Distribution Shift Robustness

Overview

The field asks how models can remain accurate, calibrated, and fair when data change across time, locations, populations, or conditions, including shifts in inputs, labels, and their relationships. Approaches include distributionally robust optimization, invariant and causal representations, targeted data augmentation, test-time adaptation, and conformal methods that adjust predictions or quantify uncertainty. Evaluation increasingly uses temporal and cross-domain tests, realistic corruption benchmarks, and worst-group measures rather than relying only on random train-test splits. A central tension is that adapting to a new domain can improve its performance but also cause instability or forget earlier capabilities, while guarantees that cover broad shifts may produce overly conservative predictions.

Latest papers

302 papers23 papers in the last four weeks, down 41% on the four weeks before. 0.2% of all new papers.

Apr 6, 2023

Reliable learning in challenging environments

The problem of designing learners that provide guarantees that their predictions are provably correct is of increasing importance in machine learning. However, learning theoretic guarantees have only been considered in very specific settings. In this work, we consider the design and analysis of reliable learners in challenging test-time environments as encountered in modern machine learning problems: namely adversarial' test-time attacks (in several variations) and natural' distribution shifts. In this work, we provide a reliable learner with provably optimal guarantees in such settings. We discuss practical implementations of the learner and further show that our algorithm achieves strong positive performance guarantees on several natural examples: for example, linear separators under log-concave distributions or smooth boundary classifiers under smooth probability distributions.
Sep 5, 2022

Learning from a Biased Sample

The empirical risk minimization approach to data-driven decision making requires access to training data drawn under the same conditions as those that will be faced when the decision rule is deployed. However, in a number of settings, we may be concerned that our training sample is biased in the sense that some groups (characterized by either observable or unobservable attributes) may be under- or over-represented relative to the general population; and in this setting empirical risk minimization over the training set may fail to yield rules that perform well at deployment. We propose a model of sampling bias called conditional ΓΓ-biased sampling, where observed covariates can affect the probability of sample selection arbitrarily much but the amount of unexplained variation in the probability of sample selection is bounded by a constant factor. Applying the distributionally robust optimization framework, we propose a method for learning a decision rule that minimizes the worst-case risk incurred under a family of test distributions that can generate the training distribution under ΓΓ-biased sampling. We apply a result of Rockafellar and Uryasev to show that this problem is equivalent to an augmented convex risk minimization problem. We give statistical guarantees for learning a model that is robust to sampling bias via the method of sieves, and propose a deep learning algorithm whose loss function captures our robust learning target. We empirically validate our proposed method in a case study on prediction of mental health scores from health survey data and a case study on ICU length of stay prediction.