Metraipolis › archive › dc_0f3100e69343

Parameter-Level Attribution of Symmetry in Trained Networks Though Parameter-Wise Functional Sensitivity — arXiv:2608.24700

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.24700
License (as archived)
CC0-1.0
Archived
2026-10-18T06:36:22.334Z
SHA-256 of the archived text
706e3ad1033f49684dc27a8996eb5bd3bc00ebd615f307f5f1fdb86597e533cf
Stable URL of this page
https://metraipolis.com/documento/dc_0f3100e69343

How to cite

Parameter-Level Attribution of Symmetry in Trained Networks Though Parameter-Wise Functional Sensitivity — arXiv:2608.24700 — arXiv (en), https://arxiv.org/abs/2608.24700 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_0f3100e69343 (Metraipolis archive dc_0f3100e69343, sha256 706e3ad1033f49684dc27a8996eb5bd3bc00ebd615f307f5f1fdb86597e533cf)
BibTeX
@misc{dc0f3100e69343,
  title = {Parameter-Level Attribution of Symmetry in Trained Networks Though Parameter-Wise Functional Sensitivity — arXiv:2608.24700},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.24700},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_0f3100e69343; sha256 706e3ad1033f49684dc27a8996eb5bd3bc00ebd615f307f5f1fdb86597e533cf},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.24700. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

When a network has learned a function with a known symmetry, can that symmetry be moved through the parametrisation---is there a motion in parameter space realising the group action in function space? We formulate this as a lifting problem for the realisation map $\Phi:\theta\mapsto f_\theta$, and show that a smooth parameter-space action exists only if the tangent space to the function's symmetry orbit lies within the image of $\mathrm d\Phi_\theta$, whose columns are the \emph{functional sensitivities} of individual parameters. This condition is also sufficient for pointwise first-order lifting. Relaxing it in least squares yields two local parameter directions: one following the symmetry orbit, one descending towards the equivariant subspace, with residuals measuring what the parametrisation cannot reach. On a rotationally invariant classifier we find these directions induce their predicted function-space motion, but only locally: recomputed directions track the orbit and reduce the equivariance defect, while directions held fixed depart from both after training. The same holds for Hamiltonian neural networks trained on a rotationally symmetric potential, even though the architecture does not explicitly enforce the symmetry.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-26 in cs.LG).
Categorie arXiv: Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2608.24700 — 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 2026-10-18T06:28:52.271Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-08-26?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).