Kore Initial Clustering for Unsupervised Domain Adaptation

Kyungsik Lee, Youngmi Jun, EunJi Kim, Suhyun Kim, Seong Hwang, Jonghyun Choi

2024

Abstract

In unsupervised domain adaptation (UDA) literature, there exists an array of techniques to derive domain adaptive features. Among them, a particularly successful family of approaches of pseudo-labeling the unlabeled target data has shown promising results. Yet, the majority of the existing methods primarily focus on leveraging only the target domain knowledge for pseudo-labeling while insufficiently considering the source domain knowledge. Here, we hypothesize that quality pseudo labels obtained via classical K-means clustering considering both the source and target domains bring simple yet significant benefits. In particular, we propose to assign pseudo labels to the target domain’s instances better aligned with the source domain labels by a simple method that modifies K-means clustering by emphasizing the strengthened notion of centroids, namely, Kore Initial Clustering (KIC). The proposed KIC is readily utilizable with a wide array of UDA models, consistently improving the UDA performance on multiple UDA datasets including Office-Home and Office-31, demonstrating the efficacy of pseudo labels in UDA.

Paper Citation

in Harvard Style

Lee K., Jun Y., Kim E., Kim S., Hwang S. and Choi J. (2024). Kore Initial Clustering for Unsupervised Domain Adaptation. In Proceedings of the 19th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications - Volume 2: VISAPP; ISBN 978-989-758-679-8, SciTePress, pages 425-432. DOI: 10.5220/0012353200003660

in Bibtex Style

@conference{visapp24,
author={Kyungsik Lee and Youngmi Jun and EunJi Kim and Suhyun Kim and Seong Hwang and Jonghyun Choi},
title={Kore Initial Clustering for Unsupervised Domain Adaptation},
booktitle={Proceedings of the 19th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications - Volume 2: VISAPP},
year={2024},
pages={425-432},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0012353200003660},
isbn={978-989-758-679-8},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 19th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications - Volume 2: VISAPP
TI - Kore Initial Clustering for Unsupervised Domain Adaptation
SN - 978-989-758-679-8
AU - Lee K.
AU - Jun Y.
AU - Kim E.
AU - Kim S.
AU - Hwang S.
AU - Choi J.
PY - 2024
SP - 425
EP - 432
DO - 10.5220/0012353200003660
PB - SciTePress