Regularized Learning for Domain Adaptation under Label Shifts

TL;DR

This paper introduces RLLS, a novel domain adaptation method that corrects label shifts by estimating importance weights and provides a generalization bound independent of data dimensions, with improved accuracy demonstrated on CIFAR-10 and MNIST.

Contribution

It presents the first generalization bound for label shift without target labels and introduces a regularized estimator for small-sample scenarios.

Findings

RLLS improves classification accuracy in low-sample regimes.

The generalization bound is independent of data dimensions.

Experiments show superior performance over previous methods.

Abstract

We propose Regularized Learning under Label shifts (RLLS), a principled and a practical domain-adaptation algorithm to correct for shifts in the label distribution between a source and a target domain. We first estimate importance weights using labeled source data and unlabeled target data, and then train a classifier on the weighted source samples. We derive a generalization bound for the classifier on the target domain which is independent of the (ambient) data dimensions, and instead only depends on the complexity of the function class. To the best of our knowledge, this is the first generalization bound for the label-shift problem where the labels in the target domain are not available. Based on this bound, we propose a regularized estimator for the small-sample regime which accounts for the uncertainty in the estimated weights. Experiments on the CIFAR-10 and MNIST datasets show that RLLS improves classification accuracy, especially in the low sample and large-shift regimes, compared to previous methods.