A Metacognitive Framework for Test-Time Adaptation: Taxonomy and Research Challenges
Proceedings of the AAAI Symposium Series · 06.11.2026 · 10 (1) · 147-162
DOI: 10.1609/aaaiss.v10i1.43402 ↗Kurz gefasst
Fachartikel in der Zeitschrift Proceedings of the AAAI Symposium Series, online erschienen am 06.11.2026. Verfasst von Dario Loi und 5 weiteren Autor:innen.
Abstract (Originalsprache)
In Test-Time Adaptation, when out-of-distribution samples interleave unpredictably with corrupted in-distribution samples, adaptation without corrective measures is prone to significant performance decrease. Existing protocols rely on unreliable cues (e.g., predictive entropy) which are unable to distinguish corrupted in-distribution data from semantically novel open-set samples. To address this issue, we formalize Test-Time Adaptation through a two-level metacognitive framework that decouples task execution from control, facilitating a unified taxonomy that categorizes existing Test-Time Adaptation protocols across three core axes, i.e., which knowledge is monitored, which part of the models are targeted for adaptation, and how adaptation occurs. We provide evidence that robust corrective measures in Test-Time Adaptation require monitoring grounded in structured internal representations to reliably estimate model competence. Lastly, we frame monitoring as a resource-allocation problem, examine neuromorphic substrates for bounded self-regulation, and outline key open challenges.
Zitieren
APA
Loi, D., Onorato, G., Coviello, A., Rifat, S., Bastian, N. D., & Restuccia, F. (2026). A Metacognitive Framework for Test-Time Adaptation: Taxonomy and Research Challenges. Proceedings of the AAAI Symposium Series, 10(1), 147-162. https://doi.org/10.1609/aaaiss.v10i1.43402
Vancouver
Loi D, Onorato G, Coviello A, Rifat S, Bastian ND, Restuccia F. A Metacognitive Framework for Test-Time Adaptation: Taxonomy and Research Challenges. Proceedings of the AAAI Symposium Series. 2026;10(1):147-162. doi:10.1609/aaaiss.v10i1.43402
BibTeX
@article{Loi2026,
author = {Loi, Dario and Onorato, Gabriele and Coviello, Antonella and Rifat, Shahriar and Bastian, Nathaniel D. and Restuccia, Francesco},
title = {A Metacognitive Framework for Test-Time Adaptation: Taxonomy and Research Challenges},
journal = {Proceedings of the AAAI Symposium Series},
year = {2026},
volume = {10},
number = {1},
pages = {147--162},
doi = {10.1609/aaaiss.v10i1.43402}
}
EUVIMED