Metraipolis › archive › dc_2cc5fc48db0a

MuLoRA: Spectrally Balanced Low-Rank Adaptation for Continual Learning — arXiv:2610.02283

Source
arXiv
Original
https://arxiv.org/abs/2610.02283
License (as archived)
CC0-1.0
Archived
2026-10-05T23:08:37.924Z
SHA-256 of the archived text
cd463c90a88277e2fbeddbc4da4997020a4cabadf611c5a450f73e093d26f98a
Stable URL of this page
https://metraipolis.com/documento/dc_2cc5fc48db0a

How to cite

MuLoRA: Spectrally Balanced Low-Rank Adaptation for Continual Learning — arXiv:2610.02283 — arXiv, https://arxiv.org/abs/2610.02283 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_2cc5fc48db0a (Metraipolis archive dc_2cc5fc48db0a, sha256 cd463c90a88277e2fbeddbc4da4997020a4cabadf611c5a450f73e093d26f98a)
BibTeX
@misc{dc2cc5fc48db0a,
  title = {MuLoRA: Spectrally Balanced Low-Rank Adaptation for Continual Learning — arXiv:2610.02283},
  howpublished = {arXiv},
  url = {https://arxiv.org/abs/2610.02283},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_2cc5fc48db0a; sha256 cd463c90a88277e2fbeddbc4da4997020a4cabadf611c5a450f73e093d26f98a},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv, https://arxiv.org/abs/2610.02283. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Low-rank adaptation (LoRA) provides a parameter-efficient approach to continual learning, but its nominal rank can conceal a loss of effective adaptation capacity. We identify \emph{spectral plasticity collapse}: during sequential adaptation, update energy becomes concentrated in a small subset of singular modes, leaving much of the available low-rank space underutilized. This exposes a limitation of interference avoidance alone: protecting historical representations does not ensure that the remaining adaptation capacity is responsive to new tasks or effectively utilized. To address this problem, we propose \texttt{MuLoRA}, which jointly controls capacity allocation and utilization. First, historical whitening identifies input directions with strong current-task response relative to accumulated historical response, yielding a task-adaptive basis that remains fixed during training. Second, approximate polar orthogonalization of momentum updates reduces spectral concentration within theselected space. An orthonormal basis connects these mechanisms by transferring the factor-update spectrum exactly tothe induced weight update. We establish a max--min characterization of exact subspace selection and derive cumulative spectral bounds under controlled cross-step anisotropy. Across five class-incremental benchmarks and eight incremental settings, \texttt{MuLoRA} achieves the highest mean accuracy in 15 of 16 reported metrics.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.LG, nuovi lavori del giorno).
Categorie arXiv: Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2610.02283 — 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 2026-10-05T23:01:38.557Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).