these are costly to obtain and often limited by human annotators' biases. To achieve robust generalization, it is essential to develop algorithms for automatic environment discovery within datasets. Current proposals, which divide examples based on their training error, suffer from one fundamental problem. These methods introduce hyper-parameters and early-stopping criteria, which require a validation set with human-annotated environments, the very information subject to discovery. In this paper, we propose Cross-Risk-Minimization (XRM) to address this issue. XRM trains twin networks, each learning from one random half of the training data, while imitating confident held-out mistakes made by its sibling. XRM provides a recipe for hyper-parameter tuning, does not require early-stopping, and can discover environments for all training and validation data. Algorithms built on top of XRM environments achieve oracle worst-group-accuracy, addressing a long-standing challenge in OOD generalization. Code available at \url{https://github.com/facebookresearch/XRM}.
format Preprint
id arxiv_https___arxiv_org_abs_2309_16748
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Discovering environments with XRM
Pezeshki, Mohammad
Bouchacourt, Diane
Ibrahim, Mark
Ballas, Nicolas
Vincent, Pascal
Lopez-Paz, David
Machine Learning
Artificial Intelligence
Environment annotations are essential for the success of many out-of-distribution (OOD) generalization methods. Unfortunately, these are costly to obtain and often limited by human annotators' biases. To achieve robust generalization, it is essential to develop algorithms for automatic environment discovery within datasets. Current proposals, which divide examples based on their training error, suffer from one fundamental problem. These methods introduce hyper-parameters and early-stopping criteria, which require a validation set with human-annotated environments, the very information subject to discovery. In this paper, we propose Cross-Risk-Minimization (XRM) to address this issue. XRM trains twin networks, each learning from one random half of the training data, while imitating confident held-out mistakes made by its sibling. XRM provides a recipe for hyper-parameter tuning, does not require early-stopping, and can discover environments for all training and validation data. Algorithms built on top of XRM environments achieve oracle worst-group-accuracy, addressing a long-standing challenge in OOD generalization. Code available at \url{https://github.com/facebookresearch/XRM}.
title Discovering environments with XRM
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2309.16748