Learning Less Generalizable Patterns for Better Test-Time Adaptation

Thomas Duboudin, Emmanuel Dellandréa, Corentin Abgrall, Gilles Hénaff, Liming Chen

2023

Abstract

Deep neural networks often fail to generalize outside of their training distribution, particularly when only a single data domain is available during training. While test-time adaptation has yielded encouraging results in this setting, we argue that to reach further improvements, these approaches should be combined with training procedure modifications aiming to learn a more diverse set of patterns. Indeed, test-time adaptation methods usually have to rely on a limited representation because of the shortcut learning phenomenon: only a subset of the available predictive patterns is learned with standard training. In this paper, we first show that the combined use of existing training-time strategies and test-time batch normalization, a simple adaptation method, does not always improve upon the test-time adaptation alone on the PACS benchmark. Furthermore, experiments on Office-Home show that very few training-time methods improve upon standard training, with or without test-time batch normalization. Therefore, we propose a novel approach that mitigates the shortcut learning behavior by having an additional classification branch learn less predictive and generalizable patterns. Our experiments show that our method improves upon the state-of-the-art results on both benchmarks and benefits the most to test-time batch normalization.

Paper Citation

in Harvard Style

Duboudin T., Dellandréa E., Abgrall C., Hénaff G. and Chen L. (2023). Learning Less Generalizable Patterns for Better Test-Time Adaptation. In Proceedings of the 18th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2023) - Volume 5: VISAPP; ISBN 978-989-758-634-7, SciTePress, pages 349-358. DOI: 10.5220/0011893800003417

in Bibtex Style

@conference{visapp23,
author={Thomas Duboudin and Emmanuel Dellandréa and Corentin Abgrall and Gilles Hénaff and Liming Chen},
title={Learning Less Generalizable Patterns for Better Test-Time Adaptation},
booktitle={Proceedings of the 18th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2023) - Volume 5: VISAPP},
year={2023},
pages={349-358},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0011893800003417},
isbn={978-989-758-634-7},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 18th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2023) - Volume 5: VISAPP
TI - Learning Less Generalizable Patterns for Better Test-Time Adaptation
SN - 978-989-758-634-7
AU - Duboudin T.
AU - Dellandréa E.
AU - Abgrall C.
AU - Hénaff G.
AU - Chen L.
PY - 2023
SP - 349
EP - 358
DO - 10.5220/0011893800003417
PB - SciTePress