In this paper, we propose Test-Time Training, a general approach for improving the performance of predictive models when training and test data come from different distributions. We turn a single unlabeled test sample into a self-supervised learning problem, on which we update the model parameters before making a prediction. This also extends naturally to data in an online stream. Our simple approach leads to improvements on diverse image classification benchmarks aimed at evaluating robustness to distribution shifts.
@article{arxiv.1909.13231,
title = {Test-Time Training with Self-Supervision for Generalization under Distribution Shifts},
author = {Yu Sun and Xiaolong Wang and Zhuang Liu and John Miller and Alexei A. Efros and Moritz Hardt},
journal= {arXiv preprint arXiv:1909.13231},
year = {2020}
}