Return of Frustratingly Easy Domain Adaptation

TLDR

Domain shifts between training and test data degrade machine learning performance, and while supervised adaptation exists, target domains are often unlabeled, necessitating unsupervised methods. The authors propose CORAL, a simple, effective, and efficient unsupervised domain adaptation method. CORAL reduces domain shift by aligning the second‑order statistics of source and target distributions without requiring target labels. Despite its simplicity—implementable in four lines of MATLAB—CORAL achieves strong performance across extensive benchmark evaluations.

Abstract

Unlike human learning, machine learning often fails to handle changes between training (source) and test (target) input distributions. Such domain shifts, common in practical scenarios, severely damage the performance of conventional machine learning methods. Supervised domain adaptation methods have been proposed for the case when the target data have labels, including some that perform very well despite being ``frustratingly easy'' to implement. However, in practice, the target domain is often unlabeled, requiring unsupervised adaptation. We propose a simple, effective, and efficient method for unsupervised domain adaptation called CORrelation ALignment (CORAL). CORAL minimizes domain shift by aligning the second-order statistics of source and target distributions, without requiring any target labels. Even though it is extraordinarily simple--it can be implemented in four lines of Matlab code--CORAL performs remarkably well in extensive evaluations on standard benchmark datasets.