Invariant Risk Minimization
2019-07-05Code Available1· sign in to hype
Martin Arjovsky, Léon Bottou, Ishaan Gulrajani, David Lopez-Paz
Code Available — Be the first to reproduce this paper.
ReproduceCode
- github.com/facebookresearch/InvariantRiskMinimizationOfficialIn paperpytorch★ 434
- github.com/thuml/Transfer-Learning-Librarypytorch★ 3,889
- github.com/facebookresearch/DomainBedpytorch★ 1,604
- github.com/kakaobrain/irm-empirical-studypytorch★ 27
- github.com/ycq091044/manydgpytorch★ 24
- github.com/claudiashi57/nicepytorch★ 10
- github.com/lingxiaoyuan/ood_mechanicspytorch★ 7
- github.com/hasanjawad001/cglearnpytorch★ 6
- github.com/rwchakra/exmappytorch★ 3
- github.com/chunyangx/IRM_researchpytorch★ 1
Abstract
We introduce Invariant Risk Minimization (IRM), a learning paradigm to estimate invariant correlations across multiple training distributions. To achieve this goal, IRM learns a data representation such that the optimal classifier, on top of that data representation, matches for all training distributions. Through theory and experiments, we show how the invariances learned by IRM relate to the causal structures governing the data and enable out-of-distribution generalization.
Tasks
Benchmark Results
| Dataset | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| Colored-MNIST(with spurious correlation) | MLP-IRM | Accuracy | 66.9 | — | Unverified |