AEO amplifies the entropy gap between known and unknown samples to let a pretrained multimodal model adapt online to open-set distribution shift, beating prior TTA baselines.
Robust TTA methods (Niu et al., 2022; Zhou et al.,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Towards Robust Multimodal Open-set Test-time Adaptation via Adaptive Entropy-aware Optimization
AEO amplifies the entropy gap between known and unknown samples to let a pretrained multimodal model adapt online to open-set distribution shift, beating prior TTA baselines.