Metraipolis › archive › dc_412de5bbfa22

Can Circuit Alignment Predict OOD Generalization? — arXiv:2609.31996

Source: arXiv (en)
Original: https://arxiv.org/abs/2609.31996
License (as archived): CC0-1.0
Archived: 2026-10-07T22:36:12.090Z
SHA-256 of the archived text: 343d9eeee22bd364969dd4a0957b8b69590bc2b5d7bbbb15972edc8c1cb55486
Stable URL of this page: https://metraipolis.com/documento/dc_412de5bbfa22

How to cite

Can Circuit Alignment Predict OOD Generalization? — arXiv:2609.31996 — arXiv (en), https://arxiv.org/abs/2609.31996 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_412de5bbfa22 (Metraipolis archive dc_412de5bbfa22, sha256 343d9eeee22bd364969dd4a0957b8b69590bc2b5d7bbbb15972edc8c1cb55486)

BibTeX

@misc{dc412de5bbfa22,
  title = {Can Circuit Alignment Predict OOD Generalization? — arXiv:2609.31996},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.31996},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_412de5bbfa22; sha256 343d9eeee22bd364969dd4a0957b8b69590bc2b5d7bbbb15972edc8c1cb55486},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.31996. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Can out-of-distribution (OOD) generalization be predicted from a trained model's weights alone, without any target-domain data? Existing representational similarity metrics (CKA, SVCCA, RSA) compare activations rather than forecast generalization. We show they are provably insensitive to structural rerouting in the computational graph, the very change distribution shift induces. We close this gap with the Circuit Alignment Score (CAS), which compares class-specific circuits across domains via graph kernels, decomposed into same-class coherence and cross-class confusion. Casting CAS as a Lebesgue integral over the domain distribution, we prove its Monte Carlo estimate recovers the ground-truth ranking of learners by OOD accuracy, with pairwise inversion error vanishing at rate $O(1/M)$, where $M$ is the number of sampled domains. Across $48$ learners on PACS, CAS attains $0.88$ rank correlation with OOD accuracy, versus $0.58$ (CKA), $0.23$ (SVCCA), and $0.14$ (RSA), with similar trends on other benchmarks and even against data-dependent methods, making it the first provably consistent predictor of distributional robustness requiring neither target-domain data nor labels. The code is available at: this https URL

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-29 in cs.LG).
Categorie arXiv: Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2609.31996 — 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 2026-10-07T22:14:27.574Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).