Enhancing Distributional Stability among Sub-populations

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Liu, Jiashuo, Wu, Jiayun, Peng, Jie, Wu, Xiaoyu, Zheng, Yang, Li, Bo, Cui, Peng
Natura: Preprint
Pubblicazione: 2022
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866910329620922368
author Liu, Jiashuo
Wu, Jiayun
Peng, Jie
Wu, Xiaoyu
Zheng, Yang
Li, Bo
Cui, Peng
author_facet Liu, Jiashuo
Wu, Jiayun
Peng, Jie
Wu, Xiaoyu
Zheng, Yang
Li, Bo
Cui, Peng
contents Enhancing the stability of machine learning algorithms under distributional shifts is at the heart of the Out-of-Distribution (OOD) Generalization problem. Derived from causal learning, recent works of invariant learning pursue strict invariance with multiple training environments. Although intuitively reasonable, strong assumptions on the availability and quality of environments are made to learn the strict invariance property. In this work, we come up with the ``distributional stability" notion to mitigate such limitations. It quantifies the stability of prediction mechanisms among sub-populations down to a prescribed scale. Based on this, we propose the learnability assumption and derive the generalization error bound under distribution shifts. Inspired by theoretical analyses, we propose our novel stable risk minimization (SRM) algorithm to enhance the model's stability w.r.t. shifts in prediction mechanisms ($Y|X$-shifts). Experimental results are consistent with our intuition and validate the effectiveness of our algorithm. The code can be found at https://github.com/LJSthu/SRM.
format Preprint
id arxiv_https___arxiv_org_abs_2206_02990
institution arXiv
publishDate 2022
record_format arxiv
spellingShingle Enhancing Distributional Stability among Sub-populations
Liu, Jiashuo
Wu, Jiayun
Peng, Jie
Wu, Xiaoyu
Zheng, Yang
Li, Bo
Cui, Peng
Machine Learning
Enhancing the stability of machine learning algorithms under distributional shifts is at the heart of the Out-of-Distribution (OOD) Generalization problem. Derived from causal learning, recent works of invariant learning pursue strict invariance with multiple training environments. Although intuitively reasonable, strong assumptions on the availability and quality of environments are made to learn the strict invariance property. In this work, we come up with the ``distributional stability" notion to mitigate such limitations. It quantifies the stability of prediction mechanisms among sub-populations down to a prescribed scale. Based on this, we propose the learnability assumption and derive the generalization error bound under distribution shifts. Inspired by theoretical analyses, we propose our novel stable risk minimization (SRM) algorithm to enhance the model's stability w.r.t. shifts in prediction mechanisms ($Y|X$-shifts). Experimental results are consistent with our intuition and validate the effectiveness of our algorithm. The code can be found at https://github.com/LJSthu/SRM.
title Enhancing Distributional Stability among Sub-populations
topic Machine Learning
url https://arxiv.org/abs/2206.02990