Generalized Robust Test-Time Adaptation in Continuous Dynamic Scenarios

TL;DR

This paper introduces GRoTTA, a novel method for robust test-time adaptation that effectively handles simultaneous covariate and label shifts in continuous data streams, improving model stability and performance in real-world scenarios.

Contribution

The paper proposes a comprehensive framework, GRoTTA, that addresses both covariate and label shifts during test-time adaptation with novel strategies like balanced prediction, batch normalization recalibration, and bias-guided output adaptation.

Findings

GRoTTA significantly outperforms existing methods under practical test-time adaptation scenarios.

The approach effectively mitigates the effects of continual covariate and label shifts.

Experimental results demonstrate improved stability and accuracy in real-world data streams.

Abstract

Test-time adaptation (TTA) adapts the pre-trained models to test distributions during the inference phase exclusively employing unlabeled test data streams, which holds great value for the deployment of models in real-world applications. Numerous studies have achieved promising performance on simplistic test streams, characterized by independently and uniformly sampled test data originating from a fixed target data distribution. However, these methods frequently prove ineffective in practical scenarios, where both continual covariate shift and continual label shift occur simultaneously, i.e., data and label distributions change concurrently and continually over time. In this study, a more challenging Practical Test-Time Adaptation (PTTA) setup is introduced, which takes into account the concurrent presence of continual covariate shift and continual label shift, and we propose a Generalized Robust Test-Time Adaptation (GRoTTA) method to effectively address the difficult problem. We start by steadily adapting the model through Robust Parameter Adaptation to make balanced predictions for test samples. To be specific, firstly, the effects of continual label shift are eliminated by enforcing the model to learn from a uniform label distribution and introducing recalibration of batch normalization to ensure stability. Secondly, the continual covariate shift is alleviated by employing a source knowledge regularization with the teacher-student model to update parameters. Considering the potential information in the test stream, we further refine the balanced predictions by Bias-Guided Output Adaptation, which exploits latent structure in the feature space and is adaptive to the imbalanced label distribution. Extensive experiments demonstrate GRoTTA outperforms the existing competitors by a large margin under PTTA setting, rendering it highly conducive for adoption in real-world applications.