Metraipolis › archive › dc_652dc0318a5d

Symmetry Breaking at the Edge of Stability via Central Flow Analysis — arXiv:2611.06718

Source
arXiv (en)
Original
https://arxiv.org/abs/2611.06718
License (as archived)
CC0-1.0
Archived
2026-11-05T18:22:26.266Z
SHA-256 of the archived text
0d9d4c4d4ff4cf050b8720077fb31392f8e57d8fbebb4b3b5014748b5de0b38c
Stable URL of this page
https://metraipolis.com/documento/dc_652dc0318a5d

How to cite

Symmetry Breaking at the Edge of Stability via Central Flow Analysis — arXiv:2611.06718 — arXiv (en), https://arxiv.org/abs/2611.06718 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_652dc0318a5d (Metraipolis archive dc_652dc0318a5d, sha256 0d9d4c4d4ff4cf050b8720077fb31392f8e57d8fbebb4b3b5014748b5de0b38c)
BibTeX
@misc{dc652dc0318a5d,
  title = {Symmetry Breaking at the Edge of Stability via Central Flow Analysis — arXiv:2611.06718},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.06718},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_652dc0318a5d; sha256 0d9d4c4d4ff4cf050b8720077fb31392f8e57d8fbebb4b3b5014748b5de0b38c},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.06718. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Continuous symmetries in neural networks lead to conserved quantities that are invariant under gradient flow, such as layer balancedness in linear and ReLU networks. Finite-step gradient descent need not preserve these laws, particularly at the edge of stability (EOS), where the iterates oscillate around a smoother trajectory, which can be described by the central flow dynamics (Cohen et al., 2025). We study how gradient flow conservation laws evolve under the central flow description of this regime. We show that the curvature correction induces a finite-dimensional drift determined by the oscillation covariance along critical parameter directions. We thus take the first steps to analyse which conservation laws remain and are broken at the EOS according to the central flow dynamics by analysing a scalar two-layer model, two-layer matrix factorisation, and a two-layer ReLU network, where the known conservation laws are shown to be complete (Marcotte et al., 2023).

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/2611.06718 — 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 2026-11-05T18:15:49.025Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).