Towards Reliable Test-Time Adaptation: Style Invariance as a Correctness Likelihood

TL;DR

This paper introduces SICL, a style-invariance framework for test-time adaptation that improves predictive uncertainty calibration in dynamic, real-world conditions without additional training.

Contribution

SICL leverages style invariance to estimate correctness likelihood, providing a plug-and-play calibration method compatible with any TTA approach.

Findings

Reduces calibration error by 13 percentage points on average.

Compatible with various TTA methods and architectures.

Effective in real-world, dynamic test scenarios.

Abstract

Test-time adaptation (TTA) enables efficient adaptation of deployed models, yet it often leads to poorly calibrated predictive uncertainty - a critical issue in high-stakes domains such as autonomous driving, finance, and healthcare. Existing calibration methods typically assume fixed models or static distributions, resulting in degraded performance under real-world, dynamic test conditions. To address these challenges, we introduce Style Invariance as a Correctness Likelihood (SICL), a framework that leverages style-invariance for robust uncertainty estimation. SICL estimates instance-wise correctness likelihood by measuring prediction consistency across style-altered variants, requiring only the model's forward pass. This makes it a plug-and-play, backpropagation-free calibration module compatible with any TTA method. Comprehensive evaluations across four baselines, five TTA methods, and two realistic scenarios with three model architecture demonstrate that SICL reduces calibration error by an average of 13 percentage points compared to conventional calibration approaches.