Distributionally Robust Learning with Stable Adversarial Training

Abstract

Machine learning algorithms with empirical risk minimization are vulnerable under distributional shifts due to the greedy adoption of all the correlations found in training data. There is an emerging literature on tackling this problem by minimizing the worst-case risk over an uncertainty set. However, existing methods mostly construct ambiguity sets by treating all variables equally regardless of the stability of their correlations with the target, resulting in the overwhelmingly-large uncertainty set and low confidence of the learner. In this paper, we propose a novel Stable Adversarial Learning (SAL) algorithm that leverages heterogeneous data sources to construct a more practical uncertainty set and conduct differentiated robustness optimization, where covariates are differentiated according to the stability of their correlations with the target. We theoretically show that our method is tractable for stochastic gradient-based optimization and provide the performance guarantees for our method. Empirical studies on both simulation and real datasets validate the effectiveness of our method in terms of uniformly good performance across unknown distributional shifts.

Accepted by Transactions on Knowledge and Data Engineering. Extension of AAAI paper (arXiv:2006.04414v2). arXiv admin note: substantial text overlap with arXiv:2006.04414

References in corpus (8)

• Domain Generalization via Invariant Feature Representation
• An Empirical Evaluation of Deep Learning on Highway Driving
• Distributionally Robust Neural Networks for Group Shifts: On the Importance of Regularization for Worst-Case Generalization
• Invariant Risk Minimization
• Does Distributionally Robust Supervised Learning Give Robust Classifiers?
• Causally Regularized Learning with Agnostic Data Selection Bias
• An Investigation of Why Overparameterization Exacerbates Spurious Correlations
• Incorporating Unlabeled Data into Distributionally Robust Learning