Improving Entropy-Based Test-Time Adaptation from a Clustering View

TL;DR

This paper presents a novel clustering perspective on entropy-based test-time adaptation, introducing improvements that enhance model performance under domain shifts by refining label assignment and update strategies.

Contribution

It introduces a new clustering view for entropy-based test-time adaptation and proposes specific enhancements like robust label assignment and similarity constraints.

Findings

Achieves consistent improvements across various datasets.

Enhances label assignment robustness during adaptation.

Improves test-time adaptation performance significantly.

Abstract

Domain shift is a common problem in the realistic world, where training data and test data follow different data distributions. To deal with this problem, fully test-time adaptation (TTA) leverages the unlabeled data encountered during test time to adapt the model. In particular, entropy-based TTA (EBTTA) methods, which minimize the prediction's entropy on test samples, have shown great success. In this paper, we introduce a new clustering perspective on the EBTTA. It is an iterative algorithm: 1) in the assignment step, the forward process of the EBTTA models is the assignment of labels for these test samples, and 2) in the updating step, the backward process is the update of the model via the assigned samples. This new perspective allows us to explore how entropy minimization influences test-time adaptation. Accordingly, this observation can guide us to put forward the improvement of EBTTA. We propose to improve EBTTA from the assignment step and the updating step, where robust label assignment, similarity-preserving constraint, sample selection, and gradient accumulation are proposed to explicitly utilize more information. Experimental results demonstrate that our method can achieve consistent improvements on various datasets. Code is provided in the supplementary material.