Towards Real-World Test-Time Adaptation: Tri-Net Self-Training with Balanced Normalization

TL;DR

This paper introduces TRIBE, a test-time adaptation method using a tri-net architecture with balanced batch normalization and anchored self-training, effectively handling real-world scenarios with class imbalance and continual domain shifts.

Contribution

It proposes a novel tri-net framework with balanced batchnorm and anchored self-training to improve test-time adaptation in challenging real-world conditions.

Findings

TRIBE outperforms existing methods on four real-world TTA datasets.

Balanced batchnorm mitigates bias towards majority classes.

Anchored self-training stabilizes adaptation under continual domain shifts.

Abstract

Test-Time Adaptation aims to adapt source domain model to testing data at inference stage with success demonstrated in adapting to unseen corruptions. However, these attempts may fail under more challenging real-world scenarios. Existing works mainly consider real-world test-time adaptation under non-i.i.d. data stream and continual domain shift. In this work, we first complement the existing real-world TTA protocol with a globally class imbalanced testing set. We demonstrate that combining all settings together poses new challenges to existing methods. We argue the failure of state-of-the-art methods is first caused by indiscriminately adapting normalization layers to imbalanced testing data. To remedy this shortcoming, we propose a balanced batchnorm layer to swap out the regular batchnorm at inference stage. The new batchnorm layer is capable of adapting without biasing towards majority classes. We are further inspired by the success of self-training (ST) in learning from unlabeled data and adapt ST for test-time adaptation. However, ST alone is prone to over adaption which is responsible for the poor performance under continual domain shift. Hence, we propose to improve self-training under continual domain shift by regularizing model updates with an anchored loss. The final TTA model, termed as TRIBE, is built upon a tri-net architecture with balanced batchnorm layers. We evaluate TRIBE on four datasets representing real-world TTA settings. TRIBE consistently achieves the state-of-the-art performance across multiple evaluation protocols. The code is available at https://github.com/Gorilla-Lab-SCUT/TRIBE.