Predict then Interpolate: A Simple Algorithm to Learn Stable Classifiers
arXiv:2105.12628
Abstract
We propose Predict then Interpolate (PI), a simple algorithm for learning correlations that are stable across environments. The algorithm follows from the intuition that when using a classifier trained on one environment to make predictions on examples from another environment, its mistakes are informative as to which correlations are unstable. In this work, we prove that by interpolating the distributions of the correct predictions and the wrong predictions, we can uncover an oracle distribution where the unstable correlation vanishes. Since the oracle interpolation coefficients are not accessible, we use group distributionally robust optimization to minimize the worst-case risk across all such interpolations. We evaluate our method on both text classification and image classification. Empirical results demonstrate that our algorithm is able to learn robust classifiers (outperforms IRM by 23.85% on synthetic environments and 12.41% on natural environments). Our code and data are available at https://github.com/YujiaBao/Predict-then-Interpolate.
arXiv commentICML 2021
References in corpus (8)
- Adam: A Method for Stochastic Optimization
- Distributionally Robust Neural Networks for Group Shifts: On the Importance of Regularization for Worst-Case Generalization
- Invariant Risk Minimization
- Out-of-Distribution Generalization via Risk Extrapolation (REx)
- Does Distributionally Robust Supervised Learning Give Robust Classifiers?
- In Search of Lost Domain Generalization
- An Empirical Study of Invariant Risk Minimization
- Enforcing Predictive Invariance across Structured Biomedical Domains