Robust Generalization despite Distribution Shift via Minimum Discriminating Information
This addresses the challenge of robust generalization for machine learning models in real-world scenarios with distribution shifts, though it appears incremental as it builds on existing principles like minimum discriminating information and distributionally robust optimization.
The paper tackles the problem of training models that generalize under distribution shift by incorporating partial structural knowledge of the test distribution, using minimum discriminating information and distributionally robust optimization to derive explicit generalization bounds. It demonstrates applications in training classifiers on biased data and off-policy evaluation in Markov Decision Processes.
Training models that perform well under distribution shifts is a central challenge in machine learning. In this paper, we introduce a modeling framework where, in addition to training data, we have partial structural knowledge of the shifted test distribution. We employ the principle of minimum discriminating information to embed the available prior knowledge, and use distributionally robust optimization to account for uncertainty due to the limited samples. By leveraging large deviation results, we obtain explicit generalization bounds with respect to the unknown shifted distribution. Lastly, we demonstrate the versatility of our framework by demonstrating it on two rather distinct applications: (1) training classifiers on systematically biased data and (2) off-policy evaluation in Markov Decision Processes.