Metraipolis › archive › dc_02cdb8ad7c2f

Hierarchical Response Preservation for Continual Adaptation of Zero-Shot Graph-Text Models — arXiv:2609.33607

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.33607
License (as archived)
CC0-1.0
Archived
2026-10-07T22:37:37.261Z
SHA-256 of the archived text
77cb52006c50ec9527b7d01445f3c63d196d8d61255fc96d7ef9dd24f770ba24
Stable URL of this page
https://metraipolis.com/documento/dc_02cdb8ad7c2f

How to cite

Hierarchical Response Preservation for Continual Adaptation of Zero-Shot Graph-Text Models — arXiv:2609.33607 — arXiv (en), https://arxiv.org/abs/2609.33607 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_02cdb8ad7c2f (Metraipolis archive dc_02cdb8ad7c2f, sha256 77cb52006c50ec9527b7d01445f3c63d196d8d61255fc96d7ef9dd24f770ba24)
BibTeX
@misc{dc02cdb8ad7c2f,
  title = {Hierarchical Response Preservation for Continual Adaptation of Zero-Shot Graph-Text Models — arXiv:2609.33607},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.33607},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_02cdb8ad7c2f; sha256 77cb52006c50ec9527b7d01445f3c63d196d8d61255fc96d7ef9dd24f770ba24},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.33607. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Pretrained graph-text models align graph representations with textual semantics, enabling recognition of unseen classes and transfer across graph domains. However, as graph data and classes continually arrive, models should learn from new supervision while retaining their zero-shot transfer capabilities and historical task knowledge. Two challenges arise: (i) new classes can overturn historical predictions despite preserved distinctions among historical classes, and (ii) overly strict response preservation can stall learning of new classes. To address these challenges, we propose Hierarchical Response Preservation (HiRP). HiRP represents this competition through a hierarchical response that keeps each historical-class probability and sums new-class probabilities, preserving historical distinctions and aggregate competition while allowing distinctions within the new class group to adapt. It further uses the geometry induced by this response to guide constrained updates, retaining useful adaptation directions while controlling response drift. Across three class-incremental settings, HiRP achieves absolute gains of 1.84-7.95 percentage points in average accuracy over the strongest compared baseline in each setting, while mitigating zero-shot transfer degradation.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-29 in cs.LG).
Categorie arXiv: Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2609.33607 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-07T22:14:27.998Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).