Learning Under Adversarial and Interventional Shifts
This work addresses robustness to distribution shifts for machine learning practitioners, but it is incremental as it builds on existing adversarial and interventional approaches.
The paper tackles the problem of designing machine learning models robust to distribution shifts by combining adversarial and interventional methods into a new formulation called RISe, which is shown to be effective through experiments on synthetic and real-world healthcare datasets.
Machine learning models are often trained on data from one distribution and deployed on others. So it becomes important to design models that are robust to distribution shifts. Most of the existing work focuses on optimizing for either adversarial shifts or interventional shifts. Adversarial methods lack expressivity in representing plausible shifts as they consider shifts to joint distributions in the data. Interventional methods allow more expressivity but provide robustness to unbounded shifts, resulting in overly conservative models. In this work, we combine the complementary strengths of the two approaches and propose a new formulation, RISe, for designing robust models against a set of distribution shifts that are at the intersection of adversarial and interventional shifts. We employ the distributionally robust optimization framework to optimize the resulting objective in both supervised and reinforcement learning settings. Extensive experimentation with synthetic and real world datasets from healthcare demonstrate the efficacy of the proposed approach.