Entropy is not Enough for Test-Time Adaptation: From the Perspective of Disentangled Factors

TL;DR

This paper reveals the limitations of entropy as a confidence metric in test-time adaptation and introduces DeYO, a novel method that uses a new confidence measure based on disentangled factors to improve adaptation robustness.

Contribution

The paper identifies entropy's unreliability in biased scenarios and proposes DeYO, which employs a new confidence metric, PLPD, for more effective test-time adaptation.

Findings

DeYO outperforms baseline methods in various scenarios.

Entropy is unreliable for confidence estimation in biased data.

PLPD effectively measures shape influence on predictions.

Abstract

Test-time adaptation (TTA) fine-tunes pre-trained deep neural networks for unseen test data. The primary challenge of TTA is limited access to the entire test dataset during online updates, causing error accumulation. To mitigate it, TTA methods have utilized the model output's entropy as a confidence metric that aims to determine which samples have a lower likelihood of causing error. Through experimental studies, however, we observed the unreliability of entropy as a confidence metric for TTA under biased scenarios and theoretically revealed that it stems from the neglect of the influence of latent disentangled factors of data on predictions. Building upon these findings, we introduce a novel TTA method named Destroy Your Object (DeYO), which leverages a newly proposed confidence metric named Pseudo-Label Probability Difference (PLPD). PLPD quantifies the influence of the shape of an object on prediction by measuring the difference between predictions before and after applying an object-destructive transformation. DeYO consists of sample selection and sample weighting, which employ entropy and PLPD concurrently. For robust adaptation, DeYO prioritizes samples that dominantly incorporate shape information when making predictions. Our extensive experiments demonstrate the consistent superiority of DeYO over baseline methods across various scenarios, including biased and wild. Project page is publicly available at https://whitesnowdrop.github.io/DeYO/.