Controllable Continual Test-Time Adaptation

TL;DR

This paper introduces C-CoTTA, a novel method for continual test-time adaptation that controls domain shifts to prevent error accumulation and improve model robustness without access to source data.

Contribution

We propose a controllable approach to CTTA that explicitly prevents category encroachment and reduces sensitivity to domain shifts, addressing limitations of existing suppression-based methods.

Findings

C-CoTTA outperforms existing methods in accuracy during continual test-time adaptation.

Qualitative analyses confirm the theoretical advantages of controlling domain shifts.

The method reduces the impact of uncontrollable domain transformations on model performance.

Abstract

Continual Test-Time Adaptation (CTTA) is an emerging and challenging task where a model trained in a source domain must adapt to continuously changing conditions during testing, without access to the original source data. CTTA is prone to error accumulation due to uncontrollable domain shifts, leading to blurred decision boundaries between categories. Existing CTTA methods primarily focus on suppressing domain shifts, which proves inadequate during the unsupervised test phase. In contrast, we introduce a novel approach that guides rather than suppresses these shifts. Specifically, we propose $\textbf{C}$ontrollable $\textbf{Co}$ntinual $\textbf{T}$est-$\textbf{T}$ime $\textbf{A}$daptation (C-CoTTA), which explicitly prevents any single category from encroaching on others, thereby mitigating the mutual influence between categories caused by uncontrollable shifts. Moreover, our method reduces the sensitivity of model to domain transformations, thereby minimizing the magnitude of category shifts. Extensive quantitative experiments demonstrate the effectiveness of our method, while qualitative analyses, such as t-SNE plots, confirm the theoretical validity of our approach.