Understanding Self-Training for Gradual Domain Adaptation

TL;DR

This paper provides a theoretical analysis and practical insights into self-training for gradual domain adaptation, demonstrating improved accuracy on datasets with evolving data distributions.

Contribution

It introduces the first non-vacuous error bound for self-training under gradual shifts and highlights the importance of regularization and label sharpening.

Findings

Self-training performs well for small Wasserstein-infinity shifts.

Regularization and label sharpening are crucial even with infinite data.

Higher accuracies achieved on rotating MNIST and Portraits datasets.

Abstract

Machine learning systems must adapt to data distributions that evolve over time, in applications ranging from sensor networks and self-driving car perception modules to brain-machine interfaces. We consider gradual domain adaptation, where the goal is to adapt an initial classifier trained on a source domain given only unlabeled data that shifts gradually in distribution towards a target domain. We prove the first non-vacuous upper bound on the error of self-training with gradual shifts, under settings where directly adapting to the target domain can result in unbounded error. The theoretical analysis leads to algorithmic insights, highlighting that regularization and label sharpening are essential even when we have infinite data, and suggesting that self-training works particularly well for shifts with small Wasserstein-infinity distance. Leveraging the gradual shift structure leads to higher accuracies on a rotating MNIST dataset and a realistic Portraits dataset.