Test-Time Adaptation with Binary Feedback

Taeckyung Lee, Sorn Chottananurak, Junsu Kim, Jinwoo Shin, Taesik Gong, Sung-Ju Lee
Proceedings of the 42nd International Conference on Machine Learning, PMLR 267:33005-33024, 2025.

Abstract

Deep learning models perform poorly when domain shifts exist between training and test data. Test-time adaptation (TTA) is a paradigm to mitigate this issue by adapting pre-trained models using only unlabeled test samples. However, existing TTA methods can fail under severe domain shifts, while recent active TTA approaches requiring full-class labels are impractical due to high labeling costs. To address this issue, we introduce a new setting of TTA with binary feedback, which uses a few binary feedbacks from annotators to indicate whether model predictions are correct, thereby significantly reducing the labeling burden of annotators. Under the setting, we propose BiTTA, a novel dual-path optimization framework that leverages reinforcement learning to balance binary feedback-guided adaptation on uncertain samples with agreement-based self-adaptation on confident predictions. Experiments show BiTTA achieves substantial accuracy improvements over state-of-the-art baselines, demonstrating its effectiveness in handling severe distribution shifts with minimal labeling effort.

Cite this Paper

BibTeX
@InProceedings{pmlr-v267-lee25g, title = {Test-Time Adaptation with Binary Feedback}, author = {Lee, Taeckyung and Chottananurak, Sorn and Kim, Junsu and Shin, Jinwoo and Gong, Taesik and Lee, Sung-Ju}, booktitle = {Proceedings of the 42nd International Conference on Machine Learning}, pages = {33005--33024}, year = {2025}, editor = {Singh, Aarti and Fazel, Maryam and Hsu, Daniel and Lacoste-Julien, Simon and Berkenkamp, Felix and Maharaj, Tegan and Wagstaff, Kiri and Zhu, Jerry}, volume = {267}, series = {Proceedings of Machine Learning Research}, month = {13--19 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v267/main/assets/lee25g/lee25g.pdf}, url = {https://proceedings.mlr.press/v267/lee25g.html}, abstract = {Deep learning models perform poorly when domain shifts exist between training and test data. Test-time adaptation (TTA) is a paradigm to mitigate this issue by adapting pre-trained models using only unlabeled test samples. However, existing TTA methods can fail under severe domain shifts, while recent active TTA approaches requiring full-class labels are impractical due to high labeling costs. To address this issue, we introduce a new setting of TTA with binary feedback, which uses a few binary feedbacks from annotators to indicate whether model predictions are correct, thereby significantly reducing the labeling burden of annotators. Under the setting, we propose BiTTA, a novel dual-path optimization framework that leverages reinforcement learning to balance binary feedback-guided adaptation on uncertain samples with agreement-based self-adaptation on confident predictions. Experiments show BiTTA achieves substantial accuracy improvements over state-of-the-art baselines, demonstrating its effectiveness in handling severe distribution shifts with minimal labeling effort.} }
Endnote
%0 Conference Paper %T Test-Time Adaptation with Binary Feedback %A Taeckyung Lee %A Sorn Chottananurak %A Junsu Kim %A Jinwoo Shin %A Taesik Gong %A Sung-Ju Lee %B Proceedings of the 42nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2025 %E Aarti Singh %E Maryam Fazel %E Daniel Hsu %E Simon Lacoste-Julien %E Felix Berkenkamp %E Tegan Maharaj %E Kiri Wagstaff %E Jerry Zhu %F pmlr-v267-lee25g %I PMLR %P 33005--33024 %U https://proceedings.mlr.press/v267/lee25g.html %V 267 %X Deep learning models perform poorly when domain shifts exist between training and test data. Test-time adaptation (TTA) is a paradigm to mitigate this issue by adapting pre-trained models using only unlabeled test samples. However, existing TTA methods can fail under severe domain shifts, while recent active TTA approaches requiring full-class labels are impractical due to high labeling costs. To address this issue, we introduce a new setting of TTA with binary feedback, which uses a few binary feedbacks from annotators to indicate whether model predictions are correct, thereby significantly reducing the labeling burden of annotators. Under the setting, we propose BiTTA, a novel dual-path optimization framework that leverages reinforcement learning to balance binary feedback-guided adaptation on uncertain samples with agreement-based self-adaptation on confident predictions. Experiments show BiTTA achieves substantial accuracy improvements over state-of-the-art baselines, demonstrating its effectiveness in handling severe distribution shifts with minimal labeling effort.
APA
Lee, T., Chottananurak, S., Kim, J., Shin, J., Gong, T. & Lee, S.. (2025). Test-Time Adaptation with Binary Feedback. Proceedings of the 42nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 267:33005-33024 Available from https://proceedings.mlr.press/v267/lee25g.html.

Related Material