Search papers, labs, and topics across Lattice.
This paper investigates the finite-sample generalization error of robust learning frameworks, specifically Distributionally Robust Optimization (DRO) and Robust Satisficing (RS), under distributional shifts. By deriving explicit error bounds that characterize the trade-off between sensitivity to shifts and regularization penalties, the authors provide a systematic comparison of these methods in the target environment. Additionally, they introduce information-directed hyperparameter calibrations that enhance the performance of both methods when partial shift information is available, demonstrating their complementary behaviors in practical applications such as network lot-sizing.
Robust learning methods can be systematically compared under distributional shifts, revealing unexpected complementary behaviors that enhance performance with partial information.
Distributional shifts arise when the target deployment environment differs from the source environment that generated the training data. Robust learning frameworks such as Distributionally Robust Optimization (DRO) and Robust Satisficing (RS) aim to address this challenge, yet their finite-sample guarantees under such shifts, and their systematic comparison, remain underexplored: existing analyses typically establish guarantees either in the source environment or for adversarial worst-case performance over an ambiguity set. This paper instead studies generalization error in the target environment---the excess loss under the shifted target distribution. Our contributions are threefold. First, we derive finite-sample generalization error bounds in the shifted target environment for both DRO and RS. These bounds explicitly characterize the trade-off between reduced sensitivity to shift and the regularization penalty induced by each method's robustness hyperparameter, and they avoid the curse of dimensionality associated with Wasserstein empirical concentration. Second, when partial shift information such as shift magnitude or direction is available, we propose information-directed hyperparameter calibrations and compare the two methods given the same information. Under these calibrations, and in the partial-information regimes we study, DRO and RS exhibit complementary theoretical and empirical behavior. Finally, we apply the framework to a network lot-sizing problem, using it to interpret how robust policies respond to positive shifts in the demand distribution. Together, these results fill a gap in understanding the statistical properties of robust learning methods under distributional shifts and provide a principled basis for comparing DRO and RS.
