Metraipolis › archive › dc_c4f3023ebfdf
TASSO: TAsk-Specific Subspace Optimization for Continual Learning of Vision-Language Models — arXiv:2608.21487
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2608.21487
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-18T07:42:16.938Z
- SHA-256 of the archived text
54f2fcf787a3a770c79f8b8b5897b1d81deaa42fece49bf9fe0bc93803d98751- Stable URL of this page
https://metraipolis.com/documento/dc_c4f3023ebfdf
How to cite
TASSO: TAsk-Specific Subspace Optimization for Continual Learning of Vision-Language Models — arXiv:2608.21487 — arXiv (en), https://arxiv.org/abs/2608.21487 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_c4f3023ebfdf (Metraipolis archive dc_c4f3023ebfdf, sha256 54f2fcf787a3a770c79f8b8b5897b1d81deaa42fece49bf9fe0bc93803d98751)
BibTeX
@misc{dcc4f3023ebfdf,
title = {TASSO: TAsk-Specific Subspace Optimization for Continual Learning of Vision-Language Models — arXiv:2608.21487},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2608.21487},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_c4f3023ebfdf; sha256 54f2fcf787a3a770c79f8b8b5897b1d81deaa42fece49bf9fe0bc93803d98751},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.21487. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Vision-Language Models (VLMs) exhibit strong zero-shot capabilities, making them an attractive solution for continual learning across diverse tasks. However, during continual adaptation, both catastrophic forgetting and zero-shot degradation occur, severely degrading performance. In this paper, we introduce TASSO, a new paradigm that efficiently preserves the latent space geometry while ensuring network plasticity. We achieve this with two complementary techniques: subspace learning and geometry-aware knowledge distillation. Specifically, we first learn a sequence of task-specific low-rank projectors, which we use to project the latent representations before optimizing cross-entropy. Secondly, we employ a geodesic-distance-based loss that distills knowledge from the previous-task model while effectively preserving the latent space geometry. These design choices not only avoid unnecessary parameter updates along the full embedding dimensions but also improve learning by focusing on task-specific manifolds. Moreover, the geometry-aware distillation provides strong regularization and significantly reduces both catastrophic forgetting and zero-shot degradation throughout the continual learning sequence. Experimental results with the CLIP vision language model in the multi-domain task incremental and class incremental learning benchmarks demonstrate clear improvements over state-of-the-art methods in mitigating forgetting and preserving zero-shot capabilities. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-25 in cs.AI). Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV) ; Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2608.21487 — 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 2026-10-18T07:37:52.816Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-08-25?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).