PuSH - Publikationsserver des Helmholtz Zentrums München

Ambekar, S. ; Hasny, M. ; Daza, L. ; Lang, D. ; Schnabel, J.A.*

Hierarchical Adaptive networks with Task vectors for Test-Time Adaptation.

In: (2026 IEEE/CVF Winter Conference on Applications of Computer Vision, WACV 2026, 6-10 March 2026, Tucson). 2026. 4661-4672 (Proceedings 2026 IEEE Cvf Winter Conference on Applications of Computer Vision Wacv 2026)
DOI
Test-time adaptation allows pretrained models to adjust to incoming data streams, addressing distribution shifts between source and target domains. However, standard methods rely on single-dimensional linear classification layers, which often fail to handle diverse and complex shifts. We propose Hierarchical Adaptive Networks with Task Vectors (Hi-Vec), which leverages multiple layers of increasing size for dynamic test-time adaptation. By decomposing the encoder's representation space into such hierarchically organized layers, Hi-Vec, in a plug-and-play manner, allows existing methods to adapt to shifts of varying complexity. Our contributions are threefold: First, we propose dynamic layer selection for automatic identification of the optimal layer for adaptation to each test batch. Second, we propose a mechanism that merges weights from the dynamic layer to other layers, ensuring all layers receive target information. Third, we propose linear layer agreement that acts as a gating function, preventing erroneous fine-tuning by adaptation on noisy batches. We rigorously evaluate the performance of Hi-Vec in challenging scenarios and on multiple target datasets, proving its strong capability to advance state-of-the-art methods. Our results show that Hi-Vec improves robustness, addresses uncertainty, and handles limited batch sizes and increased outlier rates. Code: https://github.com/ambekarsameer96/Hi-Vec
Altmetric
Weitere Metriken?
[➜Einloggen]
Zusatzinfos bearbeiten [➜Einloggen]
Publikationstyp Artikel: Konferenzbeitrag
Schlagwörter Hierachical Adaptation ; Task Vectors ; Test-time Adaptation
ISSN (print) / ISBN [9798331555115]
Konferenztitel 2026 IEEE/CVF Winter Conference on Applications of Computer Vision, WACV 2026
Konferzenzdatum 6-10 March 2026
Konferenzort Tucson
Quellenangaben Band: , Heft: , Seiten: 4661-4672 Artikelnummer: , Supplement: ,
Institut(e) Institute for Machine Learning in Biomed Imaging (IML)