Metraipolis › archive › dc_abdff34050bb

Information-Theoretic Decoupled Prompt Tuning for Continual Learning — arXiv:2609.26257

Source: arXiv (en)
Original: https://arxiv.org/abs/2609.26257
License (as archived): CC0-1.0
Archived: 2026-10-08T15:07:53.471Z
SHA-256 of the archived text: 1d5676711d7ea112ab0bf16802cceb2d35c2aefb481eb444e05d2a4e46c96751
Stable URL of this page: https://metraipolis.com/documento/dc_abdff34050bb

How to cite

Information-Theoretic Decoupled Prompt Tuning for Continual Learning — arXiv:2609.26257 — arXiv (en), https://arxiv.org/abs/2609.26257 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_abdff34050bb (Metraipolis archive dc_abdff34050bb, sha256 1d5676711d7ea112ab0bf16802cceb2d35c2aefb481eb444e05d2a4e46c96751)

BibTeX

@misc{dcabdff34050bb,
  title = {Information-Theoretic Decoupled Prompt Tuning for Continual Learning — arXiv:2609.26257},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.26257},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_abdff34050bb; sha256 1d5676711d7ea112ab0bf16802cceb2d35c2aefb481eb444e05d2a4e46c96751},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.26257. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Continual learning (CL) aims to incrementally acquire knowledge from sequential data while avoiding catastrophic forgetting. Recently, prompt tuning has attracted increasing attention as an efficient approach for adapting pre-trained models to CL tasks. However, existing prompt design paradigms commonly suffer from retrieval dependence and classifier bias, which make model adaptation sensitive to prompt selection and bias predictions toward newly arrived classes. To address these challenges, we propose Decoupled Prompt Tuning for Continual Learning (DPT4CL), which decouples the CLIP textual prompt into a task-shared prompt distribution and class-specific prompts. The task-shared prompt distribution is derived by optimizing an Information Bottleneck objective to facilitate cross-task knowledge transfer and alleviate classifier bias, while class-specific prompts enhance inter-class separability without relying on explicit prompt retrieval. Furthermore, we establish a unified excess risk bound from an information-theoretic perspective, providing theoretical support for the robust generalization and forgetting mitigation of the proposed framework. Extensive experiments on standard CL benchmarks demonstrate that DPT4CL achieves state-of-the-art performance. The source code is available at this https URL

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-23 in cs.LG).
Categorie arXiv: Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2609.26257 — 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 2026-10-08T14:37:29.475Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-09-23?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).