Metraipolis › archive › dc_d9747ef43f2d

Nonlinear Maps Between Irreducible Representations in Graph Neural Networks — arXiv:2610.41059

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.41059
License (as archived)
CC0-1.0
Archived
2026-11-02T07:49:25.412Z
SHA-256 of the archived text
4ddb2ec55f4c19db042b9ae8f0c56f4f0ad144d25031361776f7881fc7ae65e8
Stable URL of this page
https://metraipolis.com/documento/dc_d9747ef43f2d

How to cite

Nonlinear Maps Between Irreducible Representations in Graph Neural Networks — arXiv:2610.41059 — arXiv (en), https://arxiv.org/abs/2610.41059 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_d9747ef43f2d (Metraipolis archive dc_d9747ef43f2d, sha256 4ddb2ec55f4c19db042b9ae8f0c56f4f0ad144d25031361776f7881fc7ae65e8)
BibTeX
@misc{dcd9747ef43f2d,
  title = {Nonlinear Maps Between Irreducible Representations in Graph Neural Networks — arXiv:2610.41059},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.41059},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_d9747ef43f2d; sha256 4ddb2ec55f4c19db042b9ae8f0c56f4f0ad144d25031361776f7881fc7ae65e8},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.41059. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Equivariant linear maps cannot mix non-isomorphic irreducible representations, whereas nonlinear activations can. We develop a degree-resolved version of this nonlinear interaction picture and use the resulting block inventory to construct Specht Triple, a residual on unordered graph triples. On a synthetic structural benchmark with a frozen GIN backbone, Specht Triple reduces aggregate standardised test MSE by 11.05% relative to a parameter-matched Projector Control using the same irreducible projectors and complete linear equivariant space. The results support irreducible decomposition as an inductive bias for nonlinear higher-order equivariant processing.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.LG, nuovi lavori del giorno).
Categorie arXiv: Machine Learning (cs.LG) ; Representation Theory (math.RT).
Pagina del lavoro: https://arxiv.org/abs/2610.41059 — 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 2026-11-02T07:40:00.173Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).