SOURCE-LINKED INTELLIGENCE

Distilling Image Prototypes for Guided Test-Time Adaptation

arXiv · AI, language, vision and robotics · article · Sep 9, 2026 · UTC

Test-Time Adaptation (TTA) enhances the robustness of models against distribution shifts but faces two critical challenges: error accumulation from noisy pseudo-labels and catastrophic forgetting of source knowledge. Uncertainty-based approaches designed to mitigate error accumulation often yield overconfident or computationally expensive estimates, while strategies intended to prevent forgetting via prototype replay rely on static representations that easily become misaligned as the model adapts. To address these issues, this paper proposes a novel framework, Distilling Image Prototype for Gu

Read original source ↗ Open in workspace

recordType

paper

region

Global